notaz.gp2x.de / pcsx_rearmed.git / blame
| Commit | Line | Data |
|---|---|---|
| 75e28f62 E |
1 | /* |
| 2 | * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com> | |
| 3 | * | |
| 4 | * This program is free software; you can redistribute it and/or | |
| 5 | * modify it under the terms of the GNU General Public License as | |
| 6 | * published by the Free Software Foundation; either version 2 of | |
| 7 | * the License, or (at your option) any later version. | |
| 8 | * | |
| 9 | * This program is distributed in the hope that it will be useful, | |
| 10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
| 11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
| 12 | * General Public License for more details. | |
| 13 | */ | |
| 14 | ||
| 15 | #include <stdio.h> | |
| 16 | #include <stdlib.h> | |
| 77e1e479 | 17 | #include <stdint.h> |
| 2da2fc76 | 18 | #include <stddef.h> |
| 75e28f62 | 19 | #include <string.h> |
| 2d658c89 | 20 | #include <assert.h> |
| 75e28f62 E |
21 | |
| 22 | #include "common.h" | |
| a4021361 | 23 | #ifndef NEON_BUILD |
| 24 | #include "vector_ops.h" | |
| 25 | #endif | |
| 37725e8c | 26 | #include "psx_gpu_simd.h" |
| 2da2fc76 | 27 | #include "psx_gpu_offsets.h" |
| 75e28f62 | 28 | |
| aafce833 | 29 | #if 0 |
| 30 | void dump_r_d(const char *name, void *dump); | |
| 31 | void dump_r_q(const char *name, void *dump); | |
| 32 | #define dumprd(n) dump_r_d(#n, n.e) | |
| 33 | #define dumprq(n) dump_r_q(#n, n.e) | |
| 34 | #endif | |
| 35 | ||
| 75e28f62 E |
36 | u32 span_pixels = 0; |
| 37 | u32 span_pixel_blocks = 0; | |
| 75e28f62 E |
38 | u32 spans = 0; |
| 39 | u32 triangles = 0; | |
| 40 | u32 sprites = 0; | |
| 41 | u32 sprites_4bpp = 0; | |
| 42 | u32 sprites_8bpp = 0; | |
| 43 | u32 sprites_16bpp = 0; | |
| 44 | u32 sprite_blocks = 0; | |
| 45 | u32 sprites_untextured = 0; | |
| 46 | u32 lines = 0; | |
| 47 | u32 trivial_rejects = 0; | |
| 48 | u32 texels_4bpp = 0; | |
| 49 | u32 texels_8bpp = 0; | |
| 50 | u32 texels_16bpp = 0; | |
| 51 | u32 texel_blocks_4bpp = 0; | |
| 52 | u32 texel_blocks_8bpp = 0; | |
| 53 | u32 texel_blocks_16bpp = 0; | |
| 54 | u32 texel_blocks_untextured = 0; | |
| 55 | u32 blend_blocks = 0; | |
| 75e28f62 E |
56 | u32 render_buffer_flushes = 0; |
| 57 | u32 state_changes = 0; | |
| 58 | u32 left_split_triangles = 0; | |
| 59 | u32 flat_triangles = 0; | |
| 60 | u32 clipped_triangles = 0; | |
| 61 | u32 zero_block_spans = 0; | |
| 62 | u32 texture_cache_loads = 0; | |
| 3867c6ef | 63 | u32 false_modulated_blocks = 0; |
| 75e28f62 | 64 | |
| a2cb152a | 65 | #define stats_add(stat, count) // stat += count |
| 66 | ||
| c1817bd9 | 67 | /* double size for enhancement */ |
| 68 | u32 reciprocal_table[512 * 2]; | |
| 75e28f62 E |
69 | |
| 70 | ||
| 71 | typedef s32 fixed_type; | |
| 72 | ||
| 73 | #define EDGE_STEP_BITS 32 | |
| 74 | #define FIXED_BITS 12 | |
| 75 | ||
| 76 | #define fixed_center(value) \ | |
| 77 | ((((fixed_type)(value)) << FIXED_BITS) + (1 << (FIXED_BITS - 1))) \ | |
| 78 | ||
| 79 | #define int_to_fixed(value) \ | |
| 80 | (((fixed_type)(value)) << FIXED_BITS) \ | |
| 81 | ||
| 82 | #define fixed_to_int(value) \ | |
| 83 | ((value) >> FIXED_BITS) \ | |
| 84 | ||
| 85 | #define fixed_to_double(value) \ | |
| 86 | ((value) / (double)(1 << FIXED_BITS)) \ | |
| 87 | ||
| 88 | #define double_to_fixed(value) \ | |
| 89 | (fixed_type)(((value) * (double)(1 << FIXED_BITS))) \ | |
| 90 | ||
| 91 | typedef void (setup_blocks_function_type)(psx_gpu_struct *psx_gpu); | |
| 92 | typedef void (texture_blocks_function_type)(psx_gpu_struct *psx_gpu); | |
| 93 | typedef void (shade_blocks_function_type)(psx_gpu_struct *psx_gpu); | |
| 94 | typedef void (blend_blocks_function_type)(psx_gpu_struct *psx_gpu); | |
| 95 | ||
| 96 | typedef void (setup_sprite_function_type)(psx_gpu_struct *psx_gpu, s32 x, | |
| 97 | s32 y, s32 u, s32 v, s32 width, s32 height, u32 color); | |
| 98 | ||
| 99 | struct render_block_handler_struct | |
| 100 | { | |
| 101 | void *setup_blocks; | |
| 102 | texture_blocks_function_type *texture_blocks; | |
| 103 | shade_blocks_function_type *shade_blocks; | |
| 104 | blend_blocks_function_type *blend_blocks; | |
| 105 | }; | |
| 106 | ||
| 2bbbb7af | 107 | #ifndef NEON_BUILD |
| 75e28f62 E |
108 | |
| 109 | u32 fixed_reciprocal(u32 denominator, u32 *_shift) | |
| 110 | { | |
| 111 | u32 shift = __builtin_clz(denominator); | |
| 112 | u32 denominator_normalized = denominator << shift; | |
| 113 | ||
| 114 | double numerator = (1ULL << 62) + denominator_normalized; | |
| 115 | double numerator_b; | |
| 116 | ||
| 117 | double denominator_normalized_dp_b; | |
| 118 | u64 denominator_normalized_dp_u64; | |
| 119 | ||
| 120 | u32 reciprocal; | |
| 121 | double reciprocal_dp; | |
| 122 | ||
| 123 | u64 numerator_u64 = (denominator_normalized >> 10) | | |
| 124 | ((u64)(62 + 1023) << 52); | |
| 125 | *((u64 *)(&numerator_b)) = numerator_u64; | |
| 126 | ||
| 127 | denominator_normalized_dp_u64 = | |
| 128 | (u64)(denominator_normalized << 21) | | |
| 129 | ((u64)((denominator_normalized >> 11) + ((1022 + 31) << 20)) << 32); | |
| 130 | *((u64 *)(&denominator_normalized_dp_b)) = denominator_normalized_dp_u64; | |
| 131 | ||
| 132 | // Implement with a DP divide | |
| 133 | reciprocal_dp = numerator / denominator_normalized_dp_b; | |
| 134 | reciprocal = reciprocal_dp; | |
| 135 | ||
| 136 | if(reciprocal == 0x80000001) | |
| 137 | reciprocal = 0x80000000; | |
| 138 | ||
| 139 | *_shift = 62 - shift; | |
| 140 | return reciprocal; | |
| 141 | } | |
| 142 | ||
| 143 | double reciprocal_estimate(double a) | |
| 144 | { | |
| 145 | int q, s; | |
| 146 | double r; | |
| 147 | ||
| 148 | q = (int)(a * 512.0); | |
| 149 | /* a in units of 1/512 rounded down */ | |
| 150 | r = 1.0 / (((double)q + 0.5) / 512.0); /* reciprocal r */ | |
| 151 | s = (int)(256.0 * r + 0.5); | |
| 152 | ||
| 153 | /* r in units of 1/256 rounded to nearest */ | |
| 154 | ||
| 155 | return (double)s / 256.0; | |
| 156 | } | |
| 157 | ||
| 158 | u32 reciprocal_estimate_u32(u32 value) | |
| 159 | { | |
| 160 | u64 dp_value_u64; | |
| 161 | volatile double dp_value; | |
| 162 | volatile u64 *dp_value_ptr = (volatile u64 *)&dp_value; | |
| 163 | ||
| 164 | if((value >> 31) == 0) | |
| 165 | return 0xFFFFFFFF; | |
| 166 | ||
| 167 | dp_value_u64 = (0x3FEULL << (31 + 21)) | ((u64)(value & 0x7FFFFFFF) << 21); | |
| 168 | ||
| 169 | *dp_value_ptr = dp_value_u64; | |
| 170 | ||
| 171 | dp_value = reciprocal_estimate(dp_value); | |
| 172 | dp_value_u64 = *dp_value_ptr; | |
| 173 | ||
| 174 | return (0x80000000 | ((dp_value_u64 >> 21) & 0x7FFFFFFF)); | |
| 175 | } | |
| 176 | ||
| 177 | u32 fixed_reciprocal_nr(u32 value, u32 *_shift) | |
| 178 | { | |
| 179 | u32 shift = __builtin_clz(value); | |
| 180 | u32 value_normalized = value << shift; | |
| 181 | ||
| 182 | *_shift = 62 - shift; | |
| 183 | ||
| 184 | value_normalized -= 2; | |
| 185 | ||
| 186 | u32 reciprocal_normalized = reciprocal_estimate_u32(value_normalized) >> 1; | |
| 187 | ||
| 188 | u32 temp = -(((u64)value_normalized * (u32)reciprocal_normalized) >> 31); | |
| 189 | reciprocal_normalized = (((u64)reciprocal_normalized * temp) >> 31); | |
| 190 | temp = -(((u64)value_normalized * (u32)reciprocal_normalized) >> 31); | |
| 191 | reciprocal_normalized = (((u64)reciprocal_normalized * temp) >> 31); | |
| 192 | temp = -(((u64)value_normalized * (u32)reciprocal_normalized) >> 31); | |
| 193 | reciprocal_normalized = (((u64)reciprocal_normalized * temp) >> 31); | |
| 194 | ||
| 195 | return reciprocal_normalized; | |
| 196 | } | |
| 197 | ||
| 198 | #endif | |
| 199 | ||
| 200 | ||
| 201 | s32 triangle_signed_area_x2(s32 x0, s32 y0, s32 x1, s32 y1, s32 x2, s32 y2) | |
| 202 | { | |
| 203 | return ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)); | |
| 204 | } | |
| 205 | ||
| 206 | u32 texture_region_mask(s32 x1, s32 y1, s32 x2, s32 y2) | |
| 207 | { | |
| 208 | s32 coverage_x, coverage_y; | |
| 209 | ||
| 210 | u32 mask_up_left; | |
| 211 | u32 mask_down_right; | |
| 212 | ||
| 213 | coverage_x = x2 >> 6; | |
| 214 | coverage_y = y2 >> 8; | |
| 215 | ||
| 216 | if(coverage_x < 0) | |
| 217 | coverage_x = 0; | |
| 218 | ||
| 219 | if(coverage_x > 31) | |
| 220 | coverage_x = 31; | |
| 221 | ||
| 222 | mask_down_right = ~(0xFFFFFFFF << (coverage_x + 1)) & 0xFFFF; | |
| 223 | ||
| 224 | if(coverage_y >= 1) | |
| 225 | mask_down_right |= mask_down_right << 16; | |
| 226 | ||
| 227 | coverage_x = x1 >> 6; | |
| 228 | ||
| f707f14b | 229 | mask_up_left = 0xFFFF0000; |
| 230 | if(coverage_x > 0) | |
| 231 | mask_up_left <<= coverage_x; | |
| 75e28f62 E |
232 | |
| 233 | coverage_y = y1 >> 8; | |
| 234 | if(coverage_y <= 0) | |
| 235 | mask_up_left |= mask_up_left >> 16; | |
| 236 | ||
| 237 | return mask_up_left & mask_down_right; | |
| 238 | } | |
| 239 | ||
| 240 | u32 invalidate_texture_cache_region(psx_gpu_struct *psx_gpu, u32 x1, u32 y1, | |
| 241 | u32 x2, u32 y2) | |
| 242 | { | |
| 243 | u32 mask = texture_region_mask(x1, y1, x2, y2); | |
| 244 | ||
| 245 | psx_gpu->dirty_textures_4bpp_mask |= mask; | |
| 246 | psx_gpu->dirty_textures_8bpp_mask |= mask; | |
| 247 | psx_gpu->dirty_textures_8bpp_alternate_mask |= mask; | |
| 248 | ||
| 249 | return mask; | |
| 250 | } | |
| 251 | ||
| 252 | u32 invalidate_texture_cache_region_viewport(psx_gpu_struct *psx_gpu, u32 x1, | |
| 253 | u32 y1, u32 x2, u32 y2) | |
| 254 | { | |
| 255 | u32 mask = texture_region_mask(x1, y1, x2, y2) & | |
| 256 | psx_gpu->viewport_mask; | |
| 3867c6ef | 257 | |
| 75e28f62 E |
258 | psx_gpu->dirty_textures_4bpp_mask |= mask; |
| 259 | psx_gpu->dirty_textures_8bpp_mask |= mask; | |
| 260 | psx_gpu->dirty_textures_8bpp_alternate_mask |= mask; | |
| 261 | ||
| 262 | return mask; | |
| 263 | } | |
| 264 | ||
| 47c15995 | 265 | static void update_texture_cache_region_(psx_gpu_struct *psx_gpu, |
| 266 | u32 x1, u32 y1, u32 x2, u32 y2) | |
| 05740673 | 267 | { |
| 268 | u32 mask = texture_region_mask(x1, y1, x2, y2); | |
| 269 | u32 texture_page; | |
| 270 | u8 *texture_page_ptr; | |
| 271 | u16 *vram_ptr; | |
| 272 | u32 texel_block; | |
| 273 | u32 sub_x, sub_y; | |
| 274 | ||
| 275 | psx_gpu->dirty_textures_8bpp_mask |= mask; | |
| 276 | psx_gpu->dirty_textures_8bpp_alternate_mask |= mask; | |
| 277 | ||
| 278 | if ((psx_gpu->dirty_textures_4bpp_mask & mask) == 0 && | |
| 279 | (x1 & 3) == 0 && (y1 & 15) == 0 && x2 - x1 < 4 && y2 - y1 < 16) | |
| 280 | { | |
| 281 | texture_page = ((x1 / 64) & 15) + (y1 / 256) * 16; | |
| 282 | texture_page_ptr = psx_gpu->texture_4bpp_cache[texture_page]; | |
| 283 | texture_page_ptr += (x1 / 4 & 15) * 16*16 + (y1 / 16 & 15) * 16*16*16; | |
| 284 | vram_ptr = psx_gpu->vram_ptr + x1 + y1 * 1024; | |
| 285 | sub_x = 4; | |
| 286 | sub_y = 16; | |
| 287 | ||
| 288 | while(sub_y) | |
| 289 | { | |
| 290 | while(sub_x) | |
| 291 | { | |
| 292 | texel_block = *vram_ptr; | |
| 293 | ||
| 294 | texture_page_ptr[0] = texel_block & 0xF; | |
| 295 | texture_page_ptr[1] = (texel_block >> 4) & 0xF; | |
| 296 | texture_page_ptr[2] = (texel_block >> 8) & 0xF; | |
| 297 | texture_page_ptr[3] = texel_block >> 12; | |
| 298 | ||
| 299 | vram_ptr++; | |
| 300 | texture_page_ptr += 4; | |
| 301 | ||
| 302 | sub_x--; | |
| 303 | } | |
| 304 | ||
| 305 | vram_ptr -= 4; | |
| 306 | sub_x = 4; | |
| 307 | ||
| 308 | sub_y--; | |
| 309 | vram_ptr += 1024; | |
| 310 | } | |
| 311 | } | |
| 312 | else | |
| 313 | { | |
| 314 | psx_gpu->dirty_textures_4bpp_mask |= mask; | |
| 315 | } | |
| 316 | } | |
| 75e28f62 | 317 | |
| 47c15995 | 318 | void update_texture_cache_region(psx_gpu_struct *psx_gpu, u32 x1, u32 y1, |
| 319 | u32 x2, u32 y2) | |
| 320 | { | |
| 321 | s32 w = x2 - x1; | |
| 322 | do | |
| 323 | { | |
| 324 | x2 = x1 + w; | |
| 325 | if (x2 > 1023) | |
| 326 | x2 = 1023; | |
| 327 | update_texture_cache_region_(psx_gpu, x1, y1, x2, y2); | |
| 328 | w -= x2 - x1; | |
| 329 | x1 = 0; | |
| 330 | } | |
| 331 | while (unlikely(w > 0)); | |
| 332 | } | |
| 333 | ||
| 2bbbb7af | 334 | #ifndef NEON_BUILD |
| 75e28f62 E |
335 | |
| 336 | void update_texture_4bpp_cache(psx_gpu_struct *psx_gpu) | |
| 337 | { | |
| 338 | u32 current_texture_page = psx_gpu->current_texture_page; | |
| 3867c6ef | 339 | u8 *texture_page_ptr = psx_gpu->texture_page_base; |
| 75e28f62 E |
340 | u16 *vram_ptr = psx_gpu->vram_ptr; |
| 341 | ||
| 342 | u32 texel_block; | |
| 343 | u32 tile_x, tile_y; | |
| 344 | u32 sub_x, sub_y; | |
| 345 | ||
| 346 | vram_ptr += (current_texture_page >> 4) * 256 * 1024; | |
| 347 | vram_ptr += (current_texture_page & 0xF) * 64; | |
| 348 | ||
| 349 | texture_cache_loads++; | |
| 350 | ||
| 351 | tile_y = 16; | |
| 352 | tile_x = 16; | |
| 353 | sub_x = 4; | |
| 354 | sub_y = 16; | |
| 355 | ||
| 356 | psx_gpu->dirty_textures_4bpp_mask &= ~(psx_gpu->current_texture_mask); | |
| 357 | ||
| 358 | while(tile_y) | |
| 359 | { | |
| 360 | while(tile_x) | |
| 361 | { | |
| 362 | while(sub_y) | |
| 363 | { | |
| 364 | while(sub_x) | |
| 365 | { | |
| 366 | texel_block = *vram_ptr; | |
| b7ed0632 | 367 | |
| 75e28f62 E |
368 | texture_page_ptr[0] = texel_block & 0xF; |
| 369 | texture_page_ptr[1] = (texel_block >> 4) & 0xF; | |
| 370 | texture_page_ptr[2] = (texel_block >> 8) & 0xF; | |
| 371 | texture_page_ptr[3] = texel_block >> 12; | |
| 372 | ||
| 373 | vram_ptr++; | |
| 374 | texture_page_ptr += 4; | |
| 375 | ||
| 376 | sub_x--; | |
| 377 | } | |
| 378 | ||
| 379 | vram_ptr -= 4; | |
| 380 | sub_x = 4; | |
| 381 | ||
| 382 | sub_y--; | |
| 383 | vram_ptr += 1024; | |
| 384 | } | |
| 385 | ||
| 386 | sub_y = 16; | |
| 387 | ||
| 388 | vram_ptr -= (1024 * 16) - 4; | |
| 389 | tile_x--; | |
| 390 | } | |
| 391 | ||
| 392 | tile_x = 16; | |
| 393 | ||
| 394 | vram_ptr += (16 * 1024) - (4 * 16); | |
| 395 | tile_y--; | |
| 396 | } | |
| 397 | } | |
| 398 | ||
| 399 | void update_texture_8bpp_cache_slice(psx_gpu_struct *psx_gpu, | |
| 400 | u32 texture_page) | |
| 401 | { | |
| 3867c6ef | 402 | u16 *texture_page_ptr = psx_gpu->texture_page_base; |
| 75e28f62 E |
403 | u16 *vram_ptr = psx_gpu->vram_ptr; |
| 404 | ||
| 405 | u32 tile_x, tile_y; | |
| 406 | u32 sub_y; | |
| 407 | ||
| 408 | vec_8x16u texels; | |
| 409 | ||
| 410 | texture_cache_loads++; | |
| 411 | ||
| 412 | vram_ptr += (texture_page >> 4) * 256 * 1024; | |
| 413 | vram_ptr += (texture_page & 0xF) * 64; | |
| 414 | ||
| 415 | if((texture_page ^ psx_gpu->current_texture_page) & 0x1) | |
| 416 | texture_page_ptr += (8 * 16) * 8; | |
| 417 | ||
| 418 | tile_x = 8; | |
| 419 | tile_y = 16; | |
| 420 | ||
| 421 | sub_y = 16; | |
| 422 | ||
| 423 | while(tile_y) | |
| 424 | { | |
| 425 | while(tile_x) | |
| 426 | { | |
| 427 | while(sub_y) | |
| 428 | { | |
| 429 | load_128b(texels, vram_ptr); | |
| 430 | store_128b(texels, texture_page_ptr); | |
| 431 | ||
| 432 | texture_page_ptr += 8; | |
| 433 | vram_ptr += 1024; | |
| 434 | ||
| 435 | sub_y--; | |
| 436 | } | |
| 437 | ||
| 438 | sub_y = 16; | |
| 439 | ||
| 440 | vram_ptr -= (1024 * 16); | |
| 441 | vram_ptr += 8; | |
| 442 | ||
| 443 | tile_x--; | |
| 444 | } | |
| 445 | ||
| 446 | tile_x = 8; | |
| 447 | ||
| 448 | vram_ptr -= (8 * 8); | |
| 449 | vram_ptr += (16 * 1024); | |
| 450 | ||
| 451 | texture_page_ptr += (8 * 16) * 8; | |
| 452 | tile_y--; | |
| 453 | } | |
| 454 | } | |
| 455 | ||
| 456 | #endif | |
| 457 | ||
| 458 | ||
| 459 | void update_texture_8bpp_cache(psx_gpu_struct *psx_gpu) | |
| 460 | { | |
| 461 | u32 current_texture_page = psx_gpu->current_texture_page; | |
| 462 | u32 update_textures = | |
| 463 | psx_gpu->dirty_textures_8bpp_mask & psx_gpu->current_texture_mask; | |
| 464 | ||
| 465 | psx_gpu->dirty_textures_8bpp_mask &= ~update_textures; | |
| 466 | ||
| 467 | if(update_textures & (1 << current_texture_page)) | |
| 468 | { | |
| 469 | update_texture_8bpp_cache_slice(psx_gpu, current_texture_page); | |
| 470 | update_textures &= ~(1 << current_texture_page); | |
| 471 | } | |
| 472 | ||
| 473 | if(update_textures) | |
| 474 | { | |
| 475 | u32 adjacent_texture_page = ((current_texture_page + 1) & 0xF) | | |
| 476 | (current_texture_page & 0x10); | |
| 477 | ||
| 478 | update_texture_8bpp_cache_slice(psx_gpu, adjacent_texture_page); | |
| 479 | } | |
| 480 | } | |
| 481 | ||
| 75e28f62 E |
482 | void flush_render_block_buffer(psx_gpu_struct *psx_gpu) |
| 483 | { | |
| f1359c57 | 484 | if((psx_gpu->render_mode & RENDER_INTERLACE_ENABLED) && |
| 69b09c0d E |
485 | (psx_gpu->primitive_type == PRIMITIVE_TYPE_SPRITE)) |
| 486 | { | |
| 487 | u32 num_blocks_dest = 0; | |
| 488 | block_struct *block_src = psx_gpu->blocks; | |
| 489 | block_struct *block_dest = psx_gpu->blocks; | |
| 490 | ||
| 491 | u16 *vram_ptr = psx_gpu->vram_ptr; | |
| 492 | u32 i; | |
| 493 | ||
| f1359c57 | 494 | if(psx_gpu->render_mode & RENDER_INTERLACE_ODD) |
| 69b09c0d E |
495 | { |
| 496 | for(i = 0; i < psx_gpu->num_blocks; i++) | |
| 497 | { | |
| 498 | u32 fb_offset = (u32)((u8 *)block_src->fb_ptr - (u8 *)vram_ptr); | |
| 499 | if(fb_offset & (1 << 11)) | |
| 500 | { | |
| 501 | *block_dest = *block_src; | |
| 502 | num_blocks_dest++; | |
| 503 | block_dest++; | |
| 504 | } | |
| 505 | block_src++; | |
| 506 | } | |
| 507 | } | |
| 508 | else | |
| 509 | { | |
| 510 | for(i = 0; i < psx_gpu->num_blocks; i++) | |
| 511 | { | |
| 512 | u32 fb_offset = (u32)((u8 *)block_src->fb_ptr - (u8 *)vram_ptr); | |
| 513 | if((fb_offset & (1 << 11)) == 0) | |
| 514 | { | |
| 515 | *block_dest = *block_src; | |
| 516 | num_blocks_dest++; | |
| 517 | block_dest++; | |
| 518 | } | |
| 519 | block_src++; | |
| 520 | } | |
| 521 | } | |
| 522 | ||
| 523 | psx_gpu->num_blocks = num_blocks_dest; | |
| 524 | } | |
| 525 | ||
| 75e28f62 E |
526 | if(psx_gpu->num_blocks) |
| 527 | { | |
| 528 | render_block_handler_struct *render_block_handler = | |
| 529 | psx_gpu->render_block_handler; | |
| 530 | ||
| db2804fb | 531 | #if defined(__arm__) && defined(NEON_BUILD) && !defined(SIMD_BUILD) |
| 532 | // the asm doesn't bother to save callee-save vector regs, so do it here | |
| 533 | __asm__ __volatile__("":::"q4","q5","q6","q7"); | |
| 534 | #endif | |
| 535 | ||
| 75e28f62 E |
536 | render_block_handler->texture_blocks(psx_gpu); |
| 537 | render_block_handler->shade_blocks(psx_gpu); | |
| 538 | render_block_handler->blend_blocks(psx_gpu); | |
| 539 | ||
| 3867c6ef | 540 | #ifdef PROFILE |
| 75e28f62 E |
541 | span_pixel_blocks += psx_gpu->num_blocks; |
| 542 | render_buffer_flushes++; | |
| 3867c6ef | 543 | #endif |
| 75e28f62 E |
544 | |
| 545 | psx_gpu->num_blocks = 0; | |
| db2804fb | 546 | #if defined(__arm__) && defined(NEON_BUILD) && !defined(SIMD_BUILD) |
| 547 | __asm__ __volatile__("":::"q4","q5","q6","q7"); | |
| 548 | #endif | |
| 75e28f62 E |
549 | } |
| 550 | } | |
| 551 | ||
| 552 | ||
| 2bbbb7af | 553 | #ifndef NEON_BUILD |
| 75e28f62 E |
554 | |
| 555 | #define setup_gradient_calculation_input(set, vertex) \ | |
| 556 | /* First type is: uvrg bxxx xxxx */\ | |
| 557 | /* Second type is: yyyy ybyy uvrg */\ | |
| 558 | /* Since x_a and y_c are the same the same variable is used for both. */\ | |
| 559 | x##set##_a_y##set##_c.e[0] = vertex->u; \ | |
| 560 | x##set##_a_y##set##_c.e[1] = vertex->v; \ | |
| 561 | x##set##_a_y##set##_c.e[2] = vertex->r; \ | |
| 562 | x##set##_a_y##set##_c.e[3] = vertex->g; \ | |
| 563 | dup_4x16b(x##set##_b, vertex->x); \ | |
| 564 | dup_4x16b(x##set##_c, vertex->x); \ | |
| 565 | dup_4x16b(y##set##_a, vertex->y); \ | |
| 566 | dup_4x16b(y##set##_b, vertex->y); \ | |
| 567 | x##set##_b.e[0] = vertex->b; \ | |
| 568 | y##set##_b.e[1] = vertex->b \ | |
| 569 | ||
| 570 | ||
| ee060c58 | 571 | void compute_all_gradients(psx_gpu_struct * __restrict__ psx_gpu, |
| 572 | const vertex_struct * __restrict__ a, const vertex_struct * __restrict__ b, | |
| 573 | const vertex_struct * __restrict__ c) | |
| 75e28f62 E |
574 | { |
| 575 | u32 triangle_area = psx_gpu->triangle_area; | |
| 576 | u32 winding_mask_scalar; | |
| 577 | ||
| 578 | u32 triangle_area_shift; | |
| 579 | u64 triangle_area_reciprocal = | |
| 580 | fixed_reciprocal(triangle_area, &triangle_area_shift); | |
| 581 | triangle_area_shift = -(triangle_area_shift - FIXED_BITS); | |
| 582 | ||
| 583 | // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) = | |
| 584 | // ( d0 * d1 ) - ( d2 * d3 ) = | |
| 585 | // ( m0 ) - ( m1 ) = gradient | |
| 586 | ||
| 587 | // This is split to do 12 elements at a time over three sets: a, b, and c. | |
| 588 | // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so | |
| 589 | // two of the slots are unused. | |
| 590 | ||
| 591 | // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as | |
| 592 | // is g. | |
| 593 | ||
| 594 | vec_4x16s x0_a_y0_c, x0_b, x0_c; | |
| 595 | vec_4x16s y0_a, y0_b; | |
| 596 | vec_4x16s x1_a_y1_c, x1_b, x1_c; | |
| 597 | vec_4x16s y1_a, y1_b; | |
| 598 | vec_4x16s x2_a_y2_c, x2_b, x2_c; | |
| 599 | vec_4x16s y2_a, y2_b; | |
| 600 | ||
| 601 | vec_4x32u uvrg_base; | |
| 602 | vec_4x32u b_base; | |
| c6063f89 | 603 | vec_4x32u uvrgb_phase; |
| 75e28f62 E |
604 | |
| 605 | vec_4x16s d0_a_d3_c, d0_b, d0_c; | |
| 606 | vec_4x16s d1_a, d1_b, d1_c_d2_a; | |
| 607 | vec_4x16s d2_b, d2_c; | |
| 608 | vec_4x16s d3_a, d3_b; | |
| 609 | ||
| 610 | vec_4x32s m0_a, m0_b, m0_c; | |
| 611 | vec_4x32s m1_a, m1_b, m1_c; | |
| 612 | ||
| 613 | vec_4x32u gradient_area_a, gradient_area_c; | |
| 614 | vec_2x32u gradient_area_b; | |
| 615 | ||
| 616 | vec_4x32u gradient_area_sign_a, gradient_area_sign_c; | |
| 617 | vec_2x32u gradient_area_sign_b; | |
| 618 | vec_4x32u winding_mask; | |
| 619 | ||
| 620 | vec_2x64u gradient_wide_a0, gradient_wide_a1; | |
| 621 | vec_2x64u gradient_wide_c0, gradient_wide_c1; | |
| 622 | vec_2x64u gradient_wide_b; | |
| 623 | ||
| 624 | vec_4x32u gradient_a, gradient_c; | |
| 625 | vec_2x32u gradient_b; | |
| 626 | vec_16x8s gradient_shift; | |
| 627 | ||
| 628 | setup_gradient_calculation_input(0, a); | |
| 629 | setup_gradient_calculation_input(1, b); | |
| 630 | setup_gradient_calculation_input(2, c); | |
| 631 | ||
| c6063f89 | 632 | dup_4x32b(uvrgb_phase, psx_gpu->uvrgb_phase); |
| 75e28f62 E |
633 | shl_long_4x16b(uvrg_base, x0_a_y0_c, 16); |
| 634 | shl_long_4x16b(b_base, x0_b, 16); | |
| 635 | ||
| c6063f89 | 636 | add_4x32b(uvrg_base, uvrg_base, uvrgb_phase); |
| 637 | add_4x32b(b_base, b_base, uvrgb_phase); | |
| 75e28f62 E |
638 | |
| 639 | // Can probably pair these, but it'll require careful register allocation | |
| 640 | sub_4x16b(d0_a_d3_c, x1_a_y1_c, x0_a_y0_c); | |
| 641 | sub_4x16b(d1_c_d2_a, x2_a_y2_c, x1_a_y1_c); | |
| 642 | ||
| 643 | sub_4x16b(d0_b, x1_b, x0_b); | |
| 644 | sub_4x16b(d0_c, x1_c, x0_c); | |
| 645 | ||
| 646 | sub_4x16b(d1_a, y2_a, y1_a); | |
| 647 | sub_4x16b(d1_b, y2_b, y1_b); | |
| 648 | ||
| 649 | sub_4x16b(d2_b, x2_b, x1_b); | |
| 650 | sub_4x16b(d2_c, x2_c, x1_c); | |
| 651 | ||
| 652 | sub_4x16b(d3_a, y1_a, y0_a); | |
| 653 | sub_4x16b(d3_b, y1_b, y0_b); | |
| 654 | ||
| 655 | mul_long_4x16b(m0_a, d0_a_d3_c, d1_a); | |
| 656 | mul_long_4x16b(m0_b, d0_b, d1_b); | |
| 657 | mul_long_4x16b(m0_c, d0_c, d1_c_d2_a); | |
| 658 | ||
| 659 | mul_long_4x16b(m1_a, d1_c_d2_a, d3_a); | |
| 660 | mul_long_4x16b(m1_b, d2_b, d3_b); | |
| 661 | mul_long_4x16b(m1_c, d2_c, d0_a_d3_c); | |
| 662 | ||
| 663 | sub_4x32b(gradient_area_a, m0_a, m1_a); | |
| 664 | sub_2x32b(gradient_area_b, m0_b.low, m1_b.low); | |
| 665 | sub_4x32b(gradient_area_c, m0_c, m1_c); | |
| 666 | ||
| 667 | cmpltz_4x32b(gradient_area_sign_a, gradient_area_a); | |
| 668 | cmpltz_2x32b(gradient_area_sign_b, gradient_area_b); | |
| 669 | cmpltz_4x32b(gradient_area_sign_c, gradient_area_c); | |
| 670 | ||
| 671 | abs_4x32b(gradient_area_a, gradient_area_a); | |
| 672 | abs_2x32b(gradient_area_b, gradient_area_b); | |
| 673 | abs_4x32b(gradient_area_c, gradient_area_c); | |
| 674 | ||
| 675 | winding_mask_scalar = -psx_gpu->triangle_winding; | |
| 676 | ||
| 677 | dup_4x32b(winding_mask, winding_mask_scalar); | |
| 678 | eor_4x32b(gradient_area_sign_a, gradient_area_sign_a, winding_mask); | |
| 679 | eor_2x32b(gradient_area_sign_b, gradient_area_sign_b, winding_mask); | |
| 680 | eor_4x32b(gradient_area_sign_c, gradient_area_sign_c, winding_mask); | |
| 681 | ||
| 682 | mul_scalar_long_2x32b(gradient_wide_a0, | |
| 683 | vector_cast(vec_2x32s, gradient_area_a.low), | |
| 684 | (s64)triangle_area_reciprocal); | |
| 685 | mul_scalar_long_2x32b(gradient_wide_a1, | |
| 686 | vector_cast(vec_2x32s, gradient_area_a.high), | |
| 687 | (s64)triangle_area_reciprocal); | |
| 688 | mul_scalar_long_2x32b(gradient_wide_b, | |
| 689 | vector_cast(vec_2x32s, gradient_area_b), | |
| 690 | (s64)triangle_area_reciprocal); | |
| 691 | mul_scalar_long_2x32b(gradient_wide_c0, | |
| 692 | vector_cast(vec_2x32s, gradient_area_c.low), | |
| 693 | (s64)triangle_area_reciprocal); | |
| 694 | mul_scalar_long_2x32b(gradient_wide_c1, | |
| 695 | vector_cast(vec_2x32s, gradient_area_c.high), | |
| 696 | (s64)triangle_area_reciprocal); | |
| 697 | ||
| 698 | dup_16x8b(gradient_shift, triangle_area_shift); | |
| 699 | shl_reg_2x64b(gradient_wide_a0, gradient_wide_a0, | |
| 700 | vector_cast(vec_2x64u, gradient_shift)); | |
| 701 | shl_reg_2x64b(gradient_wide_a1, gradient_wide_a1, | |
| 702 | vector_cast(vec_2x64u, gradient_shift)); | |
| 703 | shl_reg_2x64b(gradient_wide_b, gradient_wide_b, | |
| 704 | vector_cast(vec_2x64u, gradient_shift)); | |
| 705 | shl_reg_2x64b(gradient_wide_c0, gradient_wide_c0, | |
| 706 | vector_cast(vec_2x64u, gradient_shift)); | |
| 707 | shl_reg_2x64b(gradient_wide_c1, gradient_wide_c1, | |
| 708 | vector_cast(vec_2x64u, gradient_shift)); | |
| 709 | ||
| 710 | mov_narrow_2x64b(gradient_a.low, gradient_wide_a0); | |
| 711 | mov_narrow_2x64b(gradient_a.high, gradient_wide_a1); | |
| 712 | mov_narrow_2x64b(gradient_b, gradient_wide_b); | |
| 713 | mov_narrow_2x64b(gradient_c.low, gradient_wide_c0); | |
| 714 | mov_narrow_2x64b(gradient_c.high, gradient_wide_c1); | |
| 715 | ||
| 716 | shl_4x32b(gradient_a, gradient_a, 4); | |
| 717 | shl_2x32b(gradient_b, gradient_b, 4); | |
| 718 | shl_4x32b(gradient_c, gradient_c, 4); | |
| 719 | ||
| 720 | eor_4x32b(gradient_a, gradient_a, gradient_area_sign_a); | |
| 721 | eor_2x32b(gradient_b, gradient_b, gradient_area_sign_b); | |
| 722 | eor_4x32b(gradient_c, gradient_c, gradient_area_sign_c); | |
| 723 | ||
| 724 | sub_4x32b(gradient_a, gradient_a, gradient_area_sign_a); | |
| 725 | sub_2x32b(gradient_b, gradient_b, gradient_area_sign_b); | |
| 726 | sub_4x32b(gradient_c, gradient_c, gradient_area_sign_c); | |
| 727 | ||
| 728 | u32 left_adjust = a->x; | |
| 729 | mls_scalar_4x32b(uvrg_base, gradient_a, left_adjust); | |
| 730 | mls_scalar_2x32b(b_base.low, gradient_b, left_adjust); | |
| 731 | ||
| 732 | vec_4x32u uvrg_dx2; | |
| 733 | vec_2x32u b_dx2; | |
| 734 | ||
| 735 | vec_4x32u uvrg_dx3; | |
| 736 | vec_2x32u b_dx3; | |
| 737 | ||
| 738 | vec_4x32u zero; | |
| 739 | ||
| 740 | eor_4x32b(zero, zero, zero); | |
| 741 | add_4x32b(uvrg_dx2, gradient_a, gradient_a); | |
| 742 | add_2x32b(b_dx2, gradient_b, gradient_b); | |
| 743 | add_4x32b(uvrg_dx3, gradient_a, uvrg_dx2); | |
| 744 | add_2x32b(b_dx3, gradient_b, b_dx2); | |
| 745 | ||
| 746 | // Can be done with vst4, assuming that the zero, dx, dx2, and dx3 are | |
| 747 | // lined up properly | |
| 748 | psx_gpu->u_block_span.e[0] = zero.e[0]; | |
| 749 | psx_gpu->u_block_span.e[1] = gradient_a.e[0]; | |
| 750 | psx_gpu->u_block_span.e[2] = uvrg_dx2.e[0]; | |
| 751 | psx_gpu->u_block_span.e[3] = uvrg_dx3.e[0]; | |
| 752 | ||
| 753 | psx_gpu->v_block_span.e[0] = zero.e[1]; | |
| 754 | psx_gpu->v_block_span.e[1] = gradient_a.e[1]; | |
| 755 | psx_gpu->v_block_span.e[2] = uvrg_dx2.e[1]; | |
| 756 | psx_gpu->v_block_span.e[3] = uvrg_dx3.e[1]; | |
| 757 | ||
| 758 | psx_gpu->r_block_span.e[0] = zero.e[2]; | |
| 759 | psx_gpu->r_block_span.e[1] = gradient_a.e[2]; | |
| 760 | psx_gpu->r_block_span.e[2] = uvrg_dx2.e[2]; | |
| 761 | psx_gpu->r_block_span.e[3] = uvrg_dx3.e[2]; | |
| 762 | ||
| 763 | psx_gpu->g_block_span.e[0] = zero.e[3]; | |
| 764 | psx_gpu->g_block_span.e[1] = gradient_a.e[3]; | |
| 765 | psx_gpu->g_block_span.e[2] = uvrg_dx2.e[3]; | |
| 766 | psx_gpu->g_block_span.e[3] = uvrg_dx3.e[3]; | |
| 767 | ||
| 768 | psx_gpu->b_block_span.e[0] = zero.e[0]; | |
| 769 | psx_gpu->b_block_span.e[1] = gradient_b.e[0]; | |
| 770 | psx_gpu->b_block_span.e[2] = b_dx2.e[0]; | |
| 771 | psx_gpu->b_block_span.e[3] = b_dx3.e[0]; | |
| 772 | ||
| 773 | psx_gpu->uvrg = uvrg_base; | |
| 774 | psx_gpu->b = b_base.e[0]; | |
| 775 | ||
| 776 | psx_gpu->uvrg_dx = gradient_a; | |
| 777 | psx_gpu->uvrg_dy = gradient_c; | |
| 778 | psx_gpu->b_dy = gradient_b.e[1]; | |
| 779 | } | |
| 780 | #endif | |
| 781 | ||
| 782 | #define vector_check(_a, _b) \ | |
| 783 | if(memcmp(&_a, &_b, sizeof(_b))) \ | |
| 784 | { \ | |
| 785 | if(sizeof(_b) == 8) \ | |
| 786 | { \ | |
| 787 | printf("mismatch on %s vs %s: (%x %x) vs (%x %x)\n", \ | |
| 788 | #_a, #_b, _a.e[0], _a.e[1], _b.e[0], _b.e[1]); \ | |
| 789 | } \ | |
| 790 | else \ | |
| 791 | { \ | |
| 792 | printf("mismatch on %s vs %s: (%x %x %x %x) vs (%x %x %x %x)\n", \ | |
| 793 | #_a, #_b, _a.e[0], _a.e[1], _a.e[2], _a.e[3], _b.e[0], _b.e[1], \ | |
| 794 | _b.e[2], _b.e[3]); \ | |
| 795 | } \ | |
| 796 | } \ | |
| 797 | ||
| 798 | #define scalar_check(_a, _b) \ | |
| 799 | if(_a != _b) \ | |
| 800 | printf("mismatch on %s %s: %x vs %x\n", #_a, #_b, _a, _b) \ | |
| 801 | ||
| 802 | ||
| 2d658c89 | 803 | #if !defined(NEON_BUILD) && !defined(NDEBUG) |
| 804 | static void setup_spans_debug_check(psx_gpu_struct *psx_gpu, | |
| 805 | edge_data_struct *span_edge_data_element) | |
| 806 | { | |
| 807 | u32 _num_spans = span_edge_data_element - psx_gpu->span_edge_data; | |
| 808 | if (_num_spans > MAX_SPANS) | |
| 809 | *(volatile int *)0 = 1; | |
| 810 | if (_num_spans < psx_gpu->num_spans) | |
| 811 | { | |
| 812 | if(span_edge_data_element->num_blocks > MAX_BLOCKS_PER_ROW) | |
| 813 | *(volatile int *)0 = 2; | |
| 814 | if(span_edge_data_element->y >= 2048) | |
| 815 | *(volatile int *)0 = 3; | |
| 816 | } | |
| 817 | } | |
| c111e8f8 | 818 | #else |
| 2d658c89 | 819 | #define setup_spans_debug_check(psx_gpu, span_edge_data_element) |
| c111e8f8 | 820 | #endif |
| 821 | ||
| 75e28f62 E |
822 | #define setup_spans_prologue_alternate_yes() \ |
| 823 | vec_2x64s alternate_x; \ | |
| 824 | vec_2x64s alternate_dx_dy; \ | |
| 825 | vec_4x32s alternate_x_32; \ | |
| aafce833 | 826 | vec_4x16u alternate_x_16; \ |
| 75e28f62 E |
827 | \ |
| 828 | vec_4x16u alternate_select; \ | |
| 829 | vec_4x16s y_mid_point; \ | |
| 830 | \ | |
| 831 | s32 y_b = v_b->y; \ | |
| 832 | s64 edge_alt; \ | |
| 833 | s32 edge_dx_dy_alt; \ | |
| 834 | u32 edge_shift_alt \ | |
| 835 | ||
| 836 | #define setup_spans_prologue_alternate_no() \ | |
| 837 | ||
| 838 | #define setup_spans_prologue(alternate_active) \ | |
| 839 | edge_data_struct *span_edge_data; \ | |
| 840 | vec_4x32u *span_uvrg_offset; \ | |
| 841 | u32 *span_b_offset; \ | |
| 842 | \ | |
| 843 | s32 clip; \ | |
| 844 | \ | |
| 845 | vec_2x64s edges_xy; \ | |
| 846 | vec_2x32s edges_dx_dy; \ | |
| 847 | vec_2x32u edge_shifts; \ | |
| 848 | \ | |
| 849 | vec_2x64s left_x, right_x; \ | |
| 850 | vec_2x64s left_dx_dy, right_dx_dy; \ | |
| 851 | vec_4x32s left_x_32, right_x_32; \ | |
| 852 | vec_8x16s left_right_x_16; \ | |
| 853 | vec_4x16s y_x4; \ | |
| 854 | vec_8x16s left_edge; \ | |
| 855 | vec_8x16s right_edge; \ | |
| 856 | vec_4x16u span_shift; \ | |
| 857 | \ | |
| 858 | vec_2x32u c_0x01; \ | |
| 859 | vec_4x16u c_0x04; \ | |
| 860 | vec_4x16u c_0xFFFE; \ | |
| 861 | vec_4x16u c_0x07; \ | |
| 862 | \ | |
| 863 | vec_2x32s x_starts; \ | |
| 864 | vec_2x32s x_ends; \ | |
| 865 | \ | |
| 866 | s32 x_a = v_a->x; \ | |
| 867 | s32 x_b = v_b->x; \ | |
| 868 | s32 x_c = v_c->x; \ | |
| 869 | s32 y_a = v_a->y; \ | |
| 870 | s32 y_c = v_c->y; \ | |
| 871 | \ | |
| 872 | vec_4x32u uvrg = psx_gpu->uvrg; \ | |
| 873 | vec_4x32u uvrg_dy = psx_gpu->uvrg_dy; \ | |
| 874 | u32 b = psx_gpu->b; \ | |
| 875 | u32 b_dy = psx_gpu->b_dy; \ | |
| 876 | \ | |
| 877 | dup_2x32b(c_0x01, 0x01); \ | |
| 878 | setup_spans_prologue_alternate_##alternate_active() \ | |
| 879 | ||
| 880 | #define setup_spans_prologue_b() \ | |
| 881 | span_edge_data = psx_gpu->span_edge_data; \ | |
| 882 | span_uvrg_offset = psx_gpu->span_uvrg_offset; \ | |
| 883 | span_b_offset = psx_gpu->span_b_offset; \ | |
| 884 | \ | |
| 885 | vec_8x16u c_0x0001; \ | |
| 2d658c89 | 886 | vec_4x16u c_max_blocks_per_row; \ |
| 75e28f62 E |
887 | \ |
| 888 | dup_8x16b(c_0x0001, 0x0001); \ | |
| 889 | dup_8x16b(left_edge, psx_gpu->viewport_start_x); \ | |
| 890 | dup_8x16b(right_edge, psx_gpu->viewport_end_x); \ | |
| 891 | add_8x16b(right_edge, right_edge, c_0x0001); \ | |
| 892 | dup_4x16b(c_0x04, 0x04); \ | |
| 893 | dup_4x16b(c_0x07, 0x07); \ | |
| 894 | dup_4x16b(c_0xFFFE, 0xFFFE); \ | |
| 2d658c89 | 895 | dup_4x16b(c_max_blocks_per_row, MAX_BLOCKS_PER_ROW); \ |
| 75e28f62 E |
896 | |
| 897 | ||
| 898 | #define compute_edge_delta_x2() \ | |
| 899 | { \ | |
| 900 | vec_2x32s heights; \ | |
| 901 | vec_2x32s height_reciprocals; \ | |
| 902 | vec_2x32s heights_b; \ | |
| 903 | vec_4x32u widths; \ | |
| 904 | \ | |
| 905 | u32 edge_shift = reciprocal_table[height]; \ | |
| 906 | \ | |
| 907 | dup_2x32b(heights, height); \ | |
| 908 | sub_2x32b(widths, x_ends, x_starts); \ | |
| 909 | \ | |
| 910 | dup_2x32b(edge_shifts, edge_shift); \ | |
| 911 | sub_2x32b(heights_b, heights, c_0x01); \ | |
| 7d5140f5 | 912 | shr_2x32b(height_reciprocals, edge_shifts, 10); \ |
| 75e28f62 E |
913 | \ |
| 914 | mla_2x32b(heights_b, x_starts, heights); \ | |
| 915 | bic_immediate_4x16b(vector_cast(vec_4x16u, edge_shifts), 0xE0); \ | |
| 916 | mul_2x32b(edges_dx_dy, widths, height_reciprocals); \ | |
| 917 | mul_long_2x32b(edges_xy, heights_b, height_reciprocals); \ | |
| 918 | } \ | |
| 919 | ||
| 920 | #define compute_edge_delta_x3(start_c, height_a, height_b) \ | |
| 921 | { \ | |
| 922 | vec_2x32s heights; \ | |
| 923 | vec_2x32s height_reciprocals; \ | |
| 924 | vec_2x32s heights_b; \ | |
| 925 | vec_2x32u widths; \ | |
| 926 | \ | |
| 927 | u32 width_alt; \ | |
| 928 | s32 height_b_alt; \ | |
| 929 | u32 height_reciprocal_alt; \ | |
| 930 | \ | |
| 931 | heights.e[0] = height_a; \ | |
| 932 | heights.e[1] = height_b; \ | |
| 933 | \ | |
| 934 | edge_shifts.e[0] = reciprocal_table[height_a]; \ | |
| 935 | edge_shifts.e[1] = reciprocal_table[height_b]; \ | |
| 936 | edge_shift_alt = reciprocal_table[height_minor_b]; \ | |
| 937 | \ | |
| 938 | sub_2x32b(widths, x_ends, x_starts); \ | |
| 939 | width_alt = x_c - start_c; \ | |
| 940 | \ | |
| 7d5140f5 E |
941 | shr_2x32b(height_reciprocals, edge_shifts, 10); \ |
| 942 | height_reciprocal_alt = edge_shift_alt >> 10; \ | |
| 75e28f62 E |
943 | \ |
| 944 | bic_immediate_4x16b(vector_cast(vec_4x16u, edge_shifts), 0xE0); \ | |
| 945 | edge_shift_alt &= 0x1F; \ | |
| 946 | \ | |
| 947 | sub_2x32b(heights_b, heights, c_0x01); \ | |
| 948 | height_b_alt = height_minor_b - 1; \ | |
| 949 | \ | |
| 950 | mla_2x32b(heights_b, x_starts, heights); \ | |
| 951 | height_b_alt += height_minor_b * start_c; \ | |
| 952 | \ | |
| 953 | mul_long_2x32b(edges_xy, heights_b, height_reciprocals); \ | |
| 954 | edge_alt = (s64)height_b_alt * height_reciprocal_alt; \ | |
| 955 | \ | |
| 956 | mul_2x32b(edges_dx_dy, widths, height_reciprocals); \ | |
| 957 | edge_dx_dy_alt = width_alt * height_reciprocal_alt; \ | |
| 958 | } \ | |
| 959 | ||
| 960 | ||
| 961 | #define setup_spans_adjust_y_up() \ | |
| 962 | sub_4x32b(y_x4, y_x4, c_0x04) \ | |
| 963 | ||
| 964 | #define setup_spans_adjust_y_down() \ | |
| 965 | add_4x32b(y_x4, y_x4, c_0x04) \ | |
| 966 | ||
| 967 | #define setup_spans_adjust_interpolants_up() \ | |
| 968 | sub_4x32b(uvrg, uvrg, uvrg_dy); \ | |
| 969 | b -= b_dy \ | |
| 970 | ||
| 971 | #define setup_spans_adjust_interpolants_down() \ | |
| 972 | add_4x32b(uvrg, uvrg, uvrg_dy); \ | |
| 973 | b += b_dy \ | |
| 974 | ||
| 975 | ||
| 976 | #define setup_spans_clip_interpolants_increment() \ | |
| 977 | mla_scalar_4x32b(uvrg, uvrg_dy, clip); \ | |
| 978 | b += b_dy * clip \ | |
| 979 | ||
| 980 | #define setup_spans_clip_interpolants_decrement() \ | |
| 981 | mls_scalar_4x32b(uvrg, uvrg_dy, clip); \ | |
| 982 | b -= b_dy * clip \ | |
| 983 | ||
| 984 | #define setup_spans_clip_alternate_yes() \ | |
| 985 | edge_alt += edge_dx_dy_alt * (s64)(clip) \ | |
| 986 | ||
| 987 | #define setup_spans_clip_alternate_no() \ | |
| 988 | ||
| 989 | #define setup_spans_clip(direction, alternate_active) \ | |
| 990 | { \ | |
| 991 | clipped_triangles++; \ | |
| 992 | mla_scalar_long_2x32b(edges_xy, edges_dx_dy, (s64)clip); \ | |
| 993 | setup_spans_clip_alternate_##alternate_active(); \ | |
| 994 | setup_spans_clip_interpolants_##direction(); \ | |
| 995 | } \ | |
| 996 | ||
| 997 | ||
| 998 | #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \ | |
| 999 | { \ | |
| 1000 | vec_2x64u edge_shifts_64; \ | |
| 1001 | vec_2x64s edges_dx_dy_64; \ | |
| 1002 | \ | |
| 1003 | mov_wide_2x32b(edge_shifts_64, edge_shifts); \ | |
| 1004 | shl_variable_2x64b(edges_xy, edges_xy, edge_shifts_64); \ | |
| 1005 | \ | |
| 1006 | mov_wide_2x32b(edges_dx_dy_64, edges_dx_dy); \ | |
| 1007 | shl_variable_2x64b(edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64); \ | |
| 1008 | \ | |
| 1009 | left_x.e[0] = edges_xy.e[left_index]; \ | |
| 1010 | right_x.e[0] = edges_xy.e[right_index]; \ | |
| 1011 | \ | |
| 1012 | left_dx_dy.e[0] = edges_dx_dy_64.e[left_index]; \ | |
| 1013 | left_dx_dy.e[1] = edges_dx_dy_64.e[left_index]; \ | |
| 1014 | right_dx_dy.e[0] = edges_dx_dy_64.e[right_index]; \ | |
| 1015 | right_dx_dy.e[1] = edges_dx_dy_64.e[right_index]; \ | |
| 1016 | \ | |
| 1017 | add_1x64b(left_x.high, left_x.low, left_dx_dy.low); \ | |
| 1018 | add_1x64b(right_x.high, right_x.low, right_dx_dy.low); \ | |
| 1019 | \ | |
| 1020 | add_2x64b(left_dx_dy, left_dx_dy, left_dx_dy); \ | |
| 1021 | add_2x64b(right_dx_dy, right_dx_dy, right_dx_dy); \ | |
| 1022 | } \ | |
| 1023 | ||
| 1024 | #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \ | |
| 1025 | { \ | |
| 1026 | setup_spans_adjust_edges_alternate_no(left_index, right_index); \ | |
| 1027 | s64 edge_dx_dy_alt_64; \ | |
| 1028 | \ | |
| 1029 | dup_4x16b(y_mid_point, y_b); \ | |
| 1030 | \ | |
| 1031 | edge_alt <<= edge_shift_alt; \ | |
| 1032 | edge_dx_dy_alt_64 = (s64)edge_dx_dy_alt << edge_shift_alt; \ | |
| 1033 | \ | |
| 1034 | alternate_x.e[0] = edge_alt; \ | |
| 1035 | alternate_dx_dy.e[0] = edge_dx_dy_alt_64; \ | |
| 1036 | alternate_dx_dy.e[1] = edge_dx_dy_alt_64; \ | |
| 1037 | \ | |
| 1038 | add_1x64b(alternate_x.high, alternate_x.low, alternate_dx_dy.low); \ | |
| 1039 | add_2x64b(alternate_dx_dy, alternate_dx_dy, alternate_dx_dy); \ | |
| 1040 | } \ | |
| 1041 | ||
| 1042 | ||
| 1043 | #define setup_spans_y_select_up() \ | |
| 1044 | cmplt_4x16b(alternate_select, y_x4, y_mid_point) \ | |
| 1045 | ||
| 1046 | #define setup_spans_y_select_down() \ | |
| 1047 | cmpgt_4x16b(alternate_select, y_x4, y_mid_point) \ | |
| 1048 | ||
| 1049 | #define setup_spans_y_select_alternate_yes(direction) \ | |
| 1050 | setup_spans_y_select_##direction() \ | |
| 1051 | ||
| 1052 | #define setup_spans_y_select_alternate_no(direction) \ | |
| 1053 | ||
| 1054 | #define setup_spans_alternate_select_left() \ | |
| 1055 | bit_4x16b(left_right_x_16.low, alternate_x_16, alternate_select) \ | |
| 1056 | ||
| 1057 | #define setup_spans_alternate_select_right() \ | |
| 1058 | bit_4x16b(left_right_x_16.high, alternate_x_16, alternate_select) \ | |
| 1059 | ||
| 1060 | #define setup_spans_alternate_select_none() \ | |
| 1061 | ||
| 1062 | #define setup_spans_increment_alternate_yes() \ | |
| 1063 | shr_narrow_2x64b(alternate_x_32.low, alternate_x, 32); \ | |
| 1064 | add_2x64b(alternate_x, alternate_x, alternate_dx_dy); \ | |
| 1065 | shr_narrow_2x64b(alternate_x_32.high, alternate_x, 32); \ | |
| 1066 | add_2x64b(alternate_x, alternate_x, alternate_dx_dy); \ | |
| 1067 | mov_narrow_4x32b(alternate_x_16, alternate_x_32) \ | |
| 1068 | ||
| 1069 | #define setup_spans_increment_alternate_no() \ | |
| 1070 | ||
| 1071 | #define setup_spans_set_x4(alternate, direction, alternate_active) \ | |
| 1072 | { \ | |
| 1073 | span_uvrg_offset[0] = uvrg; \ | |
| 1074 | span_b_offset[0] = b; \ | |
| 1075 | setup_spans_adjust_interpolants_##direction(); \ | |
| 1076 | \ | |
| 1077 | span_uvrg_offset[1] = uvrg; \ | |
| 1078 | span_b_offset[1] = b; \ | |
| 1079 | setup_spans_adjust_interpolants_##direction(); \ | |
| 1080 | \ | |
| 1081 | span_uvrg_offset[2] = uvrg; \ | |
| 1082 | span_b_offset[2] = b; \ | |
| 1083 | setup_spans_adjust_interpolants_##direction(); \ | |
| 1084 | \ | |
| 1085 | span_uvrg_offset[3] = uvrg; \ | |
| 1086 | span_b_offset[3] = b; \ | |
| 1087 | setup_spans_adjust_interpolants_##direction(); \ | |
| 1088 | \ | |
| 1089 | span_uvrg_offset += 4; \ | |
| 1090 | span_b_offset += 4; \ | |
| 1091 | \ | |
| 1092 | shr_narrow_2x64b(left_x_32.low, left_x, 32); \ | |
| 1093 | shr_narrow_2x64b(right_x_32.low, right_x, 32); \ | |
| 1094 | \ | |
| 1095 | add_2x64b(left_x, left_x, left_dx_dy); \ | |
| 1096 | add_2x64b(right_x, right_x, right_dx_dy); \ | |
| 1097 | \ | |
| 1098 | shr_narrow_2x64b(left_x_32.high, left_x, 32); \ | |
| 1099 | shr_narrow_2x64b(right_x_32.high, right_x, 32); \ | |
| 1100 | \ | |
| 1101 | add_2x64b(left_x, left_x, left_dx_dy); \ | |
| 1102 | add_2x64b(right_x, right_x, right_dx_dy); \ | |
| 1103 | \ | |
| 1104 | mov_narrow_4x32b(left_right_x_16.low, left_x_32); \ | |
| 1105 | mov_narrow_4x32b(left_right_x_16.high, right_x_32); \ | |
| 1106 | \ | |
| 1107 | setup_spans_increment_alternate_##alternate_active(); \ | |
| 1108 | setup_spans_y_select_alternate_##alternate_active(direction); \ | |
| 1109 | setup_spans_alternate_select_##alternate(); \ | |
| 1110 | \ | |
| 1111 | max_8x16b(left_right_x_16, left_right_x_16, left_edge); \ | |
| 1112 | min_8x16b(left_right_x_16, left_right_x_16, right_edge); \ | |
| 1113 | \ | |
| 1114 | sub_4x16b(left_right_x_16.high, left_right_x_16.high, left_right_x_16.low); \ | |
| 1115 | add_4x16b(left_right_x_16.high, left_right_x_16.high, c_0x07); \ | |
| 1116 | and_4x16b(span_shift, left_right_x_16.high, c_0x07); \ | |
| 1117 | shl_variable_4x16b(span_shift, c_0xFFFE, span_shift); \ | |
| 1118 | shr_4x16b(left_right_x_16.high, left_right_x_16.high, 3); \ | |
| 2d658c89 | 1119 | min_4x16b(left_right_x_16.high, left_right_x_16.high, c_max_blocks_per_row); \ |
| 75e28f62 E |
1120 | \ |
| 1121 | u32 i; \ | |
| 1122 | for(i = 0; i < 4; i++) \ | |
| 1123 | { \ | |
| 1124 | span_edge_data[i].left_x = left_right_x_16.low.e[i]; \ | |
| 1125 | span_edge_data[i].num_blocks = left_right_x_16.high.e[i]; \ | |
| 1126 | span_edge_data[i].right_mask = span_shift.e[i]; \ | |
| 1127 | span_edge_data[i].y = y_x4.e[i]; \ | |
| 2d658c89 | 1128 | setup_spans_debug_check(psx_gpu, &span_edge_data[i]); \ |
| 75e28f62 E |
1129 | } \ |
| 1130 | \ | |
| 1131 | span_edge_data += 4; \ | |
| 1132 | \ | |
| 1133 | setup_spans_adjust_y_##direction(); \ | |
| 1134 | } \ | |
| 1135 | ||
| 1136 | ||
| 1137 | #define setup_spans_alternate_adjust_yes() \ | |
| 1138 | edge_alt -= edge_dx_dy_alt * (s64)height_minor_a \ | |
| 1139 | ||
| 1140 | #define setup_spans_alternate_adjust_no() \ | |
| 1141 | ||
| 1142 | ||
| 1143 | #define setup_spans_down(left_index, right_index, alternate, alternate_active) \ | |
| 1144 | setup_spans_alternate_adjust_##alternate_active(); \ | |
| 1145 | if(y_c > psx_gpu->viewport_end_y) \ | |
| 1146 | height -= y_c - psx_gpu->viewport_end_y - 1; \ | |
| 1147 | \ | |
| 1148 | clip = psx_gpu->viewport_start_y - y_a; \ | |
| 1149 | if(clip > 0) \ | |
| 1150 | { \ | |
| 1151 | height -= clip; \ | |
| 1152 | y_a += clip; \ | |
| 1153 | setup_spans_clip(increment, alternate_active); \ | |
| 1154 | } \ | |
| 1155 | \ | |
| 1156 | setup_spans_prologue_b(); \ | |
| 1157 | \ | |
| 2d658c89 | 1158 | if (height > 512) \ |
| 1159 | height = 512; \ | |
| 1160 | if (height > 0) \ | |
| 75e28f62 E |
1161 | { \ |
| 1162 | y_x4.e[0] = y_a; \ | |
| 1163 | y_x4.e[1] = y_a + 1; \ | |
| 1164 | y_x4.e[2] = y_a + 2; \ | |
| 1165 | y_x4.e[3] = y_a + 3; \ | |
| 1166 | setup_spans_adjust_edges_alternate_##alternate_active(left_index, \ | |
| 1167 | right_index); \ | |
| 1168 | \ | |
| 1169 | psx_gpu->num_spans = height; \ | |
| 1170 | do \ | |
| 1171 | { \ | |
| 1172 | setup_spans_set_x4(alternate, down, alternate_active); \ | |
| 1173 | height -= 4; \ | |
| 1174 | } while(height > 0); \ | |
| ee060c58 | 1175 | if (psx_gpu->hacks_active & (AHACK_TEXTURE_ADJ_U | AHACK_TEXTURE_ADJ_V)) \ |
| 1176 | span_uvrg_offset[height - 1].low = span_uvrg_offset[height - 2].low; \ | |
| 75e28f62 E |
1177 | } \ |
| 1178 | ||
| 1179 | ||
| 1180 | #define setup_spans_alternate_pre_increment_yes() \ | |
| 1181 | edge_alt += edge_dx_dy_alt \ | |
| 1182 | ||
| 1183 | #define setup_spans_alternate_pre_increment_no() \ | |
| 1184 | ||
| 1185 | #define setup_spans_up_decrement_height_yes() \ | |
| 1186 | height-- \ | |
| 1187 | ||
| 1188 | #define setup_spans_up_decrement_height_no() \ | |
| 1189 | {} \ | |
| 1190 | ||
| 1191 | #define setup_spans_up(left_index, right_index, alternate, alternate_active) \ | |
| 1192 | setup_spans_alternate_adjust_##alternate_active(); \ | |
| 1193 | y_a--; \ | |
| 1194 | \ | |
| 1195 | if(y_c < psx_gpu->viewport_start_y) \ | |
| 1196 | height -= psx_gpu->viewport_start_y - y_c; \ | |
| 1197 | else \ | |
| 1198 | setup_spans_up_decrement_height_##alternate_active(); \ | |
| 1199 | \ | |
| 1200 | clip = y_a - psx_gpu->viewport_end_y; \ | |
| 1201 | if(clip > 0) \ | |
| 1202 | { \ | |
| 1203 | height -= clip; \ | |
| 1204 | y_a -= clip; \ | |
| 1205 | setup_spans_clip(decrement, alternate_active); \ | |
| 1206 | } \ | |
| 1207 | \ | |
| 1208 | setup_spans_prologue_b(); \ | |
| 1209 | \ | |
| 2d658c89 | 1210 | if (height > 512) \ |
| 1211 | height = 512; \ | |
| 1212 | if (height > 0) \ | |
| 75e28f62 E |
1213 | { \ |
| 1214 | y_x4.e[0] = y_a; \ | |
| 1215 | y_x4.e[1] = y_a - 1; \ | |
| 1216 | y_x4.e[2] = y_a - 2; \ | |
| 1217 | y_x4.e[3] = y_a - 3; \ | |
| 1218 | add_wide_2x32b(edges_xy, edges_xy, edges_dx_dy); \ | |
| 1219 | setup_spans_alternate_pre_increment_##alternate_active(); \ | |
| 1220 | setup_spans_adjust_edges_alternate_##alternate_active(left_index, \ | |
| 1221 | right_index); \ | |
| 1222 | setup_spans_adjust_interpolants_up(); \ | |
| 1223 | \ | |
| 1224 | psx_gpu->num_spans = height; \ | |
| 1225 | while(height > 0) \ | |
| 1226 | { \ | |
| 1227 | setup_spans_set_x4(alternate, up, alternate_active); \ | |
| 1228 | height -= 4; \ | |
| 1229 | } \ | |
| ee060c58 | 1230 | if (psx_gpu->hacks_active & AHACK_TEXTURE_ADJ_V) \ |
| 1231 | psx_gpu->span_uvrg_offset[0].low = psx_gpu->span_uvrg_offset[1].low; \ | |
| 75e28f62 E |
1232 | } \ |
| 1233 | ||
| 1234 | #define index_left 0 | |
| 1235 | #define index_right 1 | |
| 1236 | ||
| 1237 | #define setup_spans_up_up(minor, major) \ | |
| 1238 | setup_spans_prologue(yes); \ | |
| 1239 | s32 height_minor_a = y_a - y_b; \ | |
| 1240 | s32 height_minor_b = y_b - y_c; \ | |
| 1241 | s32 height = y_a - y_c; \ | |
| 1242 | \ | |
| 1243 | dup_2x32b(x_starts, x_a); \ | |
| 1244 | x_ends.e[0] = x_c; \ | |
| 1245 | x_ends.e[1] = x_b; \ | |
| 1246 | \ | |
| 1247 | compute_edge_delta_x3(x_b, height, height_minor_a); \ | |
| 1248 | setup_spans_up(index_##major, index_##minor, minor, yes) \ | |
| 1249 | ||
| 1250 | ||
| 2bbbb7af | 1251 | #ifndef NEON_BUILD |
| 75e28f62 E |
1252 | |
| 1253 | void setup_spans_up_left(psx_gpu_struct *psx_gpu, vertex_struct *v_a, | |
| 1254 | vertex_struct *v_b, vertex_struct *v_c) | |
| 1255 | { | |
| 1256 | setup_spans_up_up(left, right); | |
| 1257 | } | |
| 1258 | ||
| 1259 | void setup_spans_up_right(psx_gpu_struct *psx_gpu, vertex_struct *v_a, | |
| 1260 | vertex_struct *v_b, vertex_struct *v_c) | |
| 1261 | { | |
| 1262 | setup_spans_up_up(right, left); | |
| 1263 | } | |
| 1264 | ||
| 1265 | #define setup_spans_down_down(minor, major) \ | |
| 1266 | setup_spans_prologue(yes); \ | |
| 1267 | s32 height_minor_a = y_b - y_a; \ | |
| 1268 | s32 height_minor_b = y_c - y_b; \ | |
| 1269 | s32 height = y_c - y_a; \ | |
| 1270 | \ | |
| 1271 | dup_2x32b(x_starts, x_a); \ | |
| 1272 | x_ends.e[0] = x_c; \ | |
| 1273 | x_ends.e[1] = x_b; \ | |
| 1274 | \ | |
| 1275 | compute_edge_delta_x3(x_b, height, height_minor_a); \ | |
| 1276 | setup_spans_down(index_##major, index_##minor, minor, yes) \ | |
| 1277 | ||
| 1278 | void setup_spans_down_left(psx_gpu_struct *psx_gpu, vertex_struct *v_a, | |
| 1279 | vertex_struct *v_b, vertex_struct *v_c) | |
| 1280 | { | |
| 1281 | setup_spans_down_down(left, right); | |
| 1282 | } | |
| 1283 | ||
| 1284 | void setup_spans_down_right(psx_gpu_struct *psx_gpu, vertex_struct *v_a, | |
| 1285 | vertex_struct *v_b, vertex_struct *v_c) | |
| 1286 | { | |
| 1287 | setup_spans_down_down(right, left); | |
| 1288 | } | |
| 1289 | ||
| 1290 | #define setup_spans_up_flat() \ | |
| 1291 | s32 height = y_a - y_c; \ | |
| 1292 | \ | |
| 1293 | flat_triangles++; \ | |
| 1294 | compute_edge_delta_x2(); \ | |
| 1295 | setup_spans_up(index_left, index_right, none, no) \ | |
| 1296 | ||
| 1297 | void setup_spans_up_a(psx_gpu_struct *psx_gpu, vertex_struct *v_a, | |
| 1298 | vertex_struct *v_b, vertex_struct *v_c) | |
| 1299 | { | |
| 1300 | setup_spans_prologue(no); | |
| 1301 | x_starts.e[0] = x_a; | |
| 1302 | x_starts.e[1] = x_b; | |
| 1303 | dup_2x32b(x_ends, x_c); | |
| 1304 | ||
| 1305 | setup_spans_up_flat(); | |
| 1306 | } | |
| 1307 | ||
| 1308 | void setup_spans_up_b(psx_gpu_struct *psx_gpu, vertex_struct *v_a, | |
| 1309 | vertex_struct *v_b, vertex_struct *v_c) | |
| 1310 | { | |
| 1311 | setup_spans_prologue(no); | |
| 1312 | dup_2x32b(x_starts, x_a); | |
| 1313 | x_ends.e[0] = x_b; | |
| 1314 | x_ends.e[1] = x_c; | |
| 1315 | ||
| 1316 | setup_spans_up_flat(); | |
| 1317 | } | |
| 1318 | ||
| 1319 | #define setup_spans_down_flat() \ | |
| 1320 | s32 height = y_c - y_a; \ | |
| 1321 | \ | |
| 1322 | flat_triangles++; \ | |
| 1323 | compute_edge_delta_x2(); \ | |
| 1324 | setup_spans_down(index_left, index_right, none, no) \ | |
| 1325 | ||
| 1326 | void setup_spans_down_a(psx_gpu_struct *psx_gpu, vertex_struct *v_a, | |
| 1327 | vertex_struct *v_b, vertex_struct *v_c) | |
| 1328 | { | |
| 1329 | setup_spans_prologue(no); | |
| 1330 | x_starts.e[0] = x_a; | |
| 1331 | x_starts.e[1] = x_b; | |
| 1332 | dup_2x32b(x_ends, x_c); | |
| 1333 | ||
| 1334 | setup_spans_down_flat(); | |
| 1335 | } | |
| 1336 | ||
| 1337 | void setup_spans_down_b(psx_gpu_struct *psx_gpu, vertex_struct *v_a, | |
| 1338 | vertex_struct *v_b, vertex_struct *v_c) | |
| 1339 | { | |
| 1340 | setup_spans_prologue(no); | |
| 1341 | dup_2x32b(x_starts, x_a); | |
| 1342 | x_ends.e[0] = x_b; | |
| 1343 | x_ends.e[1] = x_c; | |
| 1344 | ||
| 1345 | setup_spans_down_flat(); | |
| 1346 | } | |
| 1347 | ||
| 1348 | void setup_spans_up_down(psx_gpu_struct *psx_gpu, vertex_struct *v_a, | |
| 1349 | vertex_struct *v_b, vertex_struct *v_c) | |
| 1350 | { | |
| 1351 | setup_spans_prologue(no); | |
| 1352 | ||
| 1353 | s32 y_b = v_b->y; | |
| 1354 | s64 edge_alt; | |
| 1355 | s32 edge_dx_dy_alt; | |
| 1356 | u32 edge_shift_alt; | |
| 1357 | ||
| 1358 | s32 middle_y = y_a; | |
| 1359 | s32 height_minor_a = y_a - y_b; | |
| 1360 | s32 height_minor_b = y_c - y_a; | |
| 1361 | s32 height_major = y_c - y_b; | |
| 1362 | ||
| 1363 | vec_2x64s edges_xy_b; | |
| 1364 | vec_2x32s edges_dx_dy_b; | |
| 1365 | vec_2x32u edge_shifts_b; | |
| 1366 | ||
| 1367 | vec_2x32s height_increment; | |
| 1368 | ||
| 1369 | x_starts.e[0] = x_a; | |
| 1370 | x_starts.e[1] = x_c; | |
| 1371 | dup_2x32b(x_ends, x_b); | |
| 1372 | ||
| 1373 | compute_edge_delta_x3(x_a, height_minor_a, height_major); | |
| 1374 | ||
| 1375 | height_increment.e[0] = 0; | |
| 1376 | height_increment.e[1] = height_minor_b; | |
| 1377 | ||
| 1378 | mla_long_2x32b(edges_xy, edges_dx_dy, height_increment); | |
| 1379 | ||
| 1380 | edges_xy_b.e[0] = edge_alt; | |
| 1381 | edges_xy_b.e[1] = edges_xy.e[1]; | |
| 1382 | ||
| 1383 | edge_shifts_b = edge_shifts; | |
| 1384 | edge_shifts_b.e[0] = edge_shift_alt; | |
| 1385 | ||
| 1386 | neg_2x32b(edges_dx_dy_b, edges_dx_dy); | |
| 1387 | edges_dx_dy_b.e[0] = edge_dx_dy_alt; | |
| 1388 | ||
| 1389 | y_a--; | |
| 1390 | ||
| 1391 | if(y_b < psx_gpu->viewport_start_y) | |
| 1392 | height_minor_a -= psx_gpu->viewport_start_y - y_b; | |
| 1393 | ||
| 1394 | clip = y_a - psx_gpu->viewport_end_y; | |
| 1395 | if(clip > 0) | |
| 1396 | { | |
| 1397 | height_minor_a -= clip; | |
| 1398 | y_a -= clip; | |
| 1399 | setup_spans_clip(decrement, no); | |
| 1400 | } | |
| 1401 | ||
| 1402 | setup_spans_prologue_b(); | |
| 1403 | ||
| 2d658c89 | 1404 | if (height_minor_a > 512) |
| 1405 | height_minor_a = 512; | |
| 1406 | if (height_minor_a > 0) | |
| 75e28f62 E |
1407 | { |
| 1408 | y_x4.e[0] = y_a; | |
| 1409 | y_x4.e[1] = y_a - 1; | |
| 1410 | y_x4.e[2] = y_a - 2; | |
| 1411 | y_x4.e[3] = y_a - 3; | |
| 1412 | add_wide_2x32b(edges_xy, edges_xy, edges_dx_dy); | |
| 1413 | setup_spans_adjust_edges_alternate_no(index_left, index_right); | |
| 1414 | setup_spans_adjust_interpolants_up(); | |
| 1415 | ||
| 1416 | psx_gpu->num_spans = height_minor_a; | |
| 1417 | while(height_minor_a > 0) | |
| 1418 | { | |
| 1419 | setup_spans_set_x4(none, up, no); | |
| 1420 | height_minor_a -= 4; | |
| 1421 | } | |
| 1422 | ||
| 1423 | span_edge_data += height_minor_a; | |
| 1424 | span_uvrg_offset += height_minor_a; | |
| 1425 | span_b_offset += height_minor_a; | |
| 1426 | } | |
| 1427 | ||
| 1428 | edges_xy = edges_xy_b; | |
| 1429 | edges_dx_dy = edges_dx_dy_b; | |
| 1430 | edge_shifts = edge_shifts_b; | |
| 1431 | ||
| 1432 | uvrg = psx_gpu->uvrg; | |
| 1433 | b = psx_gpu->b; | |
| 1434 | ||
| 1435 | y_a = middle_y; | |
| 1436 | ||
| 1437 | if(y_c > psx_gpu->viewport_end_y) | |
| 1438 | height_minor_b -= y_c - psx_gpu->viewport_end_y - 1; | |
| 1439 | ||
| 1440 | clip = psx_gpu->viewport_start_y - y_a; | |
| 1441 | if(clip > 0) | |
| 1442 | { | |
| 1443 | height_minor_b -= clip; | |
| 1444 | y_a += clip; | |
| 1445 | setup_spans_clip(increment, no); | |
| 1446 | } | |
| 1447 | ||
| 2d658c89 | 1448 | if (height_minor_b > 512) |
| 1449 | height_minor_b = 512; | |
| 1450 | if (height_minor_b > 0) | |
| 75e28f62 E |
1451 | { |
| 1452 | y_x4.e[0] = y_a; | |
| 1453 | y_x4.e[1] = y_a + 1; | |
| 1454 | y_x4.e[2] = y_a + 2; | |
| 1455 | y_x4.e[3] = y_a + 3; | |
| 1456 | setup_spans_adjust_edges_alternate_no(index_left, index_right); | |
| 1457 | ||
| b7569147 | 1458 | // FIXME: overflow corner case |
| 1459 | if(psx_gpu->num_spans + height_minor_b == MAX_SPANS) | |
| 1460 | height_minor_b &= ~3; | |
| 1461 | ||
| 75e28f62 | 1462 | psx_gpu->num_spans += height_minor_b; |
| b7569147 | 1463 | while(height_minor_b > 0) |
| 75e28f62 E |
1464 | { |
| 1465 | setup_spans_set_x4(none, down, no); | |
| 1466 | height_minor_b -= 4; | |
| b7569147 | 1467 | } |
| ee060c58 | 1468 | if (psx_gpu->hacks_active & (AHACK_TEXTURE_ADJ_U | AHACK_TEXTURE_ADJ_V)) |
| 1469 | { | |
| 1470 | span_uvrg_offset[height_minor_b - 1].low = | |
| 1471 | span_uvrg_offset[height_minor_b - 2].low; | |
| 1472 | } | |
| 75e28f62 E |
1473 | } |
| 1474 | ||
| 1475 | left_split_triangles++; | |
| 1476 | } | |
| 1477 | ||
| 1478 | #endif | |
| 1479 | ||
| ee060c58 | 1480 | // this is some hacky mess, can this be improved somehow? |
| 1481 | // ideally change things to not have to do this hack at all | |
| 1482 | void __attribute__((noinline)) | |
| 1483 | setup_blocks_uv_adj_hack(psx_gpu_struct *psx_gpu, block_struct *block, | |
| 1484 | edge_data_struct *span_edge_data, vec_4x32u *span_uvrg_offset) | |
| 1485 | { | |
| 1486 | size_t span_i = span_uvrg_offset - psx_gpu->span_uvrg_offset; | |
| 1487 | if (span_i != 0 && span_i != psx_gpu->num_spans - 1 | |
| 1488 | && !(psx_gpu->hacks_active & AHACK_TEXTURE_ADJ_U)) | |
| 1489 | return; | |
| 1490 | u32 num_blocks = span_edge_data->num_blocks - 1; | |
| 1491 | s32 offset = __builtin_ctz(span_edge_data->right_mask | 0x100) - 1; | |
| 1492 | s32 toffset = 8 * num_blocks + offset - 1; | |
| 1493 | if (toffset < 0 && !(psx_gpu->hacks_active & AHACK_TEXTURE_ADJ_U)) | |
| 1494 | return; | |
| 1495 | ||
| 1496 | toffset += span_edge_data->left_x; | |
| 1497 | s32 u_dx = psx_gpu->uvrg_dx.low.e[0]; | |
| 1498 | s32 v_dx = psx_gpu->uvrg_dx.low.e[1]; | |
| 1499 | u32 u = span_uvrg_offset->low.e[0]; | |
| 1500 | u32 v = span_uvrg_offset->low.e[1]; | |
| 1501 | u += u_dx * toffset; | |
| 1502 | v += v_dx * toffset; | |
| 1503 | u = (u >> 16) & psx_gpu->texture_mask_width; | |
| 1504 | v = (v >> 16) & psx_gpu->texture_mask_height; | |
| 1505 | if (!(psx_gpu->render_state_base & (TEXTURE_MODE_16BPP << 8))) { | |
| 1506 | // 4bpp 8bpp are swizzled | |
| 1507 | u32 u_ = u; | |
| 1508 | u = (u & 0x0f) | ((v & 0x0f) << 4); | |
| 1509 | v = (v & 0xf0) | (u_ >> 4); | |
| 1510 | } | |
| 1511 | assert(offset >= 0); | |
| 1512 | //assert(block->uv.e[offset] == ((v << 8) | u)); | |
| 1513 | block->uv.e[offset] = (v << 8) | u; | |
| 1514 | } | |
| 75e28f62 E |
1515 | |
| 1516 | #define dither_table_entry_normal(value) \ | |
| 1517 | (value) \ | |
| 1518 | ||
| 1519 | ||
| 1520 | #define setup_blocks_load_msb_mask_indirect() \ | |
| 1521 | ||
| 1522 | #define setup_blocks_load_msb_mask_direct() \ | |
| 1523 | vec_8x16u msb_mask; \ | |
| 1524 | dup_8x16b(msb_mask, psx_gpu->mask_msb); \ | |
| 1525 | ||
| 1526 | ||
| 1527 | #define setup_blocks_variables_shaded_textured(target) \ | |
| 1528 | vec_4x32u u_block; \ | |
| 1529 | vec_4x32u v_block; \ | |
| 1530 | vec_4x32u r_block; \ | |
| 1531 | vec_4x32u g_block; \ | |
| 1532 | vec_4x32u b_block; \ | |
| 1533 | vec_4x32u uvrg_dx = psx_gpu->uvrg_dx; \ | |
| 1534 | vec_4x32u uvrg_dx4; \ | |
| 1535 | vec_4x32u uvrg_dx8; \ | |
| 1536 | vec_4x32u uvrg; \ | |
| 1537 | u32 b_dx = psx_gpu->b_block_span.e[1]; \ | |
| 1538 | u32 b_dx4 = b_dx << 2; \ | |
| 1539 | u32 b_dx8 = b_dx << 3; \ | |
| 1540 | u32 b; \ | |
| 1541 | \ | |
| 1542 | vec_16x8u texture_mask; \ | |
| 1543 | shl_4x32b(uvrg_dx4, uvrg_dx, 2); \ | |
| 1544 | shl_4x32b(uvrg_dx8, uvrg_dx, 3); \ | |
| 1545 | dup_8x8b(texture_mask.low, psx_gpu->texture_mask_width); \ | |
| 1546 | dup_8x8b(texture_mask.high, psx_gpu->texture_mask_height) \ | |
| 1547 | ||
| 1548 | #define setup_blocks_variables_shaded_untextured(target) \ | |
| 1549 | vec_4x32u r_block; \ | |
| 1550 | vec_4x32u g_block; \ | |
| 1551 | vec_4x32u b_block; \ | |
| 1552 | vec_4x32u rgb_dx; \ | |
| 1553 | vec_4x32u rgb_dx4; \ | |
| 1554 | vec_4x32u rgb_dx8; \ | |
| 1555 | vec_4x32u rgb; \ | |
| 1556 | \ | |
| 1557 | vec_8x8u d64_0x07; \ | |
| 1558 | vec_8x8u d64_1; \ | |
| 1559 | vec_8x8u d64_4; \ | |
| 1560 | vec_8x8u d64_128; \ | |
| 1561 | \ | |
| 1562 | dup_8x8b(d64_0x07, 0x07); \ | |
| 1563 | dup_8x8b(d64_1, 1); \ | |
| 1564 | dup_8x8b(d64_4, 4); \ | |
| 1565 | dup_8x8b(d64_128, 128); \ | |
| 1566 | \ | |
| 1567 | rgb_dx.low = psx_gpu->uvrg_dx.high; \ | |
| 1568 | rgb_dx.e[2] = psx_gpu->b_block_span.e[1]; \ | |
| 1569 | shl_4x32b(rgb_dx4, rgb_dx, 2); \ | |
| 1570 | shl_4x32b(rgb_dx8, rgb_dx, 3) \ | |
| 1571 | ||
| 1572 | #define setup_blocks_variables_unshaded_textured(target) \ | |
| 1573 | vec_4x32u u_block; \ | |
| 1574 | vec_4x32u v_block; \ | |
| 1575 | vec_2x32u uv_dx = psx_gpu->uvrg_dx.low; \ | |
| 1576 | vec_2x32u uv_dx4; \ | |
| 1577 | vec_2x32u uv_dx8; \ | |
| 1578 | vec_2x32u uv = psx_gpu->uvrg.low; \ | |
| 1579 | \ | |
| 1580 | vec_16x8u texture_mask; \ | |
| 1581 | shl_2x32b(uv_dx4, uv_dx, 2); \ | |
| 1582 | shl_2x32b(uv_dx8, uv_dx, 3); \ | |
| 1583 | dup_8x8b(texture_mask.low, psx_gpu->texture_mask_width); \ | |
| 1584 | dup_8x8b(texture_mask.high, psx_gpu->texture_mask_height) \ | |
| 1585 | ||
| 1586 | ||
| 1587 | #define setup_blocks_variables_unshaded_untextured_direct() \ | |
| 1588 | or_8x16b(colors, colors, msb_mask) \ | |
| 1589 | ||
| 1590 | #define setup_blocks_variables_unshaded_untextured_indirect() \ | |
| 1591 | ||
| 1592 | #define setup_blocks_variables_unshaded_untextured(target) \ | |
| 1593 | u32 color = psx_gpu->triangle_color; \ | |
| 1594 | vec_8x16u colors; \ | |
| 1595 | \ | |
| 1596 | u32 color_r = color & 0xFF; \ | |
| 1597 | u32 color_g = (color >> 8) & 0xFF; \ | |
| 1598 | u32 color_b = (color >> 16) & 0xFF; \ | |
| 1599 | \ | |
| 1600 | color = (color_r >> 3) | ((color_g >> 3) << 5) | \ | |
| 1601 | ((color_b >> 3) << 10); \ | |
| 1602 | dup_8x16b(colors, color); \ | |
| 1603 | setup_blocks_variables_unshaded_untextured_##target() \ | |
| 1604 | ||
| 1605 | #define setup_blocks_span_initialize_dithered_textured() \ | |
| 1606 | vec_8x16u dither_offsets; \ | |
| 1607 | shl_long_8x8b(dither_offsets, dither_offsets_short, 4) \ | |
| 1608 | ||
| 1609 | #define setup_blocks_span_initialize_dithered_untextured() \ | |
| 1610 | vec_8x8u dither_offsets; \ | |
| 1611 | add_8x8b(dither_offsets, dither_offsets_short, d64_4) \ | |
| 1612 | ||
| 1613 | #define setup_blocks_span_initialize_dithered(texturing) \ | |
| 1614 | u32 dither_row = psx_gpu->dither_table[y & 0x3]; \ | |
| 1615 | u32 dither_shift = (span_edge_data->left_x & 0x3) * 8; \ | |
| 1616 | vec_8x8s dither_offsets_short; \ | |
| 1617 | \ | |
| 1618 | dither_row = \ | |
| f707f14b | 1619 | (dither_row >> dither_shift) | ((u64)dither_row << (32 - dither_shift)); \ |
| 75e28f62 E |
1620 | dup_2x32b(vector_cast(vec_2x32u, dither_offsets_short), dither_row); \ |
| 1621 | setup_blocks_span_initialize_dithered_##texturing() \ | |
| 1622 | ||
| 1623 | #define setup_blocks_span_initialize_undithered(texturing) \ | |
| 1624 | ||
| 1625 | ||
| 1626 | #define setup_blocks_span_initialize_shaded_textured() \ | |
| 1627 | { \ | |
| 1628 | vec_4x32u block_span; \ | |
| 1629 | u32 offset = span_edge_data->left_x; \ | |
| 1630 | \ | |
| 1631 | uvrg = *span_uvrg_offset; \ | |
| 1632 | mla_scalar_4x32b(uvrg, uvrg_dx, offset); \ | |
| 1633 | b = *span_b_offset; \ | |
| 1634 | b += b_dx * offset; \ | |
| 1635 | \ | |
| 1636 | dup_4x32b(u_block, uvrg.e[0]); \ | |
| 1637 | dup_4x32b(v_block, uvrg.e[1]); \ | |
| 1638 | dup_4x32b(r_block, uvrg.e[2]); \ | |
| 1639 | dup_4x32b(g_block, uvrg.e[3]); \ | |
| 1640 | dup_4x32b(b_block, b); \ | |
| 1641 | \ | |
| 1642 | block_span = psx_gpu->u_block_span; \ | |
| 1643 | add_4x32b(u_block, u_block, block_span); \ | |
| 1644 | block_span = psx_gpu->v_block_span; \ | |
| 1645 | add_4x32b(v_block, v_block, block_span); \ | |
| 1646 | block_span = psx_gpu->r_block_span; \ | |
| 1647 | add_4x32b(r_block, r_block, block_span); \ | |
| 1648 | block_span = psx_gpu->g_block_span; \ | |
| 1649 | add_4x32b(g_block, g_block, block_span); \ | |
| 1650 | block_span = psx_gpu->b_block_span; \ | |
| 1651 | add_4x32b(b_block, b_block, block_span); \ | |
| 1652 | } | |
| 1653 | ||
| 1654 | #define setup_blocks_span_initialize_shaded_untextured() \ | |
| 1655 | { \ | |
| 1656 | vec_4x32u block_span; \ | |
| 1657 | u32 offset = span_edge_data->left_x; \ | |
| 1658 | \ | |
| 1659 | rgb.low = span_uvrg_offset->high; \ | |
| 1660 | rgb.high.e[0] = *span_b_offset; \ | |
| 1661 | mla_scalar_4x32b(rgb, rgb_dx, offset); \ | |
| 1662 | \ | |
| 1663 | dup_4x32b(r_block, rgb.e[0]); \ | |
| 1664 | dup_4x32b(g_block, rgb.e[1]); \ | |
| 1665 | dup_4x32b(b_block, rgb.e[2]); \ | |
| 1666 | \ | |
| 1667 | block_span = psx_gpu->r_block_span; \ | |
| 1668 | add_4x32b(r_block, r_block, block_span); \ | |
| 1669 | block_span = psx_gpu->g_block_span; \ | |
| 1670 | add_4x32b(g_block, g_block, block_span); \ | |
| 1671 | block_span = psx_gpu->b_block_span; \ | |
| 1672 | add_4x32b(b_block, b_block, block_span); \ | |
| 1673 | } \ | |
| 1674 | ||
| 1675 | #define setup_blocks_span_initialize_unshaded_textured() \ | |
| 1676 | { \ | |
| 1677 | vec_4x32u block_span; \ | |
| 1678 | u32 offset = span_edge_data->left_x; \ | |
| 1679 | \ | |
| 1680 | uv = span_uvrg_offset->low; \ | |
| 1681 | mla_scalar_2x32b(uv, uv_dx, offset); \ | |
| 1682 | \ | |
| 1683 | dup_4x32b(u_block, uv.e[0]); \ | |
| 1684 | dup_4x32b(v_block, uv.e[1]); \ | |
| 1685 | \ | |
| 1686 | block_span = psx_gpu->u_block_span; \ | |
| 1687 | add_4x32b(u_block, u_block, block_span); \ | |
| 1688 | block_span = psx_gpu->v_block_span; \ | |
| 1689 | add_4x32b(v_block, v_block, block_span); \ | |
| 1690 | } \ | |
| 1691 | ||
| 1692 | #define setup_blocks_span_initialize_unshaded_untextured() \ | |
| 1693 | ||
| 1694 | ||
| 1695 | #define setup_blocks_texture_swizzled() \ | |
| 1696 | { \ | |
| 1697 | vec_8x8u u_saved = u; \ | |
| 1698 | sli_8x8b(u, v, 4); \ | |
| 1699 | sri_8x8b(v, u_saved, 4); \ | |
| 1700 | } \ | |
| 1701 | ||
| 1702 | #define setup_blocks_texture_unswizzled() \ | |
| 1703 | ||
| 1704 | #define setup_blocks_store_shaded_textured(swizzling, dithering, target, \ | |
| 1705 | edge_type) \ | |
| 1706 | { \ | |
| 1707 | vec_8x16u u_whole; \ | |
| 1708 | vec_8x16u v_whole; \ | |
| 1709 | vec_8x16u r_whole; \ | |
| 1710 | vec_8x16u g_whole; \ | |
| 1711 | vec_8x16u b_whole; \ | |
| 1712 | \ | |
| 1713 | vec_8x8u u; \ | |
| 1714 | vec_8x8u v; \ | |
| 1715 | vec_8x8u r; \ | |
| 1716 | vec_8x8u g; \ | |
| 1717 | vec_8x8u b; \ | |
| 1718 | vec_8x16u uv; \ | |
| 1719 | \ | |
| 1720 | vec_4x32u dx4; \ | |
| 1721 | vec_4x32u dx8; \ | |
| 1722 | \ | |
| 1723 | shr_narrow_4x32b(u_whole.low, u_block, 16); \ | |
| 1724 | shr_narrow_4x32b(v_whole.low, v_block, 16); \ | |
| 1725 | shr_narrow_4x32b(r_whole.low, r_block, 16); \ | |
| 1726 | shr_narrow_4x32b(g_whole.low, g_block, 16); \ | |
| 1727 | shr_narrow_4x32b(b_whole.low, b_block, 16); \ | |
| 1728 | \ | |
| 1729 | dup_4x32b(dx4, uvrg_dx4.e[0]); \ | |
| 1730 | add_high_narrow_4x32b(u_whole.high, u_block, dx4); \ | |
| 1731 | dup_4x32b(dx4, uvrg_dx4.e[1]); \ | |
| 1732 | add_high_narrow_4x32b(v_whole.high, v_block, dx4); \ | |
| 1733 | dup_4x32b(dx4, uvrg_dx4.e[2]); \ | |
| 1734 | add_high_narrow_4x32b(r_whole.high, r_block, dx4); \ | |
| 1735 | dup_4x32b(dx4, uvrg_dx4.e[3]); \ | |
| 1736 | add_high_narrow_4x32b(g_whole.high, g_block, dx4); \ | |
| 1737 | dup_4x32b(dx4, b_dx4); \ | |
| 1738 | add_high_narrow_4x32b(b_whole.high, b_block, dx4); \ | |
| 1739 | \ | |
| 1740 | mov_narrow_8x16b(u, u_whole); \ | |
| 1741 | mov_narrow_8x16b(v, v_whole); \ | |
| 1742 | mov_narrow_8x16b(r, r_whole); \ | |
| 1743 | mov_narrow_8x16b(g, g_whole); \ | |
| 1744 | mov_narrow_8x16b(b, b_whole); \ | |
| 1745 | \ | |
| 1746 | dup_4x32b(dx8, uvrg_dx8.e[0]); \ | |
| 1747 | add_4x32b(u_block, u_block, dx8); \ | |
| 1748 | dup_4x32b(dx8, uvrg_dx8.e[1]); \ | |
| 1749 | add_4x32b(v_block, v_block, dx8); \ | |
| 1750 | dup_4x32b(dx8, uvrg_dx8.e[2]); \ | |
| 1751 | add_4x32b(r_block, r_block, dx8); \ | |
| 1752 | dup_4x32b(dx8, uvrg_dx8.e[3]); \ | |
| 1753 | add_4x32b(g_block, g_block, dx8); \ | |
| 1754 | dup_4x32b(dx8, b_dx8); \ | |
| 1755 | add_4x32b(b_block, b_block, dx8); \ | |
| 1756 | \ | |
| 1757 | and_8x8b(u, u, texture_mask.low); \ | |
| 1758 | and_8x8b(v, v, texture_mask.high); \ | |
| 1759 | setup_blocks_texture_##swizzling(); \ | |
| 1760 | \ | |
| 1761 | zip_8x16b(uv, u, v); \ | |
| 1762 | block->uv = uv; \ | |
| 1763 | block->r = r; \ | |
| 1764 | block->g = g; \ | |
| 1765 | block->b = b; \ | |
| 1766 | block->dither_offsets = vector_cast(vec_8x16u, dither_offsets); \ | |
| 1767 | block->fb_ptr = fb_ptr; \ | |
| 1768 | } \ | |
| 1769 | ||
| 1770 | #define setup_blocks_store_unshaded_textured(swizzling, dithering, target, \ | |
| 1771 | edge_type) \ | |
| 1772 | { \ | |
| 1773 | vec_8x16u u_whole; \ | |
| 1774 | vec_8x16u v_whole; \ | |
| 1775 | \ | |
| 1776 | vec_8x8u u; \ | |
| 1777 | vec_8x8u v; \ | |
| 1778 | vec_8x16u uv; \ | |
| 1779 | \ | |
| 1780 | vec_4x32u dx4; \ | |
| 1781 | vec_4x32u dx8; \ | |
| 1782 | \ | |
| 1783 | shr_narrow_4x32b(u_whole.low, u_block, 16); \ | |
| 1784 | shr_narrow_4x32b(v_whole.low, v_block, 16); \ | |
| 1785 | \ | |
| 1786 | dup_4x32b(dx4, uv_dx4.e[0]); \ | |
| 1787 | add_high_narrow_4x32b(u_whole.high, u_block, dx4); \ | |
| 1788 | dup_4x32b(dx4, uv_dx4.e[1]); \ | |
| 1789 | add_high_narrow_4x32b(v_whole.high, v_block, dx4); \ | |
| 1790 | \ | |
| 1791 | mov_narrow_8x16b(u, u_whole); \ | |
| 1792 | mov_narrow_8x16b(v, v_whole); \ | |
| 1793 | \ | |
| 1794 | dup_4x32b(dx8, uv_dx8.e[0]); \ | |
| 1795 | add_4x32b(u_block, u_block, dx8); \ | |
| 1796 | dup_4x32b(dx8, uv_dx8.e[1]); \ | |
| 1797 | add_4x32b(v_block, v_block, dx8); \ | |
| 1798 | \ | |
| 1799 | and_8x8b(u, u, texture_mask.low); \ | |
| 1800 | and_8x8b(v, v, texture_mask.high); \ | |
| 1801 | setup_blocks_texture_##swizzling(); \ | |
| 1802 | \ | |
| 1803 | zip_8x16b(uv, u, v); \ | |
| 1804 | block->uv = uv; \ | |
| 1805 | block->dither_offsets = vector_cast(vec_8x16u, dither_offsets); \ | |
| 1806 | block->fb_ptr = fb_ptr; \ | |
| 1807 | } \ | |
| 1808 | ||
| 1809 | #define setup_blocks_store_shaded_untextured_dithered() \ | |
| 1810 | addq_8x8b(r, r, dither_offsets); \ | |
| 1811 | addq_8x8b(g, g, dither_offsets); \ | |
| 1812 | addq_8x8b(b, b, dither_offsets); \ | |
| 1813 | \ | |
| 1814 | subq_8x8b(r, r, d64_4); \ | |
| 1815 | subq_8x8b(g, g, d64_4); \ | |
| 1816 | subq_8x8b(b, b, d64_4) \ | |
| 1817 | ||
| 1818 | #define setup_blocks_store_shaded_untextured_undithered() \ | |
| 1819 | ||
| 1820 | ||
| 1821 | #define setup_blocks_store_untextured_pixels_indirect_full(_pixels) \ | |
| 1822 | block->pixels = _pixels; \ | |
| 1823 | block->fb_ptr = fb_ptr \ | |
| 1824 | ||
| 1825 | #define setup_blocks_store_untextured_pixels_indirect_edge(_pixels) \ | |
| 1826 | block->pixels = _pixels; \ | |
| 1827 | block->fb_ptr = fb_ptr \ | |
| 1828 | ||
| 1829 | #define setup_blocks_store_shaded_untextured_seed_pixels_indirect() \ | |
| 1830 | mul_long_8x8b(pixels, r, d64_1) \ | |
| 1831 | ||
| 1832 | ||
| 1833 | #define setup_blocks_store_untextured_pixels_direct_full(_pixels) \ | |
| 1834 | store_8x16b(_pixels, fb_ptr) \ | |
| 1835 | ||
| 1836 | #define setup_blocks_store_untextured_pixels_direct_edge(_pixels) \ | |
| 1837 | { \ | |
| 1838 | vec_8x16u fb_pixels; \ | |
| 1839 | vec_8x16u draw_mask; \ | |
| 1840 | vec_8x16u test_mask = psx_gpu->test_mask; \ | |
| 1841 | \ | |
| 1842 | load_8x16b(fb_pixels, fb_ptr); \ | |
| 1843 | dup_8x16b(draw_mask, span_edge_data->right_mask); \ | |
| 1844 | tst_8x16b(draw_mask, draw_mask, test_mask); \ | |
| 1845 | bif_8x16b(fb_pixels, _pixels, draw_mask); \ | |
| 1846 | store_8x16b(fb_pixels, fb_ptr); \ | |
| 1847 | } \ | |
| 1848 | ||
| 1849 | #define setup_blocks_store_shaded_untextured_seed_pixels_direct() \ | |
| 1850 | pixels = msb_mask; \ | |
| 1851 | mla_long_8x8b(pixels, r, d64_1) \ | |
| 1852 | ||
| 1853 | ||
| 1854 | #define setup_blocks_store_shaded_untextured(swizzling, dithering, target, \ | |
| 1855 | edge_type) \ | |
| 1856 | { \ | |
| 1857 | vec_8x16u r_whole; \ | |
| 1858 | vec_8x16u g_whole; \ | |
| 1859 | vec_8x16u b_whole; \ | |
| 1860 | \ | |
| 1861 | vec_8x8u r; \ | |
| 1862 | vec_8x8u g; \ | |
| 1863 | vec_8x8u b; \ | |
| 1864 | \ | |
| 1865 | vec_4x32u dx4; \ | |
| 1866 | vec_4x32u dx8; \ | |
| 1867 | \ | |
| 1868 | vec_8x16u pixels; \ | |
| 1869 | \ | |
| 1870 | shr_narrow_4x32b(r_whole.low, r_block, 16); \ | |
| 1871 | shr_narrow_4x32b(g_whole.low, g_block, 16); \ | |
| 1872 | shr_narrow_4x32b(b_whole.low, b_block, 16); \ | |
| 1873 | \ | |
| 1874 | dup_4x32b(dx4, rgb_dx4.e[0]); \ | |
| 1875 | add_high_narrow_4x32b(r_whole.high, r_block, dx4); \ | |
| 1876 | dup_4x32b(dx4, rgb_dx4.e[1]); \ | |
| 1877 | add_high_narrow_4x32b(g_whole.high, g_block, dx4); \ | |
| 1878 | dup_4x32b(dx4, rgb_dx4.e[2]); \ | |
| 1879 | add_high_narrow_4x32b(b_whole.high, b_block, dx4); \ | |
| 1880 | \ | |
| 1881 | mov_narrow_8x16b(r, r_whole); \ | |
| 1882 | mov_narrow_8x16b(g, g_whole); \ | |
| 1883 | mov_narrow_8x16b(b, b_whole); \ | |
| 1884 | \ | |
| 1885 | dup_4x32b(dx8, rgb_dx8.e[0]); \ | |
| 1886 | add_4x32b(r_block, r_block, dx8); \ | |
| 1887 | dup_4x32b(dx8, rgb_dx8.e[1]); \ | |
| 1888 | add_4x32b(g_block, g_block, dx8); \ | |
| 1889 | dup_4x32b(dx8, rgb_dx8.e[2]); \ | |
| 1890 | add_4x32b(b_block, b_block, dx8); \ | |
| 1891 | \ | |
| 1892 | setup_blocks_store_shaded_untextured_##dithering(); \ | |
| 1893 | \ | |
| 1894 | shr_8x8b(r, r, 3); \ | |
| 1895 | bic_8x8b(g, g, d64_0x07); \ | |
| 1896 | bic_8x8b(b, b, d64_0x07); \ | |
| 1897 | \ | |
| 1898 | setup_blocks_store_shaded_untextured_seed_pixels_##target(); \ | |
| 1899 | mla_long_8x8b(pixels, g, d64_4); \ | |
| 1900 | mla_long_8x8b(pixels, b, d64_128) \ | |
| 1901 | \ | |
| 1902 | setup_blocks_store_untextured_pixels_##target##_##edge_type(pixels); \ | |
| 1903 | } \ | |
| 1904 | ||
| 1905 | #define setup_blocks_store_unshaded_untextured(swizzling, dithering, target, \ | |
| 1906 | edge_type) \ | |
| 1907 | setup_blocks_store_untextured_pixels_##target##_##edge_type(colors) \ | |
| 1908 | ||
| 1909 | ||
| 1910 | #define setup_blocks_store_draw_mask_textured_indirect(_block, bits) \ | |
| 1911 | (_block)->draw_mask_bits = bits \ | |
| 1912 | ||
| 1913 | #define setup_blocks_store_draw_mask_untextured_indirect(_block, bits) \ | |
| 1914 | { \ | |
| 1915 | vec_8x16u bits_mask; \ | |
| 1916 | vec_8x16u test_mask = psx_gpu->test_mask; \ | |
| 1917 | dup_8x16b(bits_mask, bits); \ | |
| 1918 | tst_8x16b(bits_mask, bits_mask, test_mask); \ | |
| 1919 | (_block)->draw_mask = bits_mask; \ | |
| 1920 | } \ | |
| 1921 | ||
| 1922 | #define setup_blocks_store_draw_mask_untextured_direct(_block, bits) \ | |
| 1923 | ||
| ee060c58 | 1924 | #define setup_blocks_uv_adj_hack_untextured(_block, edge_data, uvrg_offset) \ |
| 1925 | ||
| 1926 | #define setup_blocks_uv_adj_hack_textured(_block, edge_data, uvrg_offset) \ | |
| 1927 | { \ | |
| 1928 | u32 m_ = AHACK_TEXTURE_ADJ_U | AHACK_TEXTURE_ADJ_V; \ | |
| 1929 | if (unlikely(psx_gpu->hacks_active & m_)) \ | |
| 1930 | setup_blocks_uv_adj_hack(psx_gpu, _block, edge_data, uvrg_offset); \ | |
| 1931 | } \ | |
| 75e28f62 E |
1932 | |
| 1933 | #define setup_blocks_add_blocks_indirect() \ | |
| 1934 | num_blocks += span_num_blocks; \ | |
| 1935 | \ | |
| 1936 | if(num_blocks > MAX_BLOCKS) \ | |
| 1937 | { \ | |
| 1938 | psx_gpu->num_blocks = num_blocks - span_num_blocks; \ | |
| 1939 | flush_render_block_buffer(psx_gpu); \ | |
| 1940 | num_blocks = span_num_blocks; \ | |
| 1941 | block = psx_gpu->blocks; \ | |
| 1942 | } \ | |
| 1943 | ||
| 1944 | #define setup_blocks_add_blocks_direct() \ | |
| a2cb152a | 1945 | stats_add(texel_blocks_untextured, span_num_blocks); \ |
| 95bcdd38 | 1946 | stats_add(span_pixel_blocks, span_num_blocks); \ |
| 75e28f62 E |
1947 | |
| 1948 | ||
| 1949 | #define setup_blocks_builder(shading, texturing, dithering, sw, target) \ | |
| 1950 | void setup_blocks_##shading##_##texturing##_##dithering##_##sw##_##target( \ | |
| 1951 | psx_gpu_struct *psx_gpu) \ | |
| 1952 | { \ | |
| 1953 | setup_blocks_load_msb_mask_##target(); \ | |
| 1954 | setup_blocks_variables_##shading##_##texturing(target); \ | |
| 1955 | \ | |
| 1956 | edge_data_struct *span_edge_data = psx_gpu->span_edge_data; \ | |
| 1957 | vec_4x32u *span_uvrg_offset = psx_gpu->span_uvrg_offset; \ | |
| 1958 | u32 *span_b_offset = psx_gpu->span_b_offset; \ | |
| 1959 | \ | |
| 1960 | block_struct *block = psx_gpu->blocks + psx_gpu->num_blocks; \ | |
| 1961 | \ | |
| 1962 | u32 num_spans = psx_gpu->num_spans; \ | |
| 1963 | \ | |
| 1964 | u16 *fb_ptr; \ | |
| 1965 | u32 y; \ | |
| 1966 | \ | |
| 1967 | u32 num_blocks = psx_gpu->num_blocks; \ | |
| 1968 | u32 span_num_blocks; \ | |
| 1969 | \ | |
| 1970 | while(num_spans) \ | |
| 1971 | { \ | |
| 1972 | span_num_blocks = span_edge_data->num_blocks; \ | |
| 1973 | if(span_num_blocks) \ | |
| 1974 | { \ | |
| 1975 | y = span_edge_data->y; \ | |
| c1817bd9 | 1976 | fb_ptr = psx_gpu->vram_out_ptr + span_edge_data->left_x + (y * 1024); \ |
| 75e28f62 E |
1977 | \ |
| 1978 | setup_blocks_span_initialize_##shading##_##texturing(); \ | |
| 1979 | setup_blocks_span_initialize_##dithering(texturing); \ | |
| 1980 | \ | |
| 1981 | setup_blocks_add_blocks_##target(); \ | |
| 1982 | \ | |
| 1983 | s32 pixel_span = span_num_blocks * 8; \ | |
| 1984 | pixel_span -= __builtin_popcount(span_edge_data->right_mask & 0xFF); \ | |
| 1985 | span_pixels += pixel_span; \ | |
| 75e28f62 E |
1986 | \ |
| 1987 | span_num_blocks--; \ | |
| 1988 | while(span_num_blocks) \ | |
| 1989 | { \ | |
| 1990 | setup_blocks_store_##shading##_##texturing(sw, dithering, target, \ | |
| 1991 | full); \ | |
| 1992 | setup_blocks_store_draw_mask_##texturing##_##target(block, 0x00); \ | |
| 1993 | \ | |
| 1994 | fb_ptr += 8; \ | |
| 1995 | block++; \ | |
| 1996 | span_num_blocks--; \ | |
| 1997 | } \ | |
| 1998 | \ | |
| 1999 | setup_blocks_store_##shading##_##texturing(sw, dithering, target, edge); \ | |
| 2000 | setup_blocks_store_draw_mask_##texturing##_##target(block, \ | |
| 2001 | span_edge_data->right_mask); \ | |
| ee060c58 | 2002 | setup_blocks_uv_adj_hack_##texturing(block, span_edge_data, \ |
| 2003 | span_uvrg_offset); \ | |
| 75e28f62 E |
2004 | \ |
| 2005 | block++; \ | |
| 2006 | } \ | |
| 2007 | else \ | |
| 2008 | { \ | |
| 2009 | zero_block_spans++; \ | |
| 2010 | } \ | |
| 2011 | \ | |
| 2012 | num_spans--; \ | |
| 2013 | span_edge_data++; \ | |
| 2014 | span_uvrg_offset++; \ | |
| 2015 | span_b_offset++; \ | |
| 2016 | } \ | |
| 2017 | \ | |
| 2018 | psx_gpu->num_blocks = num_blocks; \ | |
| 2019 | } \ | |
| 2020 | ||
| 75e28f62 E |
2021 | |
| 2022 | //setup_blocks_builder(unshaded, untextured, undithered, unswizzled, direct); | |
| 2023 | ||
| 2bbbb7af | 2024 | #ifndef NEON_BUILD |
| 75e28f62 E |
2025 | |
| 2026 | setup_blocks_builder(shaded, textured, dithered, swizzled, indirect); | |
| 2027 | setup_blocks_builder(shaded, textured, dithered, unswizzled, indirect); | |
| 2028 | ||
| 2029 | setup_blocks_builder(unshaded, textured, dithered, unswizzled, indirect); | |
| 2030 | setup_blocks_builder(unshaded, textured, dithered, swizzled, indirect); | |
| 2031 | ||
| 2032 | setup_blocks_builder(shaded, untextured, undithered, unswizzled, indirect); | |
| 2033 | setup_blocks_builder(shaded, untextured, dithered, unswizzled, indirect); | |
| 2034 | setup_blocks_builder(shaded, untextured, undithered, unswizzled, direct); | |
| 2035 | setup_blocks_builder(shaded, untextured, dithered, unswizzled, direct); | |
| 2036 | ||
| 2037 | setup_blocks_builder(unshaded, untextured, undithered, unswizzled, indirect); | |
| 2038 | setup_blocks_builder(unshaded, untextured, undithered, unswizzled, direct); | |
| 2039 | ||
| 75e28f62 E |
2040 | void texture_blocks_untextured(psx_gpu_struct *psx_gpu) |
| 2041 | { | |
| 2042 | if(psx_gpu->primitive_type != PRIMITIVE_TYPE_SPRITE) | |
| a2cb152a | 2043 | stats_add(texel_blocks_untextured, psx_gpu->num_blocks); |
| 75e28f62 E |
2044 | } |
| 2045 | ||
| 2046 | void texture_blocks_4bpp(psx_gpu_struct *psx_gpu) | |
| 2047 | { | |
| 2048 | block_struct *block = psx_gpu->blocks; | |
| 2049 | u32 num_blocks = psx_gpu->num_blocks; | |
| a2cb152a | 2050 | stats_add(texel_blocks_4bpp, num_blocks); |
| 75e28f62 E |
2051 | |
| 2052 | vec_8x8u texels_low; | |
| 2053 | vec_8x8u texels_high; | |
| 2054 | vec_8x8u texels; | |
| 2055 | vec_8x16u pixels; | |
| 2056 | ||
| 2057 | vec_8x16u clut_a; | |
| 2058 | vec_8x16u clut_b; | |
| 2059 | vec_16x8u clut_low; | |
| 2060 | vec_16x8u clut_high; | |
| 2061 | ||
| 2062 | u8 *texture_ptr_8bpp = psx_gpu->texture_page_ptr; | |
| 2063 | u16 *clut_ptr = psx_gpu->clut_ptr; | |
| 2064 | ||
| 2065 | // Can be done with one deinterleaving load on NEON | |
| 2066 | load_8x16b(clut_a, clut_ptr); | |
| 2067 | load_8x16b(clut_b, clut_ptr + 8); | |
| 2068 | unzip_16x8b(clut_low, clut_high, clut_a, clut_b); | |
| 2069 | ||
| 2070 | if(psx_gpu->current_texture_mask & psx_gpu->dirty_textures_4bpp_mask) | |
| 2071 | update_texture_4bpp_cache(psx_gpu); | |
| 2072 | ||
| 2073 | while(num_blocks) | |
| 2074 | { | |
| 2075 | texels.e[0] = texture_ptr_8bpp[block->uv.e[0]]; | |
| 2076 | texels.e[1] = texture_ptr_8bpp[block->uv.e[1]]; | |
| 2077 | texels.e[2] = texture_ptr_8bpp[block->uv.e[2]]; | |
| 2078 | texels.e[3] = texture_ptr_8bpp[block->uv.e[3]]; | |
| 2079 | texels.e[4] = texture_ptr_8bpp[block->uv.e[4]]; | |
| 2080 | texels.e[5] = texture_ptr_8bpp[block->uv.e[5]]; | |
| 2081 | texels.e[6] = texture_ptr_8bpp[block->uv.e[6]]; | |
| 2082 | texels.e[7] = texture_ptr_8bpp[block->uv.e[7]]; | |
| 2083 | ||
| 2084 | tbl_16(texels_low, texels, clut_low); | |
| 2085 | tbl_16(texels_high, texels, clut_high); | |
| 2086 | ||
| 2087 | // Can be done with an interleaving store on NEON | |
| 2088 | zip_8x16b(pixels, texels_low, texels_high); | |
| 2089 | ||
| 2090 | block->texels = pixels; | |
| 2091 | ||
| 2092 | num_blocks--; | |
| 2093 | block++; | |
| 2094 | } | |
| 2095 | } | |
| 2096 | ||
| 2097 | void texture_blocks_8bpp(psx_gpu_struct *psx_gpu) | |
| 2098 | { | |
| 2099 | block_struct *block = psx_gpu->blocks; | |
| 2100 | u32 num_blocks = psx_gpu->num_blocks; | |
| 2101 | ||
| a2cb152a | 2102 | stats_add(texel_blocks_8bpp, num_blocks); |
| 75e28f62 E |
2103 | |
| 2104 | if(psx_gpu->current_texture_mask & psx_gpu->dirty_textures_8bpp_mask) | |
| 2105 | update_texture_8bpp_cache(psx_gpu); | |
| 2106 | ||
| 2107 | vec_8x16u texels; | |
| 2108 | u8 *texture_ptr_8bpp = psx_gpu->texture_page_ptr; | |
| 2109 | ||
| 2110 | u32 texel; | |
| 2111 | u32 offset; | |
| 2112 | u32 i; | |
| 2113 | ||
| 2114 | while(num_blocks) | |
| 2115 | { | |
| 2116 | for(i = 0; i < 8; i++) | |
| 2117 | { | |
| 2118 | offset = block->uv.e[i]; | |
| 2119 | ||
| 2120 | texel = texture_ptr_8bpp[offset]; | |
| 2121 | texels.e[i] = psx_gpu->clut_ptr[texel]; | |
| 2122 | } | |
| 2123 | ||
| 2124 | block->texels = texels; | |
| 2125 | ||
| 2126 | num_blocks--; | |
| 2127 | block++; | |
| 2128 | } | |
| 2129 | } | |
| 2130 | ||
| 2131 | void texture_blocks_16bpp(psx_gpu_struct *psx_gpu) | |
| 2132 | { | |
| 2133 | block_struct *block = psx_gpu->blocks; | |
| 2134 | u32 num_blocks = psx_gpu->num_blocks; | |
| 2135 | ||
| a2cb152a | 2136 | stats_add(texel_blocks_16bpp, num_blocks); |
| 75e28f62 E |
2137 | |
| 2138 | vec_8x16u texels; | |
| 2139 | ||
| 2140 | u16 *texture_ptr_16bpp = psx_gpu->texture_page_ptr; | |
| 2141 | u32 offset; | |
| 2142 | u32 i; | |
| 2143 | ||
| 2144 | while(num_blocks) | |
| 2145 | { | |
| 2146 | for(i = 0; i < 8; i++) | |
| 2147 | { | |
| 2148 | offset = block->uv.e[i]; | |
| 2149 | offset += ((offset & 0xFF00) * 3); | |
| 2150 | ||
| 2151 | texels.e[i] = texture_ptr_16bpp[offset]; | |
| 2152 | } | |
| 2153 | ||
| 2154 | block->texels = texels; | |
| 2155 | ||
| 2156 | num_blocks--; | |
| 2157 | block++; | |
| 2158 | } | |
| 2159 | } | |
| 2160 | ||
| 2161 | #endif | |
| 2162 | ||
| 2163 | ||
| 2164 | #define shade_blocks_load_msb_mask_indirect() \ | |
| 2165 | ||
| 2166 | #define shade_blocks_load_msb_mask_direct() \ | |
| 2167 | vec_8x16u msb_mask; \ | |
| 2168 | dup_8x16b(msb_mask, psx_gpu->mask_msb); \ | |
| 2169 | ||
| 2170 | #define shade_blocks_store_indirect(_draw_mask, _pixels) \ | |
| 2171 | block->draw_mask = _draw_mask; \ | |
| 2172 | block->pixels = _pixels \ | |
| 2173 | ||
| 2174 | #define shade_blocks_store_direct(_draw_mask, _pixels) \ | |
| 2175 | { \ | |
| 2176 | vec_8x16u fb_pixels; \ | |
| 2177 | or_8x16b(_pixels, _pixels, msb_mask); \ | |
| 2178 | load_8x16b(fb_pixels, block->fb_ptr); \ | |
| 2179 | bif_8x16b(fb_pixels, _pixels, _draw_mask); \ | |
| 2180 | store_8x16b(fb_pixels, block->fb_ptr); \ | |
| 2181 | } \ | |
| 2182 | ||
| 2183 | ||
| 3867c6ef | 2184 | #define shade_blocks_textured_false_modulated_check_dithered(target) \ |
| b7ed0632 E |
2185 | if(psx_gpu->triangle_color == 0x808080) \ |
| 2186 | { \ | |
| 2187 | false_modulated_blocks += num_blocks; \ | |
| 2188 | } \ | |
| 3867c6ef E |
2189 | |
| 2190 | #define shade_blocks_textured_false_modulated_check_undithered(target) \ | |
| 2191 | if(psx_gpu->triangle_color == 0x808080) \ | |
| 2192 | { \ | |
| 2193 | \ | |
| 2194 | shade_blocks_textured_unmodulated_##target(psx_gpu); \ | |
| 2195 | false_modulated_blocks += num_blocks; \ | |
| 2196 | return; \ | |
| 2197 | } \ | |
| 2198 | ||
| 2199 | ||
| 2200 | #define shade_blocks_textured_modulated_shaded_primitive_load(dithering, \ | |
| 2201 | target) \ | |
| 75e28f62 | 2202 | |
| 3867c6ef E |
2203 | #define shade_blocks_textured_modulated_unshaded_primitive_load(dithering, \ |
| 2204 | target) \ | |
| 75e28f62 E |
2205 | { \ |
| 2206 | u32 color = psx_gpu->triangle_color; \ | |
| 2207 | dup_8x8b(colors_r, color); \ | |
| 2208 | dup_8x8b(colors_g, color >> 8); \ | |
| 2209 | dup_8x8b(colors_b, color >> 16); \ | |
| 3867c6ef | 2210 | shade_blocks_textured_false_modulated_check_##dithering(target); \ |
| 75e28f62 E |
2211 | } \ |
| 2212 | ||
| 2213 | #define shade_blocks_textured_modulated_shaded_block_load() \ | |
| 2214 | colors_r = block->r; \ | |
| 2215 | colors_g = block->g; \ | |
| 2216 | colors_b = block->b \ | |
| 2217 | ||
| 2218 | #define shade_blocks_textured_modulated_unshaded_block_load() \ | |
| 2219 | ||
| 2220 | #define shade_blocks_textured_modulate_dithered(component) \ | |
| 2221 | pixels_##component = block->dither_offsets; \ | |
| 2222 | mla_long_8x8b(pixels_##component, texels_##component, colors_##component) \ | |
| 2223 | ||
| 2224 | #define shade_blocks_textured_modulate_undithered(component) \ | |
| 2225 | mul_long_8x8b(pixels_##component, texels_##component, colors_##component) \ | |
| 2226 | ||
| 2227 | #define shade_blocks_textured_modulated_builder(shading, dithering, target) \ | |
| 2228 | void shade_blocks_##shading##_textured_modulated_##dithering##_##target( \ | |
| 2229 | psx_gpu_struct *psx_gpu) \ | |
| 2230 | { \ | |
| 2231 | block_struct *block = psx_gpu->blocks; \ | |
| 2232 | u32 num_blocks = psx_gpu->num_blocks; \ | |
| 2233 | vec_8x16u texels; \ | |
| 2234 | \ | |
| 2235 | vec_8x8u texels_r; \ | |
| 2236 | vec_8x8u texels_g; \ | |
| 2237 | vec_8x8u texels_b; \ | |
| 2238 | \ | |
| 2239 | vec_8x8u colors_r; \ | |
| 2240 | vec_8x8u colors_g; \ | |
| 2241 | vec_8x8u colors_b; \ | |
| 2242 | \ | |
| 2243 | vec_8x8u pixels_r_low; \ | |
| 2244 | vec_8x8u pixels_g_low; \ | |
| 2245 | vec_8x8u pixels_b_low; \ | |
| 2246 | vec_8x16u pixels; \ | |
| 2247 | \ | |
| 2248 | vec_8x16u pixels_r; \ | |
| 2249 | vec_8x16u pixels_g; \ | |
| 2250 | vec_8x16u pixels_b; \ | |
| 2251 | \ | |
| 2252 | vec_8x16u draw_mask; \ | |
| 2253 | vec_8x16u zero_mask; \ | |
| 2254 | \ | |
| 2255 | vec_8x8u d64_0x07; \ | |
| 2256 | vec_8x8u d64_0x1F; \ | |
| 2257 | vec_8x8u d64_1; \ | |
| 2258 | vec_8x8u d64_4; \ | |
| 2259 | vec_8x8u d64_128; \ | |
| 2260 | \ | |
| 2261 | vec_8x16u d128_0x8000; \ | |
| 2262 | \ | |
| 2263 | vec_8x16u test_mask = psx_gpu->test_mask; \ | |
| 2264 | u32 draw_mask_bits; \ | |
| 2265 | shade_blocks_load_msb_mask_##target(); \ | |
| 2266 | \ | |
| 2267 | dup_8x8b(d64_0x07, 0x07); \ | |
| 2268 | dup_8x8b(d64_0x1F, 0x1F); \ | |
| 2269 | dup_8x8b(d64_1, 1); \ | |
| 2270 | dup_8x8b(d64_4, 4); \ | |
| 2271 | dup_8x8b(d64_128, 128); \ | |
| 2272 | \ | |
| 2273 | dup_8x16b(d128_0x8000, 0x8000); \ | |
| 2274 | \ | |
| 3867c6ef E |
2275 | shade_blocks_textured_modulated_##shading##_primitive_load(dithering, \ |
| 2276 | target); \ | |
| 75e28f62 E |
2277 | \ |
| 2278 | while(num_blocks) \ | |
| 2279 | { \ | |
| 2280 | draw_mask_bits = block->draw_mask_bits; \ | |
| 2281 | dup_8x16b(draw_mask, draw_mask_bits); \ | |
| 2282 | tst_8x16b(draw_mask, draw_mask, test_mask); \ | |
| 2283 | \ | |
| 2284 | shade_blocks_textured_modulated_##shading##_block_load(); \ | |
| 2285 | \ | |
| 2286 | texels = block->texels; \ | |
| 2287 | \ | |
| 2288 | mov_narrow_8x16b(texels_r, texels); \ | |
| 2289 | shr_narrow_8x16b(texels_g, texels, 5); \ | |
| 2290 | shr_narrow_8x16b(texels_b, texels, 7); \ | |
| 2291 | \ | |
| 2292 | and_8x8b(texels_r, texels_r, d64_0x1F); \ | |
| 2293 | and_8x8b(texels_g, texels_g, d64_0x1F); \ | |
| 2294 | shr_8x8b(texels_b, texels_b, 3); \ | |
| 2295 | \ | |
| 2296 | shade_blocks_textured_modulate_##dithering(r); \ | |
| 2297 | shade_blocks_textured_modulate_##dithering(g); \ | |
| 2298 | shade_blocks_textured_modulate_##dithering(b); \ | |
| 2299 | \ | |
| 2300 | cmpeqz_8x16b(zero_mask, texels); \ | |
| 2301 | and_8x16b(pixels, texels, d128_0x8000); \ | |
| 2302 | \ | |
| 2303 | shrq_narrow_signed_8x16b(pixels_r_low, pixels_r, 4); \ | |
| 2304 | shrq_narrow_signed_8x16b(pixels_g_low, pixels_g, 4); \ | |
| 2305 | shrq_narrow_signed_8x16b(pixels_b_low, pixels_b, 4); \ | |
| 2306 | \ | |
| 2307 | or_8x16b(zero_mask, draw_mask, zero_mask); \ | |
| 2308 | \ | |
| 2309 | shr_8x8b(pixels_r_low, pixels_r_low, 3); \ | |
| 2310 | bic_8x8b(pixels_g_low, pixels_g_low, d64_0x07); \ | |
| 2311 | bic_8x8b(pixels_b_low, pixels_b_low, d64_0x07); \ | |
| 2312 | \ | |
| 2313 | mla_long_8x8b(pixels, pixels_r_low, d64_1); \ | |
| 2314 | mla_long_8x8b(pixels, pixels_g_low, d64_4); \ | |
| 2315 | mla_long_8x8b(pixels, pixels_b_low, d64_128); \ | |
| 2316 | \ | |
| 2317 | shade_blocks_store_##target(zero_mask, pixels); \ | |
| 2318 | \ | |
| 2319 | num_blocks--; \ | |
| 2320 | block++; \ | |
| 2321 | } \ | |
| 2322 | } \ | |
| 2323 | ||
| 2bbbb7af | 2324 | #ifndef NEON_BUILD |
| 75e28f62 E |
2325 | |
| 2326 | shade_blocks_textured_modulated_builder(shaded, dithered, direct); | |
| 2327 | shade_blocks_textured_modulated_builder(shaded, undithered, direct); | |
| 2328 | shade_blocks_textured_modulated_builder(unshaded, dithered, direct); | |
| 2329 | shade_blocks_textured_modulated_builder(unshaded, undithered, direct); | |
| 2330 | ||
| 2331 | shade_blocks_textured_modulated_builder(shaded, dithered, indirect); | |
| 2332 | shade_blocks_textured_modulated_builder(shaded, undithered, indirect); | |
| 2333 | shade_blocks_textured_modulated_builder(unshaded, dithered, indirect); | |
| 2334 | shade_blocks_textured_modulated_builder(unshaded, undithered, indirect); | |
| 2335 | ||
| 2336 | #endif | |
| 2337 | ||
| 2338 | ||
| 2339 | #define shade_blocks_textured_unmodulated_builder(target) \ | |
| 2340 | void shade_blocks_textured_unmodulated_##target(psx_gpu_struct *psx_gpu) \ | |
| 2341 | { \ | |
| 2342 | block_struct *block = psx_gpu->blocks; \ | |
| 2343 | u32 num_blocks = psx_gpu->num_blocks; \ | |
| 2344 | vec_8x16u draw_mask; \ | |
| 2345 | vec_8x16u test_mask = psx_gpu->test_mask; \ | |
| 2346 | u32 draw_mask_bits; \ | |
| 2347 | \ | |
| 2348 | vec_8x16u pixels; \ | |
| 2349 | shade_blocks_load_msb_mask_##target(); \ | |
| 2350 | \ | |
| 2351 | while(num_blocks) \ | |
| 2352 | { \ | |
| 2353 | vec_8x16u zero_mask; \ | |
| 2354 | \ | |
| 2355 | draw_mask_bits = block->draw_mask_bits; \ | |
| 2356 | dup_8x16b(draw_mask, draw_mask_bits); \ | |
| 2357 | tst_8x16b(draw_mask, draw_mask, test_mask); \ | |
| 2358 | \ | |
| 2359 | pixels = block->texels; \ | |
| 2360 | \ | |
| 2361 | cmpeqz_8x16b(zero_mask, pixels); \ | |
| 2362 | or_8x16b(zero_mask, draw_mask, zero_mask); \ | |
| 2363 | \ | |
| 2364 | shade_blocks_store_##target(zero_mask, pixels); \ | |
| 2365 | \ | |
| 2366 | num_blocks--; \ | |
| 2367 | block++; \ | |
| 2368 | } \ | |
| 2369 | } \ | |
| 2370 | ||
| 3867c6ef E |
2371 | #define shade_blocks_textured_unmodulated_dithered_builder(target) \ |
| 2372 | void shade_blocks_textured_unmodulated_dithered_##target(psx_gpu_struct \ | |
| 2373 | *psx_gpu) \ | |
| 2374 | { \ | |
| 2375 | block_struct *block = psx_gpu->blocks; \ | |
| 2376 | u32 num_blocks = psx_gpu->num_blocks; \ | |
| 2377 | vec_8x16u draw_mask; \ | |
| 2378 | vec_8x16u test_mask = psx_gpu->test_mask; \ | |
| 2379 | u32 draw_mask_bits; \ | |
| 2380 | \ | |
| 2381 | vec_8x16u pixels; \ | |
| 2382 | shade_blocks_load_msb_mask_##target(); \ | |
| 2383 | \ | |
| 2384 | while(num_blocks) \ | |
| 2385 | { \ | |
| 2386 | vec_8x16u zero_mask; \ | |
| 2387 | \ | |
| 2388 | draw_mask_bits = block->draw_mask_bits; \ | |
| 2389 | dup_8x16b(draw_mask, draw_mask_bits); \ | |
| 2390 | tst_8x16b(draw_mask, draw_mask, test_mask); \ | |
| 2391 | \ | |
| 2392 | pixels = block->texels; \ | |
| 2393 | \ | |
| 2394 | cmpeqz_8x16b(zero_mask, pixels); \ | |
| 2395 | or_8x16b(zero_mask, draw_mask, zero_mask); \ | |
| 2396 | \ | |
| 2397 | shade_blocks_store_##target(zero_mask, pixels); \ | |
| 2398 | \ | |
| 2399 | num_blocks--; \ | |
| 2400 | block++; \ | |
| 2401 | } \ | |
| 2402 | } \ | |
| 75e28f62 | 2403 | |
| 2bbbb7af | 2404 | #ifndef NEON_BUILD |
| 75e28f62 E |
2405 | |
| 2406 | shade_blocks_textured_unmodulated_builder(indirect) | |
| 2407 | shade_blocks_textured_unmodulated_builder(direct) | |
| 2408 | ||
| 75e28f62 E |
2409 | void shade_blocks_unshaded_untextured_indirect(psx_gpu_struct *psx_gpu) |
| 2410 | { | |
| 2411 | } | |
| 2412 | ||
| 2413 | void shade_blocks_unshaded_untextured_direct(psx_gpu_struct *psx_gpu) | |
| 2414 | { | |
| 2415 | block_struct *block = psx_gpu->blocks; | |
| 2416 | u32 num_blocks = psx_gpu->num_blocks; | |
| 2417 | ||
| 2418 | vec_8x16u pixels = block->pixels; | |
| 2419 | shade_blocks_load_msb_mask_direct(); | |
| 2420 | ||
| 2421 | while(num_blocks) | |
| 2422 | { | |
| 2423 | shade_blocks_store_direct(block->draw_mask, pixels); | |
| 2424 | ||
| 2425 | num_blocks--; | |
| 2426 | block++; | |
| 2427 | } | |
| 2428 | } | |
| 2429 | ||
| 2430 | #endif | |
| 2431 | ||
| 2432 | void shade_blocks_shaded_untextured(psx_gpu_struct *psx_gpu) | |
| 2433 | { | |
| 2434 | } | |
| 2435 | ||
| 2436 | ||
| 2437 | #define blend_blocks_mask_evaluate_on() \ | |
| 2438 | vec_8x16u mask_pixels; \ | |
| 2439 | cmpltz_8x16b(mask_pixels, framebuffer_pixels); \ | |
| 2440 | or_8x16b(draw_mask, draw_mask, mask_pixels) \ | |
| 2441 | ||
| 2442 | #define blend_blocks_mask_evaluate_off() \ | |
| 2443 | ||
| 2444 | #define blend_blocks_average() \ | |
| 2445 | { \ | |
| 2446 | vec_8x16u pixels_no_msb; \ | |
| 2447 | vec_8x16u fb_pixels_no_msb; \ | |
| 2448 | \ | |
| 2449 | vec_8x16u d128_0x0421; \ | |
| 2450 | vec_8x16u d128_0x8000; \ | |
| 2451 | \ | |
| 2452 | dup_8x16b(d128_0x0421, 0x0421); \ | |
| 2453 | dup_8x16b(d128_0x8000, 0x8000); \ | |
| 2454 | \ | |
| 2455 | eor_8x16b(blend_pixels, pixels, framebuffer_pixels); \ | |
| 2456 | bic_8x16b(pixels_no_msb, pixels, d128_0x8000); \ | |
| 2457 | and_8x16b(blend_pixels, blend_pixels, d128_0x0421); \ | |
| 2458 | sub_8x16b(blend_pixels, pixels_no_msb, blend_pixels); \ | |
| 2459 | bic_8x16b(fb_pixels_no_msb, framebuffer_pixels, d128_0x8000); \ | |
| 2460 | average_8x16b(blend_pixels, fb_pixels_no_msb, blend_pixels); \ | |
| 2461 | } \ | |
| 2462 | ||
| 2463 | #define blend_blocks_add() \ | |
| 2464 | { \ | |
| 2465 | vec_8x16u pixels_rb, pixels_g; \ | |
| 2466 | vec_8x16u fb_rb, fb_g; \ | |
| 2467 | \ | |
| 2468 | vec_8x16u d128_0x7C1F; \ | |
| 2469 | vec_8x16u d128_0x03E0; \ | |
| 2470 | \ | |
| 2471 | dup_8x16b(d128_0x7C1F, 0x7C1F); \ | |
| 2472 | dup_8x16b(d128_0x03E0, 0x03E0); \ | |
| 2473 | \ | |
| 2474 | and_8x16b(pixels_rb, pixels, d128_0x7C1F); \ | |
| 2475 | and_8x16b(pixels_g, pixels, d128_0x03E0); \ | |
| 2476 | \ | |
| 2477 | and_8x16b(fb_rb, framebuffer_pixels, d128_0x7C1F); \ | |
| 2478 | and_8x16b(fb_g, framebuffer_pixels, d128_0x03E0); \ | |
| 2479 | \ | |
| 2480 | add_8x16b(fb_rb, fb_rb, pixels_rb); \ | |
| 2481 | add_8x16b(fb_g, fb_g, pixels_g); \ | |
| 2482 | \ | |
| 2483 | min_16x8b(vector_cast(vec_16x8u, fb_rb), vector_cast(vec_16x8u, fb_rb), \ | |
| 2484 | vector_cast(vec_16x8u, d128_0x7C1F)); \ | |
| 2485 | min_8x16b(fb_g, fb_g, d128_0x03E0); \ | |
| 2486 | \ | |
| 2487 | or_8x16b(blend_pixels, fb_rb, fb_g); \ | |
| 2488 | } \ | |
| 2489 | ||
| 2490 | #define blend_blocks_subtract() \ | |
| 2491 | { \ | |
| 2492 | vec_8x16u pixels_rb, pixels_g; \ | |
| 2493 | vec_8x16u fb_rb, fb_g; \ | |
| 2494 | \ | |
| 2495 | vec_8x16u d128_0x7C1F; \ | |
| 2496 | vec_8x16u d128_0x03E0; \ | |
| 2497 | \ | |
| 2498 | dup_8x16b(d128_0x7C1F, 0x7C1F); \ | |
| 2499 | dup_8x16b(d128_0x03E0, 0x03E0); \ | |
| 2500 | \ | |
| 2501 | and_8x16b(pixels_rb, pixels, d128_0x7C1F); \ | |
| 2502 | and_8x16b(pixels_g, pixels, d128_0x03E0); \ | |
| 2503 | \ | |
| 2504 | and_8x16b(fb_rb, framebuffer_pixels, d128_0x7C1F); \ | |
| 2505 | and_8x16b(fb_g, framebuffer_pixels, d128_0x03E0); \ | |
| 2506 | \ | |
| 2507 | subs_16x8b(vector_cast(vec_16x8u, fb_rb), \ | |
| 2508 | vector_cast(vec_16x8u, fb_rb), vector_cast(vec_16x8u, pixels_rb)); \ | |
| 2509 | subs_8x16b(fb_g, fb_g, pixels_g); \ | |
| 2510 | \ | |
| 2511 | or_8x16b(blend_pixels, fb_rb, fb_g); \ | |
| 2512 | } \ | |
| 2513 | ||
| 2514 | #define blend_blocks_add_fourth() \ | |
| 2515 | { \ | |
| 2516 | vec_8x16u pixels_rb, pixels_g; \ | |
| 2517 | vec_8x16u pixels_fourth; \ | |
| 2518 | vec_8x16u fb_rb, fb_g; \ | |
| 2519 | \ | |
| 2520 | vec_8x16u d128_0x7C1F; \ | |
| 2521 | vec_8x16u d128_0x1C07; \ | |
| 2522 | vec_8x16u d128_0x03E0; \ | |
| 2523 | vec_8x16u d128_0x00E0; \ | |
| 2524 | \ | |
| 2525 | dup_8x16b(d128_0x7C1F, 0x7C1F); \ | |
| 2526 | dup_8x16b(d128_0x1C07, 0x1C07); \ | |
| 2527 | dup_8x16b(d128_0x03E0, 0x03E0); \ | |
| 2528 | dup_8x16b(d128_0x00E0, 0x00E0); \ | |
| 2529 | \ | |
| 2530 | shr_8x16b(pixels_fourth, vector_cast(vec_8x16s, pixels), 2); \ | |
| 2531 | \ | |
| 2532 | and_8x16b(fb_rb, framebuffer_pixels, d128_0x7C1F); \ | |
| 2533 | and_8x16b(fb_g, framebuffer_pixels, d128_0x03E0); \ | |
| 2534 | \ | |
| 2535 | and_8x16b(pixels_rb, pixels_fourth, d128_0x1C07); \ | |
| 2536 | and_8x16b(pixels_g, pixels_fourth, d128_0x00E0); \ | |
| 2537 | \ | |
| 2538 | add_8x16b(fb_rb, fb_rb, pixels_rb); \ | |
| 2539 | add_8x16b(fb_g, fb_g, pixels_g); \ | |
| 2540 | \ | |
| 2541 | min_16x8b(vector_cast(vec_16x8u, fb_rb), vector_cast(vec_16x8u, fb_rb), \ | |
| 2542 | vector_cast(vec_16x8u, d128_0x7C1F)); \ | |
| 2543 | min_8x16b(fb_g, fb_g, d128_0x03E0); \ | |
| 2544 | \ | |
| 2545 | or_8x16b(blend_pixels, fb_rb, fb_g); \ | |
| 2546 | } \ | |
| 2547 | ||
| 2548 | #define blend_blocks_blended_combine_textured() \ | |
| 2549 | { \ | |
| 2550 | vec_8x16u blend_mask; \ | |
| 2551 | cmpltz_8x16b(blend_mask, pixels); \ | |
| 2552 | \ | |
| 2553 | or_immediate_8x16b(blend_pixels, blend_pixels, 0x8000); \ | |
| 2554 | bif_8x16b(blend_pixels, pixels, blend_mask); \ | |
| 2555 | } \ | |
| 2556 | ||
| 2557 | #define blend_blocks_blended_combine_untextured() \ | |
| 2558 | ||
| 2559 | ||
| 2560 | #define blend_blocks_body_blend(blend_mode, texturing) \ | |
| 2561 | { \ | |
| 2562 | blend_blocks_##blend_mode(); \ | |
| 2563 | blend_blocks_blended_combine_##texturing(); \ | |
| 2564 | } \ | |
| 2565 | ||
| 2566 | #define blend_blocks_body_average(texturing) \ | |
| 2567 | blend_blocks_body_blend(average, texturing) \ | |
| 2568 | ||
| 2569 | #define blend_blocks_body_add(texturing) \ | |
| 2570 | blend_blocks_body_blend(add, texturing) \ | |
| 2571 | ||
| 2572 | #define blend_blocks_body_subtract(texturing) \ | |
| 2573 | blend_blocks_body_blend(subtract, texturing) \ | |
| 2574 | ||
| 2575 | #define blend_blocks_body_add_fourth(texturing) \ | |
| 2576 | blend_blocks_body_blend(add_fourth, texturing) \ | |
| 2577 | ||
| 2578 | #define blend_blocks_body_unblended(texturing) \ | |
| 2579 | blend_pixels = pixels \ | |
| 2580 | ||
| 2581 | ||
| 2582 | #define blend_blocks_builder(texturing, blend_mode, mask_evaluate) \ | |
| 2583 | void \ | |
| 2584 | blend_blocks_##texturing##_##blend_mode##_##mask_evaluate(psx_gpu_struct \ | |
| 2585 | *psx_gpu) \ | |
| 2586 | { \ | |
| 2587 | block_struct *block = psx_gpu->blocks; \ | |
| 2588 | u32 num_blocks = psx_gpu->num_blocks; \ | |
| 2589 | vec_8x16u draw_mask; \ | |
| 2590 | vec_8x16u pixels; \ | |
| 2591 | vec_8x16u blend_pixels; \ | |
| 2592 | vec_8x16u framebuffer_pixels; \ | |
| 2593 | vec_8x16u msb_mask; \ | |
| 2594 | \ | |
| 2595 | u16 *fb_ptr; \ | |
| 2596 | \ | |
| 2597 | dup_8x16b(msb_mask, psx_gpu->mask_msb); \ | |
| 2598 | \ | |
| 2599 | while(num_blocks) \ | |
| 2600 | { \ | |
| 2601 | pixels = block->pixels; \ | |
| 2602 | draw_mask = block->draw_mask; \ | |
| 2603 | fb_ptr = block->fb_ptr; \ | |
| 2604 | \ | |
| 2605 | load_8x16b(framebuffer_pixels, fb_ptr); \ | |
| 2606 | \ | |
| 2607 | blend_blocks_mask_evaluate_##mask_evaluate(); \ | |
| 2608 | blend_blocks_body_##blend_mode(texturing); \ | |
| 2609 | \ | |
| 2610 | or_8x16b(blend_pixels, blend_pixels, msb_mask); \ | |
| 2611 | bif_8x16b(framebuffer_pixels, blend_pixels, draw_mask); \ | |
| 2612 | store_8x16b(framebuffer_pixels, fb_ptr); \ | |
| 2613 | \ | |
| 2614 | blend_blocks++; \ | |
| 2615 | num_blocks--; \ | |
| 2616 | block++; \ | |
| 2617 | } \ | |
| 2618 | } \ | |
| 2619 | ||
| 2bbbb7af | 2620 | #ifndef NEON_BUILD |
| 75e28f62 E |
2621 | |
| 2622 | void blend_blocks_textured_unblended_off(psx_gpu_struct *psx_gpu) | |
| 2623 | { | |
| 2624 | } | |
| 2625 | ||
| 2626 | blend_blocks_builder(textured, average, off); | |
| 2627 | blend_blocks_builder(textured, average, on); | |
| 2628 | blend_blocks_builder(textured, add, off); | |
| 2629 | blend_blocks_builder(textured, add, on); | |
| 2630 | blend_blocks_builder(textured, subtract, off); | |
| 2631 | blend_blocks_builder(textured, subtract, on); | |
| 2632 | blend_blocks_builder(textured, add_fourth, off); | |
| 2633 | blend_blocks_builder(textured, add_fourth, on); | |
| 2634 | ||
| 2635 | blend_blocks_builder(untextured, average, off); | |
| 2636 | blend_blocks_builder(untextured, average, on); | |
| 2637 | blend_blocks_builder(untextured, add, off); | |
| 2638 | blend_blocks_builder(untextured, add, on); | |
| 2639 | blend_blocks_builder(untextured, subtract, off); | |
| 2640 | blend_blocks_builder(untextured, subtract, on); | |
| 2641 | blend_blocks_builder(untextured, add_fourth, off); | |
| 2642 | blend_blocks_builder(untextured, add_fourth, on); | |
| 2643 | ||
| 2644 | blend_blocks_builder(textured, unblended, on); | |
| 2645 | ||
| 2646 | #endif | |
| 2647 | ||
| 2648 | ||
| 2649 | #define vertex_swap(_a, _b) \ | |
| 2650 | { \ | |
| 2651 | vertex_struct *temp_vertex = _a; \ | |
| 2652 | _a = _b; \ | |
| 2653 | _b = temp_vertex; \ | |
| 2654 | triangle_winding ^= 1; \ | |
| 2655 | } \ | |
| 2656 | ||
| 2657 | ||
| 2658 | // Setup blocks parametric-variables: | |
| 2659 | // SHADE TEXTURE_MAP SWIZZLING | |
| 2660 | // 0 0 x | |
| 2661 | // 0 1 0 | |
| 2662 | // 0 1 1 | |
| 2663 | // 1 0 x | |
| 2664 | // 1 1 0 | |
| 2665 | // 1 1 1 | |
| 2666 | // 8 inputs, 6 combinations | |
| 2667 | ||
| 2668 | #define setup_blocks_switch_untextured_unshaded(dithering, target) \ | |
| 2669 | setup_blocks_unshaded_untextured_undithered_unswizzled_##target \ | |
| 2670 | ||
| 2671 | #define setup_blocks_switch_untextured_shaded(dithering, target) \ | |
| 2672 | setup_blocks_shaded_untextured_##dithering##_unswizzled_##target \ | |
| 2673 | ||
| 2674 | #define setup_blocks_switch_untextured(shading, texture_mode, dithering, \ | |
| 2675 | target) \ | |
| 2676 | setup_blocks_switch_untextured_##shading(dithering, target) \ | |
| 2677 | ||
| 2678 | #define setup_blocks_switch_texture_mode_4bpp(shading) \ | |
| 2679 | setup_blocks_##shading##_textured_dithered_swizzled_indirect \ | |
| 2680 | ||
| 2681 | #define setup_blocks_switch_texture_mode_8bpp(shading) \ | |
| 2682 | setup_blocks_##shading##_textured_dithered_swizzled_indirect \ | |
| 2683 | ||
| 2684 | #define setup_blocks_switch_texture_mode_16bpp(shading) \ | |
| 2685 | setup_blocks_##shading##_textured_dithered_unswizzled_indirect \ | |
| 2686 | ||
| 2687 | #define setup_blocks_switch_textured(shading, texture_mode, dithering, target) \ | |
| 2688 | setup_blocks_switch_texture_mode_##texture_mode(shading) \ | |
| 2689 | ||
| 2690 | #define setup_blocks_switch_blended(shading, texturing, texture_mode, \ | |
| 2691 | dithering, mask_evaluate) \ | |
| 2692 | setup_blocks_switch_##texturing(shading, texture_mode, dithering, indirect) \ | |
| 2693 | ||
| 2694 | #define setup_blocks_switch_unblended_on(shading, texturing, texture_mode, \ | |
| 2695 | dithering) \ | |
| 2696 | setup_blocks_switch_##texturing(shading, texture_mode, dithering, indirect) \ | |
| 2697 | ||
| 2698 | #define setup_blocks_switch_unblended_off(shading, texturing, texture_mode, \ | |
| 2699 | dithering) \ | |
| 2700 | setup_blocks_switch_##texturing(shading, texture_mode, dithering, direct) \ | |
| 2701 | ||
| 2702 | #define setup_blocks_switch_unblended(shading, texturing, texture_mode, \ | |
| 2703 | dithering, mask_evaluate) \ | |
| 2704 | setup_blocks_switch_unblended_##mask_evaluate(shading, texturing, \ | |
| 2705 | texture_mode, dithering) \ | |
| 2706 | ||
| 2707 | #define setup_blocks_switch(shading, texturing, texture_mode, dithering, \ | |
| 2708 | blending, mask_evaluate) \ | |
| 2709 | setup_blocks_switch_##blending(shading, texturing, texture_mode, \ | |
| 2710 | dithering, mask_evaluate) \ | |
| 2711 | ||
| 2712 | ||
| 2713 | // Texture blocks: | |
| 2714 | ||
| 2715 | #define texture_blocks_switch_untextured(texture_mode) \ | |
| 2716 | texture_blocks_untextured \ | |
| 2717 | ||
| 2718 | #define texture_blocks_switch_textured(texture_mode) \ | |
| 2719 | texture_blocks_##texture_mode \ | |
| 2720 | ||
| 2721 | #define texture_blocks_switch(texturing, texture_mode) \ | |
| 2722 | texture_blocks_switch_##texturing(texture_mode) \ | |
| 2723 | ||
| 2724 | ||
| 2725 | // Shade blocks parametric-variables: | |
| 2726 | // SHADE TEXTURE_MAP MODULATE_TEXELS dither_mode | |
| 2727 | // 0 0 x x | |
| 2728 | // 0 1 0 0 | |
| 2729 | // 0 1 0 1 | |
| 2730 | // x 1 1 x | |
| 2731 | // 1 0 x 0 | |
| 2732 | // 1 0 x 1 | |
| 2733 | // 1 1 0 0 | |
| 2734 | // 1 1 0 1 | |
| 2735 | // 16 inputs, 8 combinations | |
| 2736 | ||
| 2737 | #define shade_blocks_switch_unshaded_untextured(modulation, dithering, target) \ | |
| 2738 | shade_blocks_unshaded_untextured_##target \ | |
| 2739 | ||
| 2740 | #define shade_blocks_switch_unshaded_textured_unmodulated(dithering, target) \ | |
| 2741 | shade_blocks_textured_unmodulated_##target \ | |
| 2742 | ||
| 2743 | #define shade_blocks_switch_unshaded_textured_modulated(dithering, target) \ | |
| 2744 | shade_blocks_unshaded_textured_modulated_##dithering##_##target \ | |
| 2745 | ||
| 2746 | #define shade_blocks_switch_unshaded_textured(modulation, dithering, target) \ | |
| 2747 | shade_blocks_switch_unshaded_textured_##modulation(dithering, target) \ | |
| 2748 | ||
| 2749 | #define shade_blocks_switch_unshaded(texturing, modulation, dithering, target) \ | |
| 2750 | shade_blocks_switch_unshaded_##texturing(modulation, dithering, target) \ | |
| 2751 | ||
| 2752 | #define shade_blocks_switch_shaded_untextured(modulation, dithering, target) \ | |
| 2753 | shade_blocks_shaded_untextured \ | |
| 2754 | ||
| 2755 | #define shade_blocks_switch_shaded_textured_unmodulated(dithering, target) \ | |
| 2756 | shade_blocks_textured_unmodulated_##target \ | |
| 2757 | ||
| 2758 | #define shade_blocks_switch_shaded_textured_modulated(dithering, target) \ | |
| 2759 | shade_blocks_shaded_textured_modulated_##dithering##_##target \ | |
| 2760 | ||
| 2761 | #define shade_blocks_switch_shaded_textured(modulation, dithering, target) \ | |
| 2762 | shade_blocks_switch_shaded_textured_##modulation(dithering, target) \ | |
| 2763 | ||
| 2764 | #define shade_blocks_switch_shaded(texturing, modulation, dithering, target) \ | |
| 2765 | shade_blocks_switch_shaded_##texturing(modulation, dithering, target) \ | |
| 2766 | ||
| 2767 | #define shade_blocks_switch_mask_off(shading, texturing, modulation, \ | |
| 2768 | dithering) \ | |
| 2769 | shade_blocks_switch_##shading(texturing, modulation, dithering, direct) \ | |
| 2770 | ||
| 2771 | #define shade_blocks_switch_mask_on(shading, texturing, modulation, \ | |
| 2772 | dithering) \ | |
| 2773 | shade_blocks_switch_##shading(texturing, modulation, dithering, indirect) \ | |
| 2774 | ||
| 2775 | #define shade_blocks_switch_blended(shading, texturing, modulation, dithering, \ | |
| 2776 | mask_evaluate) \ | |
| 2777 | shade_blocks_switch_##shading(texturing, modulation, dithering, indirect) \ | |
| 2778 | ||
| 2779 | #define shade_blocks_switch_unblended(shading, texturing, modulation, \ | |
| 2780 | dithering, mask_evaluate) \ | |
| 2781 | shade_blocks_switch_mask_##mask_evaluate(shading, texturing, modulation, \ | |
| 2782 | dithering) \ | |
| 2783 | ||
| 2784 | #define shade_blocks_switch(shading, texturing, modulation, dithering, \ | |
| 2785 | blending, mask_evaluate) \ | |
| 2786 | shade_blocks_switch_##blending(shading, texturing, modulation, dithering, \ | |
| 2787 | mask_evaluate) \ | |
| 2788 | ||
| 2789 | ||
| 2790 | // Blend blocks parametric-variables: | |
| 2791 | // TEXTURE_MAP BLEND BM_A BM_B mask_evaluate | |
| 2792 | // x 0 x x 0 | |
| 2793 | // x 0 x x 1 | |
| 2794 | // 0 1 0 0 0 | |
| 2795 | // 0 1 0 0 1 | |
| 2796 | // 0 1 0 1 0 | |
| 2797 | // 0 1 0 1 1 | |
| 2798 | // 0 1 1 0 0 | |
| 2799 | // 0 1 1 0 1 | |
| 2800 | // 0 1 1 1 0 | |
| 2801 | // 0 1 1 1 1 | |
| 2802 | // 1 1 0 0 0 | |
| 2803 | // 1 1 0 0 1 | |
| 2804 | // 1 1 0 1 0 | |
| 2805 | // 1 1 0 1 1 | |
| 2806 | // 1 1 1 0 0 | |
| 2807 | // 1 1 1 0 1 | |
| 2808 | // 1 1 1 1 0 | |
| 2809 | // 1 1 1 1 1 | |
| 2810 | // 32 inputs, 18 combinations | |
| 2811 | ||
| 2812 | #define blend_blocks_switch_unblended(texturing, blend_mode, mask_evaluate) \ | |
| 2813 | blend_blocks_textured_unblended_##mask_evaluate \ | |
| 2814 | ||
| 2815 | #define blend_blocks_switch_blended(texturing, blend_mode, mask_evaluate) \ | |
| 2816 | blend_blocks_##texturing##_##blend_mode##_##mask_evaluate \ | |
| 2817 | ||
| 2818 | #define blend_blocks_switch(texturing, blending, blend_mode, mask_evaluate) \ | |
| 2819 | blend_blocks_switch_##blending(texturing, blend_mode, mask_evaluate) \ | |
| 2820 | ||
| 2821 | ||
| 2822 | #define render_blocks_switch_block_modulation(texture_mode, blend_mode, \ | |
| 2823 | mask_evaluate, shading, dithering, texturing, blending, modulation) \ | |
| 2824 | { \ | |
| 2825 | setup_blocks_switch(shading, texturing, texture_mode, dithering, blending, \ | |
| 2826 | mask_evaluate), \ | |
| 2827 | texture_blocks_switch(texturing, texture_mode), \ | |
| 2828 | shade_blocks_switch(shading, texturing, modulation, dithering, blending, \ | |
| 2829 | mask_evaluate), \ | |
| 2830 | blend_blocks_switch(texturing, blending, blend_mode, mask_evaluate) \ | |
| 2831 | } \ | |
| 2832 | ||
| 2833 | #define render_blocks_switch_block_blending(texture_mode, blend_mode, \ | |
| 2834 | mask_evaluate, shading, dithering, texturing, blending) \ | |
| 2835 | render_blocks_switch_block_modulation(texture_mode, blend_mode, \ | |
| 2836 | mask_evaluate, shading, dithering, texturing, blending, modulated), \ | |
| 2837 | render_blocks_switch_block_modulation(texture_mode, blend_mode, \ | |
| 2838 | mask_evaluate, shading, dithering, texturing, blending, unmodulated) \ | |
| 2839 | ||
| 2840 | #define render_blocks_switch_block_texturing(texture_mode, blend_mode, \ | |
| 2841 | mask_evaluate, shading, dithering, texturing) \ | |
| 2842 | render_blocks_switch_block_blending(texture_mode, blend_mode, \ | |
| 2843 | mask_evaluate, shading, dithering, texturing, unblended), \ | |
| 2844 | render_blocks_switch_block_blending(texture_mode, blend_mode, \ | |
| 2845 | mask_evaluate, shading, dithering, texturing, blended) \ | |
| 2846 | ||
| 2847 | #define render_blocks_switch_block_dithering(texture_mode, blend_mode, \ | |
| 2848 | mask_evaluate, shading, dithering) \ | |
| 2849 | render_blocks_switch_block_texturing(texture_mode, blend_mode, \ | |
| 2850 | mask_evaluate, shading, dithering, untextured), \ | |
| 2851 | render_blocks_switch_block_texturing(texture_mode, blend_mode, \ | |
| 2852 | mask_evaluate, shading, dithering, textured) \ | |
| 2853 | ||
| 2854 | #define render_blocks_switch_block_shading(texture_mode, blend_mode, \ | |
| 2855 | mask_evaluate, shading) \ | |
| 2856 | render_blocks_switch_block_dithering(texture_mode, blend_mode, \ | |
| 2857 | mask_evaluate, shading, undithered), \ | |
| 2858 | render_blocks_switch_block_dithering(texture_mode, blend_mode, \ | |
| 2859 | mask_evaluate, shading, dithered) \ | |
| 2860 | ||
| 2861 | #define render_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, \ | |
| 2862 | mask_evaluate) \ | |
| 2863 | render_blocks_switch_block_shading(texture_mode, blend_mode, mask_evaluate, \ | |
| 2864 | unshaded), \ | |
| 2865 | render_blocks_switch_block_shading(texture_mode, blend_mode, mask_evaluate, \ | |
| 2866 | shaded) \ | |
| 2867 | ||
| 2868 | #define render_blocks_switch_block_blend_mode(texture_mode, blend_mode) \ | |
| 2869 | render_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, off), \ | |
| 2870 | render_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, on) \ | |
| 2871 | ||
| 2872 | #define render_blocks_switch_block_texture_mode(texture_mode) \ | |
| 2873 | render_blocks_switch_block_blend_mode(texture_mode, average), \ | |
| 2874 | render_blocks_switch_block_blend_mode(texture_mode, add), \ | |
| 2875 | render_blocks_switch_block_blend_mode(texture_mode, subtract), \ | |
| 2876 | render_blocks_switch_block_blend_mode(texture_mode, add_fourth) \ | |
| 2877 | ||
| 2878 | #define render_blocks_switch_block() \ | |
| 2879 | render_blocks_switch_block_texture_mode(4bpp), \ | |
| 2880 | render_blocks_switch_block_texture_mode(8bpp), \ | |
| 2881 | render_blocks_switch_block_texture_mode(16bpp), \ | |
| b7f5c059 | 2882 | render_blocks_switch_block_texture_mode(16bpp) \ |
| 75e28f62 E |
2883 | |
| 2884 | ||
| 2885 | render_block_handler_struct render_triangle_block_handlers[] = | |
| 2886 | { | |
| 2887 | render_blocks_switch_block() | |
| 2888 | }; | |
| 2889 | ||
| 2890 | #undef render_blocks_switch_block_modulation | |
| 2891 | ||
| 2892 | #define render_blocks_switch_block_modulation(texture_mode, blend_mode, \ | |
| 2893 | mask_evaluate, shading, dithering, texturing, blending, modulation) \ | |
| 2894 | "render flags:\n" \ | |
| 2895 | "texture mode: " #texture_mode "\n" \ | |
| 2896 | "blend mode: " #blend_mode "\n" \ | |
| 2897 | "mask evaluation: " #mask_evaluate "\n" \ | |
| 2898 | #shading "\n" \ | |
| 2899 | #dithering "\n" \ | |
| 2900 | #texturing "\n" \ | |
| 2901 | #blending "\n" \ | |
| 2902 | #modulation "\n" \ | |
| 2903 | ||
| 2904 | char *render_block_flag_strings[] = | |
| 2905 | { | |
| 2906 | render_blocks_switch_block() | |
| 2907 | }; | |
| 2908 | ||
| 2909 | ||
| 2910 | #define triangle_y_direction_up 1 | |
| 2911 | #define triangle_y_direction_flat 2 | |
| 2912 | #define triangle_y_direction_down 0 | |
| 2913 | ||
| 2914 | #define triangle_winding_positive 0 | |
| 2915 | #define triangle_winding_negative 1 | |
| 2916 | ||
| 2917 | #define triangle_set_direction(direction_variable, value) \ | |
| 2918 | u32 direction_variable = (u32)(value) >> 31; \ | |
| 2919 | if(value == 0) \ | |
| 2920 | direction_variable = 2 \ | |
| 2921 | ||
| 2922 | #define triangle_case(direction_a, direction_b, direction_c, winding) \ | |
| 2923 | case (triangle_y_direction_##direction_a | \ | |
| 2924 | (triangle_y_direction_##direction_b << 2) | \ | |
| 2925 | (triangle_y_direction_##direction_c << 4) | \ | |
| 2926 | (triangle_winding_##winding << 6)) \ | |
| 2927 | ||
| c1817bd9 | 2928 | static int prepare_triangle(psx_gpu_struct *psx_gpu, vertex_struct *vertexes, |
| f4ce97e1 | 2929 | prepared_triangle *triangle_out) |
| 75e28f62 | 2930 | { |
| f4ce97e1 | 2931 | s32 y_top, y_bottom, offset_x, offset_y, i; |
| 75e28f62 E |
2932 | s32 triangle_area; |
| 2933 | u32 triangle_winding = 0; | |
| 2934 | ||
| 2935 | vertex_struct *a = &(vertexes[0]); | |
| 2936 | vertex_struct *b = &(vertexes[1]); | |
| 2937 | vertex_struct *c = &(vertexes[2]); | |
| 2938 | ||
| 2939 | triangle_area = triangle_signed_area_x2(a->x, a->y, b->x, b->y, c->x, c->y); | |
| 2940 | ||
| 3867c6ef | 2941 | #ifdef PROFILE |
| 75e28f62 | 2942 | triangles++; |
| 3867c6ef | 2943 | #endif |
| 75e28f62 E |
2944 | |
| 2945 | if(triangle_area == 0) | |
| 2946 | { | |
| 3867c6ef | 2947 | #ifdef PROFILE |
| 75e28f62 | 2948 | trivial_rejects++; |
| 3867c6ef | 2949 | #endif |
| c1817bd9 | 2950 | return 0; |
| 75e28f62 E |
2951 | } |
| 2952 | ||
| 2953 | if(b->y < a->y) | |
| 2954 | vertex_swap(a, b); | |
| 2955 | ||
| 2956 | if(c->y < b->y) | |
| 2957 | { | |
| 2958 | vertex_swap(b, c); | |
| 2959 | ||
| 2960 | if(b->y < a->y) | |
| 2961 | vertex_swap(a, b); | |
| 2962 | } | |
| 2963 | ||
| 2964 | y_bottom = c->y; | |
| 2965 | y_top = a->y; | |
| f4ce97e1 | 2966 | offset_y = sign_extend_11bit(y_top + psx_gpu->offset_y) - y_top; |
| 75e28f62 E |
2967 | |
| 2968 | if((y_bottom - y_top) >= 512) | |
| 2969 | { | |
| 3867c6ef | 2970 | #ifdef PROFILE |
| 75e28f62 | 2971 | trivial_rejects++; |
| 3867c6ef | 2972 | #endif |
| c1817bd9 | 2973 | return 0; |
| 75e28f62 E |
2974 | } |
| 2975 | ||
| 2976 | if(triangle_area < 0) | |
| 2977 | { | |
| 2978 | triangle_area = -triangle_area; | |
| 2979 | triangle_winding ^= 1; | |
| 2980 | vertex_swap(a, c); | |
| 2981 | } | |
| 2982 | ||
| 2983 | if(b->x < a->x) | |
| 2984 | vertex_swap(a, b); | |
| 2985 | ||
| 2986 | if(c->x < b->x) | |
| 2987 | { | |
| 2988 | vertex_swap(b, c); | |
| 2989 | ||
| 2990 | if(b->x < a->x) | |
| 2991 | vertex_swap(a, b); | |
| 2992 | } | |
| 2993 | ||
| f4ce97e1 | 2994 | if(c->x - a->x >= 1024) |
| 75e28f62 | 2995 | { |
| 3867c6ef | 2996 | #ifdef PROFILE |
| 75e28f62 | 2997 | trivial_rejects++; |
| 3867c6ef | 2998 | #endif |
| c1817bd9 | 2999 | return 0; |
| 75e28f62 E |
3000 | } |
| 3001 | ||
| f4ce97e1 | 3002 | offset_x = sign_extend_11bit(a->x + psx_gpu->offset_x) - a->x; |
| 3003 | if(invalidate_texture_cache_region_viewport(psx_gpu, | |
| 3004 | a->x + offset_x, y_top + offset_y, | |
| 3005 | c->x + offset_x, y_bottom + offset_y) == 0) | |
| 75e28f62 | 3006 | { |
| 3867c6ef | 3007 | #ifdef PROFILE |
| 75e28f62 | 3008 | trivial_rejects++; |
| 3867c6ef | 3009 | #endif |
| c1817bd9 | 3010 | return 0; |
| 75e28f62 E |
3011 | } |
| 3012 | ||
| f4ce97e1 | 3013 | for (i = 0; i < 3; i++) |
| 3014 | { | |
| 3015 | vertexes[i].x += offset_x; | |
| 3016 | vertexes[i].y += offset_y; | |
| 3017 | } | |
| 3018 | ||
| 75e28f62 E |
3019 | psx_gpu->triangle_area = triangle_area; |
| 3020 | psx_gpu->triangle_winding = triangle_winding; | |
| 3021 | ||
| f4ce97e1 | 3022 | triangle_out->vertexes[0] = a; |
| 3023 | triangle_out->vertexes[1] = b; | |
| 3024 | triangle_out->vertexes[2] = c; | |
| 3025 | triangle_out->offset_x = offset_x; | |
| 3026 | triangle_out->offset_y = offset_y; | |
| c1817bd9 | 3027 | |
| 3028 | return 1; | |
| 3029 | } | |
| 3030 | ||
| 3031 | static void render_triangle_p(psx_gpu_struct *psx_gpu, | |
| 3032 | vertex_struct *vertex_ptrs[3], u32 flags) | |
| 3033 | { | |
| 3034 | psx_gpu->num_spans = 0; | |
| 3035 | ||
| 3036 | vertex_struct *a = vertex_ptrs[0]; | |
| 3037 | vertex_struct *b = vertex_ptrs[1]; | |
| 3038 | vertex_struct *c = vertex_ptrs[2]; | |
| 3039 | ||
| 75e28f62 E |
3040 | s32 y_delta_a = b->y - a->y; |
| 3041 | s32 y_delta_b = c->y - b->y; | |
| 3042 | s32 y_delta_c = c->y - a->y; | |
| 3043 | ||
| 3044 | triangle_set_direction(y_direction_a, y_delta_a); | |
| 3045 | triangle_set_direction(y_direction_b, y_delta_b); | |
| 3046 | triangle_set_direction(y_direction_c, y_delta_c); | |
| 3047 | ||
| db2804fb | 3048 | #if defined(__arm__) && defined(NEON_BUILD) && !defined(SIMD_BUILD) |
| 3049 | // the asm doesn't bother to save callee-save vector regs, so do it here | |
| 3050 | __asm__ __volatile__("vstmia %0, {q4-q7}" :: "r"(psx_gpu->saved_q4_q7) : "memory"); | |
| 3051 | #endif | |
| 3052 | ||
| 75e28f62 E |
3053 | compute_all_gradients(psx_gpu, a, b, c); |
| 3054 | ||
| 3055 | switch(y_direction_a | (y_direction_b << 2) | (y_direction_c << 4) | | |
| c1817bd9 | 3056 | (psx_gpu->triangle_winding << 6)) |
| 75e28f62 E |
3057 | { |
| 3058 | triangle_case(up, up, up, negative): | |
| 3059 | triangle_case(up, up, flat, negative): | |
| 3060 | triangle_case(up, up, down, negative): | |
| 3061 | setup_spans_up_right(psx_gpu, a, b, c); | |
| 3062 | break; | |
| 3063 | ||
| 3064 | triangle_case(flat, up, up, negative): | |
| 3065 | triangle_case(flat, up, flat, negative): | |
| 3066 | triangle_case(flat, up, down, negative): | |
| 3067 | setup_spans_up_a(psx_gpu, a, b, c); | |
| 3068 | break; | |
| 3069 | ||
| 3070 | triangle_case(down, up, up, negative): | |
| 3071 | setup_spans_up_down(psx_gpu, a, c, b); | |
| 3072 | break; | |
| 3073 | ||
| 3074 | triangle_case(down, up, flat, negative): | |
| 3075 | setup_spans_down_a(psx_gpu, a, c, b); | |
| 3076 | break; | |
| 3077 | ||
| 3078 | triangle_case(down, up, down, negative): | |
| 3079 | setup_spans_down_right(psx_gpu, a, c, b); | |
| 3080 | break; | |
| 3081 | ||
| 3082 | triangle_case(down, flat, up, negative): | |
| 3083 | triangle_case(down, flat, flat, negative): | |
| 3084 | triangle_case(down, flat, down, negative): | |
| 3085 | setup_spans_down_b(psx_gpu, a, b, c); | |
| 3086 | break; | |
| 3087 | ||
| 3088 | triangle_case(down, down, up, negative): | |
| 3089 | triangle_case(down, down, flat, negative): | |
| 3090 | triangle_case(down, down, down, negative): | |
| 3091 | setup_spans_down_left(psx_gpu, a, b, c); | |
| 3092 | break; | |
| 3093 | ||
| 3094 | triangle_case(up, up, up, positive): | |
| 3095 | triangle_case(up, up, flat, positive): | |
| 3096 | triangle_case(up, up, down, positive): | |
| 3097 | setup_spans_up_left(psx_gpu, a, b, c); | |
| 3098 | break; | |
| 3099 | ||
| 3100 | triangle_case(up, flat, up, positive): | |
| 3101 | triangle_case(up, flat, flat, positive): | |
| 3102 | triangle_case(up, flat, down, positive): | |
| 3103 | setup_spans_up_b(psx_gpu, a, b, c); | |
| 3104 | break; | |
| 3105 | ||
| 3106 | triangle_case(up, down, up, positive): | |
| 3107 | setup_spans_up_right(psx_gpu, a, c, b); | |
| 3108 | break; | |
| 3109 | ||
| 3110 | triangle_case(up, down, flat, positive): | |
| 3111 | setup_spans_up_a(psx_gpu, a, c, b); | |
| 3112 | break; | |
| 3113 | ||
| 3114 | triangle_case(up, down, down, positive): | |
| 3115 | setup_spans_up_down(psx_gpu, a, b, c); | |
| 3116 | break; | |
| 3117 | ||
| 3118 | triangle_case(flat, down, up, positive): | |
| 3119 | triangle_case(flat, down, flat, positive): | |
| 3120 | triangle_case(flat, down, down, positive): | |
| 3121 | setup_spans_down_a(psx_gpu, a, b, c); | |
| 3122 | break; | |
| 3123 | ||
| 3124 | triangle_case(down, down, up, positive): | |
| 3125 | triangle_case(down, down, flat, positive): | |
| 3126 | triangle_case(down, down, down, positive): | |
| 3127 | setup_spans_down_right(psx_gpu, a, b, c); | |
| 3128 | break; | |
| 3129 | } | |
| 3130 | ||
| 3867c6ef | 3131 | #ifdef PROFILE |
| 75e28f62 | 3132 | spans += psx_gpu->num_spans; |
| 3867c6ef | 3133 | #endif |
| 75e28f62 | 3134 | |
| f1359c57 | 3135 | if(unlikely(psx_gpu->render_mode & RENDER_INTERLACE_ENABLED)) |
| 69b09c0d E |
3136 | { |
| 3137 | u32 i; | |
| 3138 | ||
| f1359c57 | 3139 | if(psx_gpu->render_mode & RENDER_INTERLACE_ODD) |
| 69b09c0d E |
3140 | { |
| 3141 | for(i = 0; i < psx_gpu->num_spans; i++) | |
| 3142 | { | |
| 3143 | if((psx_gpu->span_edge_data[i].y & 1) == 0) | |
| 3144 | psx_gpu->span_edge_data[i].num_blocks = 0; | |
| 3145 | } | |
| 3146 | } | |
| 3147 | else | |
| 3148 | { | |
| 3149 | for(i = 0; i < psx_gpu->num_spans; i++) | |
| 3150 | { | |
| 3151 | if(psx_gpu->span_edge_data[i].y & 1) | |
| 3152 | psx_gpu->span_edge_data[i].num_blocks = 0; | |
| 3153 | } | |
| 3154 | } | |
| 3155 | } | |
| 2d658c89 | 3156 | assert(psx_gpu->span_edge_data[0].y < 1024u); |
| 69b09c0d | 3157 | |
| 75e28f62 E |
3158 | u32 render_state = flags & |
| 3159 | (RENDER_FLAGS_MODULATE_TEXELS | RENDER_FLAGS_BLEND | | |
| 3160 | RENDER_FLAGS_TEXTURE_MAP | RENDER_FLAGS_SHADE); | |
| 3161 | render_state |= psx_gpu->render_state_base; | |
| 3162 | ||
| 3163 | if((psx_gpu->render_state != render_state) || | |
| 3164 | (psx_gpu->primitive_type != PRIMITIVE_TYPE_TRIANGLE)) | |
| 3165 | { | |
| 3166 | psx_gpu->render_state = render_state; | |
| 3167 | flush_render_block_buffer(psx_gpu); | |
| 3867c6ef | 3168 | #ifdef PROFILE |
| 75e28f62 | 3169 | state_changes++; |
| 3867c6ef | 3170 | #endif |
| 75e28f62 E |
3171 | } |
| 3172 | ||
| 3173 | psx_gpu->primitive_type = PRIMITIVE_TYPE_TRIANGLE; | |
| 3174 | ||
| 3175 | psx_gpu->render_block_handler = | |
| 3176 | &(render_triangle_block_handlers[render_state]); | |
| 3177 | ((setup_blocks_function_type *)psx_gpu->render_block_handler->setup_blocks) | |
| 3178 | (psx_gpu); | |
| db2804fb | 3179 | |
| 3180 | #if defined(__arm__) && defined(NEON_BUILD) && !defined(SIMD_BUILD) | |
| 3181 | __asm__ __volatile__("vldmia %0, {q4-q7}" :: "r"(psx_gpu->saved_q4_q7)); | |
| 3182 | #endif | |
| 75e28f62 E |
3183 | } |
| 3184 | ||
| c1817bd9 | 3185 | void render_triangle(psx_gpu_struct *psx_gpu, vertex_struct *vertexes, |
| 3186 | u32 flags) | |
| 3187 | { | |
| f4ce97e1 | 3188 | prepared_triangle triangle; |
| 3189 | if (prepare_triangle(psx_gpu, vertexes, &triangle)) | |
| 3190 | render_triangle_p(psx_gpu, triangle.vertexes, flags); | |
| c1817bd9 | 3191 | } |
| 3192 | ||
| a2cb152a | 3193 | #if !defined(NEON_BUILD) || defined(SIMD_BUILD) |
| 75e28f62 E |
3194 | |
| 3195 | void texture_sprite_blocks_8bpp(psx_gpu_struct *psx_gpu) | |
| 3196 | { | |
| 3197 | block_struct *block = psx_gpu->blocks; | |
| 3198 | u32 num_blocks = psx_gpu->num_blocks; | |
| 3199 | ||
| 3200 | vec_8x16u texels; | |
| 3201 | vec_8x8u texel_indexes; | |
| 3202 | ||
| 3203 | u16 *clut_ptr = psx_gpu->clut_ptr; | |
| 3204 | u32 i; | |
| 3205 | ||
| 3206 | while(num_blocks) | |
| 3207 | { | |
| 3208 | texel_indexes = block->r; | |
| 3209 | ||
| 3210 | for(i = 0; i < 8; i++) | |
| 3211 | { | |
| 3212 | texels.e[i] = clut_ptr[texel_indexes.e[i]]; | |
| 3213 | } | |
| 3214 | ||
| 3215 | block->texels = texels; | |
| 3216 | ||
| 3217 | num_blocks--; | |
| 3218 | block++; | |
| 3219 | } | |
| 3220 | } | |
| 3221 | ||
| 3222 | #endif | |
| 3223 | ||
| 3224 | ||
| 59d15d23 | 3225 | #define setup_sprite_tiled_initialize_4bpp_clut() \ |
| 75e28f62 E |
3226 | u16 *clut_ptr = psx_gpu->clut_ptr; \ |
| 3227 | vec_8x16u clut_a, clut_b; \ | |
| 3228 | vec_16x8u clut_low, clut_high; \ | |
| 3229 | \ | |
| 3230 | load_8x16b(clut_a, clut_ptr); \ | |
| 3231 | load_8x16b(clut_b, clut_ptr + 8); \ | |
| 59d15d23 | 3232 | unzip_16x8b(clut_low, clut_high, clut_a, clut_b) \ |
| 3233 | ||
| 3234 | #define setup_sprite_tiled_initialize_4bpp() \ | |
| 3235 | setup_sprite_tiled_initialize_4bpp_clut(); \ | |
| 75e28f62 E |
3236 | \ |
| 3237 | if(psx_gpu->current_texture_mask & psx_gpu->dirty_textures_4bpp_mask) \ | |
| 3238 | update_texture_4bpp_cache(psx_gpu) \ | |
| 3239 | ||
| 3240 | #define setup_sprite_tiled_initialize_8bpp() \ | |
| 3241 | if(psx_gpu->current_texture_mask & psx_gpu->dirty_textures_8bpp_mask) \ | |
| 3242 | update_texture_8bpp_cache(psx_gpu) \ | |
| 3243 | ||
| 3244 | ||
| 3245 | #define setup_sprite_tile_fetch_texel_block_8bpp(offset) \ | |
| df740cdc | 3246 | texture_block_ptr = (u8 *)psx_gpu->texture_page_ptr + \ |
| 75e28f62 E |
3247 | ((texture_offset + offset) & texture_mask); \ |
| 3248 | \ | |
| 3249 | load_64b(texels, texture_block_ptr) \ | |
| 3250 | ||
| 3251 | ||
| 75e28f62 E |
3252 | #define setup_sprite_tile_add_blocks(tile_num_blocks) \ |
| 3253 | num_blocks += tile_num_blocks; \ | |
| 3254 | sprite_blocks += tile_num_blocks; \ | |
| 3255 | \ | |
| 3256 | if(num_blocks > MAX_BLOCKS) \ | |
| 3257 | { \ | |
| 3258 | flush_render_block_buffer(psx_gpu); \ | |
| 3259 | num_blocks = tile_num_blocks; \ | |
| 3260 | block = psx_gpu->blocks; \ | |
| 3261 | } \ | |
| 3262 | ||
| 3263 | #define setup_sprite_tile_full_4bpp(edge) \ | |
| 3264 | { \ | |
| 3265 | vec_8x8u texels_low, texels_high; \ | |
| 3266 | vec_8x16u pixels; \ | |
| 3267 | setup_sprite_tile_add_blocks(sub_tile_height * 2); \ | |
| 3268 | \ | |
| 3269 | while(sub_tile_height) \ | |
| 3270 | { \ | |
| 3271 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ | |
| 3272 | tbl_16(texels_low, texels, clut_low); \ | |
| 3273 | tbl_16(texels_high, texels, clut_high); \ | |
| 3274 | zip_8x16b(pixels, texels_low, texels_high); \ | |
| 3275 | \ | |
| 3276 | block->texels = pixels; \ | |
| 3277 | block->draw_mask_bits = left_mask_bits; \ | |
| 3278 | block->fb_ptr = fb_ptr; \ | |
| 3279 | block++; \ | |
| 3280 | \ | |
| 3281 | setup_sprite_tile_fetch_texel_block_8bpp(8); \ | |
| 3282 | tbl_16(texels_low, texels, clut_low); \ | |
| 3283 | tbl_16(texels_high, texels, clut_high); \ | |
| 3284 | zip_8x16b(pixels, texels_low, texels_high); \ | |
| 3285 | \ | |
| 3286 | block->texels = pixels; \ | |
| 3287 | block->draw_mask_bits = right_mask_bits; \ | |
| 3288 | block->fb_ptr = fb_ptr + 8; \ | |
| 3289 | block++; \ | |
| 3290 | \ | |
| 3291 | fb_ptr += 1024; \ | |
| 3292 | texture_offset += 0x10; \ | |
| 3293 | sub_tile_height--; \ | |
| 3294 | } \ | |
| 3295 | texture_offset += 0xF00; \ | |
| 3296 | psx_gpu->num_blocks = num_blocks; \ | |
| 3297 | } \ | |
| 3298 | ||
| 3299 | #define setup_sprite_tile_half_4bpp(edge) \ | |
| 3300 | { \ | |
| 3301 | vec_8x8u texels_low, texels_high; \ | |
| 3302 | vec_8x16u pixels; \ | |
| 3303 | setup_sprite_tile_add_blocks(sub_tile_height); \ | |
| 3304 | \ | |
| 3305 | while(sub_tile_height) \ | |
| 3306 | { \ | |
| 3307 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ | |
| 3308 | tbl_16(texels_low, texels, clut_low); \ | |
| 3309 | tbl_16(texels_high, texels, clut_high); \ | |
| 3310 | zip_8x16b(pixels, texels_low, texels_high); \ | |
| 3311 | \ | |
| 3312 | block->texels = pixels; \ | |
| 3313 | block->draw_mask_bits = edge##_mask_bits; \ | |
| 3314 | block->fb_ptr = fb_ptr; \ | |
| 3315 | block++; \ | |
| 3316 | \ | |
| 3317 | fb_ptr += 1024; \ | |
| 3318 | texture_offset += 0x10; \ | |
| 3319 | sub_tile_height--; \ | |
| 3320 | } \ | |
| 3321 | texture_offset += 0xF00; \ | |
| 3322 | psx_gpu->num_blocks = num_blocks; \ | |
| 3323 | } \ | |
| 3324 | ||
| 3325 | ||
| 3326 | #define setup_sprite_tile_full_8bpp(edge) \ | |
| 3327 | { \ | |
| 3328 | setup_sprite_tile_add_blocks(sub_tile_height * 2); \ | |
| 3329 | \ | |
| 3330 | while(sub_tile_height) \ | |
| 3331 | { \ | |
| 3332 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ | |
| 3333 | block->r = texels; \ | |
| 3334 | block->draw_mask_bits = left_mask_bits; \ | |
| 3335 | block->fb_ptr = fb_ptr; \ | |
| 3336 | block++; \ | |
| 3337 | \ | |
| 3338 | setup_sprite_tile_fetch_texel_block_8bpp(8); \ | |
| 3339 | block->r = texels; \ | |
| 3340 | block->draw_mask_bits = right_mask_bits; \ | |
| 3341 | block->fb_ptr = fb_ptr + 8; \ | |
| 3342 | block++; \ | |
| 3343 | \ | |
| 3344 | fb_ptr += 1024; \ | |
| 3345 | texture_offset += 0x10; \ | |
| 3346 | sub_tile_height--; \ | |
| 3347 | } \ | |
| 3348 | texture_offset += 0xF00; \ | |
| 3349 | psx_gpu->num_blocks = num_blocks; \ | |
| 3350 | } \ | |
| 3351 | ||
| 3352 | #define setup_sprite_tile_half_8bpp(edge) \ | |
| 3353 | { \ | |
| df740cdc | 3354 | setup_sprite_tile_add_blocks(sub_tile_height); \ |
| 75e28f62 E |
3355 | \ |
| 3356 | while(sub_tile_height) \ | |
| 3357 | { \ | |
| 3358 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ | |
| 3359 | block->r = texels; \ | |
| 3360 | block->draw_mask_bits = edge##_mask_bits; \ | |
| 3361 | block->fb_ptr = fb_ptr; \ | |
| 3362 | block++; \ | |
| 3363 | \ | |
| 3364 | fb_ptr += 1024; \ | |
| 3365 | texture_offset += 0x10; \ | |
| 3366 | sub_tile_height--; \ | |
| 3367 | } \ | |
| 3368 | texture_offset += 0xF00; \ | |
| 3369 | psx_gpu->num_blocks = num_blocks; \ | |
| 3370 | } \ | |
| 3371 | ||
| 3372 | ||
| 3373 | #define setup_sprite_tile_column_edge_pre_adjust_half_right() \ | |
| 3374 | texture_offset = texture_offset_base + 8; \ | |
| 3375 | fb_ptr += 8 \ | |
| 3376 | ||
| 3377 | #define setup_sprite_tile_column_edge_pre_adjust_half_left() \ | |
| 3378 | texture_offset = texture_offset_base \ | |
| 3379 | ||
| 3380 | #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \ | |
| 3381 | setup_sprite_tile_column_edge_pre_adjust_half_##edge() \ | |
| 3382 | ||
| 3383 | #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \ | |
| 3384 | texture_offset = texture_offset_base \ | |
| 3385 | ||
| 3386 | #define setup_sprite_tile_column_edge_post_adjust_half_right() \ | |
| 3387 | fb_ptr -= 8 \ | |
| 3388 | ||
| 3389 | #define setup_sprite_tile_column_edge_post_adjust_half_left() \ | |
| 3390 | ||
| 3391 | #define setup_sprite_tile_column_edge_post_adjust_half(edge) \ | |
| 3392 | setup_sprite_tile_column_edge_post_adjust_half_##edge() \ | |
| 3393 | ||
| 3394 | #define setup_sprite_tile_column_edge_post_adjust_full(edge) \ | |
| 3395 | ||
| 3396 | ||
| 59d15d23 | 3397 | #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \ |
| 3398 | x4mode) \ | |
| 75e28f62 E |
3399 | do \ |
| 3400 | { \ | |
| 3401 | sub_tile_height = column_data; \ | |
| 59d15d23 | 3402 | setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \ |
| 3403 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ | |
| 3404 | setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge); \ | |
| 75e28f62 E |
3405 | } while(0) \ |
| 3406 | ||
| 59d15d23 | 3407 | #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \ |
| 3408 | x4mode) \ | |
| 75e28f62 E |
3409 | do \ |
| 3410 | { \ | |
| 3411 | u32 tiles_remaining = column_data >> 16; \ | |
| 3412 | sub_tile_height = column_data & 0xFF; \ | |
| 59d15d23 | 3413 | setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \ |
| 3414 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ | |
| 75e28f62 E |
3415 | tiles_remaining -= 1; \ |
| 3416 | \ | |
| 3417 | while(tiles_remaining) \ | |
| 3418 | { \ | |
| 3419 | sub_tile_height = 16; \ | |
| 59d15d23 | 3420 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 75e28f62 E |
3421 | tiles_remaining--; \ |
| 3422 | } \ | |
| 3423 | \ | |
| 3424 | sub_tile_height = (column_data >> 8) & 0xFF; \ | |
| 59d15d23 | 3425 | setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \ |
| 3426 | setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge); \ | |
| 75e28f62 E |
3427 | } while(0) \ |
| 3428 | ||
| 3429 | ||
| 3430 | #define setup_sprite_column_data_single() \ | |
| 3431 | column_data = height \ | |
| 3432 | ||
| 3433 | #define setup_sprite_column_data_multi() \ | |
| 3434 | column_data = 16 - offset_v; \ | |
| 3435 | column_data |= ((height_rounded & 0xF) + 1) << 8; \ | |
| 3436 | column_data |= (tile_height - 1) << 16 \ | |
| 3437 | ||
| 3438 | ||
| 59d15d23 | 3439 | #define RIGHT_MASK_BIT_SHIFT 8 |
| 3440 | #define RIGHT_MASK_BIT_SHIFT_4x 16 | |
| 3441 | ||
| 75e28f62 | 3442 | #define setup_sprite_tile_column_width_single(texture_mode, multi_height, \ |
| 59d15d23 | 3443 | edge_mode, edge, x4mode) \ |
| 75e28f62 E |
3444 | { \ |
| 3445 | setup_sprite_column_data_##multi_height(); \ | |
| 3446 | left_mask_bits = left_block_mask | right_block_mask; \ | |
| 59d15d23 | 3447 | right_mask_bits = left_mask_bits >> RIGHT_MASK_BIT_SHIFT##x4mode; \ |
| 75e28f62 E |
3448 | \ |
| 3449 | setup_sprite_tile_column_height_##multi_height(edge_mode, edge, \ | |
| 59d15d23 | 3450 | texture_mode, x4mode); \ |
| 75e28f62 E |
3451 | } \ |
| 3452 | ||
| 3453 | #define setup_sprite_tiled_advance_column() \ | |
| 3454 | texture_offset_base += 0x100; \ | |
| 3455 | if((texture_offset_base & 0xF00) == 0) \ | |
| 3456 | texture_offset_base -= (0x100 + 0xF00) \ | |
| 3457 | ||
| 59d15d23 | 3458 | #define FB_PTR_MULTIPLIER 1 |
| 3459 | #define FB_PTR_MULTIPLIER_4x 2 | |
| 3460 | ||
| 75e28f62 | 3461 | #define setup_sprite_tile_column_width_multi(texture_mode, multi_height, \ |
| 59d15d23 | 3462 | left_mode, right_mode, x4mode) \ |
| 75e28f62 E |
3463 | { \ |
| 3464 | setup_sprite_column_data_##multi_height(); \ | |
| 59d15d23 | 3465 | s32 fb_ptr_advance_column = (16 - (1024 * height)) \ |
| 3466 | * FB_PTR_MULTIPLIER##x4mode; \ | |
| 75e28f62 E |
3467 | \ |
| 3468 | tile_width -= 2; \ | |
| 3469 | left_mask_bits = left_block_mask; \ | |
| 59d15d23 | 3470 | right_mask_bits = left_mask_bits >> RIGHT_MASK_BIT_SHIFT##x4mode; \ |
| 75e28f62 E |
3471 | \ |
| 3472 | setup_sprite_tile_column_height_##multi_height(left_mode, right, \ | |
| 59d15d23 | 3473 | texture_mode, x4mode); \ |
| 75e28f62 E |
3474 | fb_ptr += fb_ptr_advance_column; \ |
| 3475 | \ | |
| 3476 | left_mask_bits = 0x00; \ | |
| 3477 | right_mask_bits = 0x00; \ | |
| 3478 | \ | |
| 3479 | while(tile_width) \ | |
| 3480 | { \ | |
| 3481 | setup_sprite_tiled_advance_column(); \ | |
| 59d15d23 | 3482 | setup_sprite_tile_column_height_##multi_height(full, none, \ |
| 3483 | texture_mode, x4mode); \ | |
| 75e28f62 E |
3484 | fb_ptr += fb_ptr_advance_column; \ |
| 3485 | tile_width--; \ | |
| 3486 | } \ | |
| 3487 | \ | |
| 3488 | left_mask_bits = right_block_mask; \ | |
| 59d15d23 | 3489 | right_mask_bits = left_mask_bits >> RIGHT_MASK_BIT_SHIFT##x4mode; \ |
| 75e28f62 E |
3490 | \ |
| 3491 | setup_sprite_tiled_advance_column(); \ | |
| 3492 | setup_sprite_tile_column_height_##multi_height(right_mode, left, \ | |
| 59d15d23 | 3493 | texture_mode, x4mode); \ |
| 3494 | } \ | |
| 3495 | ||
| 3496 | ||
| 3497 | /* 4x stuff */ | |
| 3498 | #define setup_sprite_tiled_initialize_4bpp_4x() \ | |
| 3499 | setup_sprite_tiled_initialize_4bpp_clut() \ | |
| 3500 | ||
| 3501 | #define setup_sprite_tiled_initialize_8bpp_4x() \ | |
| 3502 | ||
| 3503 | ||
| 3504 | #define setup_sprite_tile_full_4bpp_4x(edge) \ | |
| 3505 | { \ | |
| 3506 | vec_8x8u texels_low, texels_high; \ | |
| 3507 | vec_8x16u pixels, pixels_wide; \ | |
| 3508 | setup_sprite_tile_add_blocks(sub_tile_height * 2 * 4); \ | |
| 3509 | u32 left_mask_bits_a = left_mask_bits & 0xFF; \ | |
| 3510 | u32 left_mask_bits_b = left_mask_bits >> 8; \ | |
| 3511 | u32 right_mask_bits_a = right_mask_bits & 0xFF; \ | |
| 3512 | u32 right_mask_bits_b = right_mask_bits >> 8; \ | |
| 3513 | \ | |
| 3514 | while(sub_tile_height) \ | |
| 3515 | { \ | |
| 3516 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ | |
| 3517 | tbl_16(texels_low, texels, clut_low); \ | |
| 3518 | tbl_16(texels_high, texels, clut_high); \ | |
| 3519 | zip_8x16b(pixels, texels_low, texels_high); \ | |
| 3520 | \ | |
| 3521 | zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.low, pixels.low); \ | |
| 3522 | block->texels = pixels_wide; \ | |
| 3523 | block->draw_mask_bits = left_mask_bits_a; \ | |
| 3524 | block->fb_ptr = fb_ptr; \ | |
| 3525 | block++; \ | |
| 3526 | \ | |
| 3527 | block->texels = pixels_wide; \ | |
| 3528 | block->draw_mask_bits = left_mask_bits_a; \ | |
| 3529 | block->fb_ptr = fb_ptr + 1024; \ | |
| 3530 | block++; \ | |
| 3531 | \ | |
| 3532 | zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.high, pixels.high); \ | |
| 3533 | block->texels = pixels_wide; \ | |
| 3534 | block->draw_mask_bits = left_mask_bits_b; \ | |
| 3535 | block->fb_ptr = fb_ptr + 8; \ | |
| 3536 | block++; \ | |
| 3537 | \ | |
| 3538 | block->texels = pixels_wide; \ | |
| 3539 | block->draw_mask_bits = left_mask_bits_b; \ | |
| 3540 | block->fb_ptr = fb_ptr + 1024 + 8; \ | |
| 3541 | block++; \ | |
| 3542 | \ | |
| 3543 | setup_sprite_tile_fetch_texel_block_8bpp(8); \ | |
| 3544 | tbl_16(texels_low, texels, clut_low); \ | |
| 3545 | tbl_16(texels_high, texels, clut_high); \ | |
| 3546 | zip_8x16b(pixels, texels_low, texels_high); \ | |
| 3547 | \ | |
| 3548 | zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.low, pixels.low); \ | |
| 3549 | block->texels = pixels_wide; \ | |
| 3550 | block->draw_mask_bits = right_mask_bits_a; \ | |
| 3551 | block->fb_ptr = fb_ptr + 16; \ | |
| 3552 | block++; \ | |
| 3553 | \ | |
| 3554 | block->texels = pixels_wide; \ | |
| 3555 | block->draw_mask_bits = right_mask_bits_a; \ | |
| 3556 | block->fb_ptr = fb_ptr + 1024 + 16; \ | |
| 3557 | block++; \ | |
| 3558 | \ | |
| 3559 | zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.high, pixels.high); \ | |
| 3560 | block->texels = pixels_wide; \ | |
| 3561 | block->draw_mask_bits = right_mask_bits_b; \ | |
| 3562 | block->fb_ptr = fb_ptr + 24; \ | |
| 3563 | block++; \ | |
| 3564 | \ | |
| 3565 | block->texels = pixels_wide; \ | |
| 3566 | block->draw_mask_bits = right_mask_bits_b; \ | |
| 3567 | block->fb_ptr = fb_ptr + 1024 + 24; \ | |
| 3568 | block++; \ | |
| 3569 | \ | |
| 3570 | fb_ptr += 2048; \ | |
| 3571 | texture_offset += 0x10; \ | |
| 3572 | sub_tile_height--; \ | |
| 3573 | } \ | |
| 3574 | texture_offset += 0xF00; \ | |
| 3575 | psx_gpu->num_blocks = num_blocks; \ | |
| 75e28f62 E |
3576 | } \ |
| 3577 | ||
| 59d15d23 | 3578 | #define setup_sprite_tile_half_4bpp_4x(edge) \ |
| 3579 | { \ | |
| 3580 | vec_8x8u texels_low, texels_high; \ | |
| 3581 | vec_8x16u pixels, pixels_wide; \ | |
| 3582 | setup_sprite_tile_add_blocks(sub_tile_height * 4); \ | |
| 3583 | u32 edge##_mask_bits_a = edge##_mask_bits & 0xFF; \ | |
| 3584 | u32 edge##_mask_bits_b = edge##_mask_bits >> 8; \ | |
| 3585 | \ | |
| 3586 | while(sub_tile_height) \ | |
| 3587 | { \ | |
| 3588 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ | |
| 3589 | tbl_16(texels_low, texels, clut_low); \ | |
| 3590 | tbl_16(texels_high, texels, clut_high); \ | |
| 3591 | zip_8x16b(pixels, texels_low, texels_high); \ | |
| 3592 | \ | |
| 3593 | zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.low, pixels.low); \ | |
| 3594 | block->texels = pixels_wide; \ | |
| 3595 | block->draw_mask_bits = edge##_mask_bits_a; \ | |
| 3596 | block->fb_ptr = fb_ptr; \ | |
| 3597 | block++; \ | |
| 3598 | \ | |
| 3599 | block->texels = pixels_wide; \ | |
| 3600 | block->draw_mask_bits = edge##_mask_bits_a; \ | |
| 3601 | block->fb_ptr = fb_ptr + 1024; \ | |
| 3602 | block++; \ | |
| 3603 | \ | |
| 3604 | zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.high, pixels.high); \ | |
| 3605 | block->texels = pixels_wide; \ | |
| 3606 | block->draw_mask_bits = edge##_mask_bits_b; \ | |
| 3607 | block->fb_ptr = fb_ptr + 8; \ | |
| 3608 | block++; \ | |
| 3609 | \ | |
| 3610 | block->texels = pixels_wide; \ | |
| 3611 | block->draw_mask_bits = edge##_mask_bits_b; \ | |
| 3612 | block->fb_ptr = fb_ptr + 1024 + 8; \ | |
| 3613 | block++; \ | |
| 3614 | \ | |
| 3615 | fb_ptr += 2048; \ | |
| 3616 | texture_offset += 0x10; \ | |
| 3617 | sub_tile_height--; \ | |
| 3618 | } \ | |
| 3619 | texture_offset += 0xF00; \ | |
| 3620 | psx_gpu->num_blocks = num_blocks; \ | |
| 3621 | } \ | |
| 75e28f62 | 3622 | |
| 59d15d23 | 3623 | |
| 3624 | #define setup_sprite_tile_full_8bpp_4x(edge) \ | |
| 3625 | { \ | |
| 3626 | setup_sprite_tile_add_blocks(sub_tile_height * 2 * 4); \ | |
| 3627 | vec_16x8u texels_wide; \ | |
| 3628 | u32 left_mask_bits_a = left_mask_bits & 0xFF; \ | |
| 3629 | u32 left_mask_bits_b = left_mask_bits >> 8; \ | |
| 3630 | u32 right_mask_bits_a = right_mask_bits & 0xFF; \ | |
| 3631 | u32 right_mask_bits_b = right_mask_bits >> 8; \ | |
| 3632 | \ | |
| 3633 | while(sub_tile_height) \ | |
| 3634 | { \ | |
| 3635 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ | |
| 3636 | zip_8x16b(vector_cast(vec_8x16u, texels_wide), texels, texels); \ | |
| 3637 | block->r = texels_wide.low; \ | |
| 3638 | block->draw_mask_bits = left_mask_bits_a; \ | |
| 3639 | block->fb_ptr = fb_ptr; \ | |
| 3640 | block++; \ | |
| 3641 | \ | |
| 3642 | block->r = texels_wide.low; \ | |
| 3643 | block->draw_mask_bits = left_mask_bits_a; \ | |
| 3644 | block->fb_ptr = fb_ptr + 1024; \ | |
| 3645 | block++; \ | |
| 3646 | \ | |
| 3647 | block->r = texels_wide.high; \ | |
| 3648 | block->draw_mask_bits = left_mask_bits_b; \ | |
| 3649 | block->fb_ptr = fb_ptr + 8; \ | |
| 3650 | block++; \ | |
| 3651 | \ | |
| 3652 | block->r = texels_wide.high; \ | |
| 3653 | block->draw_mask_bits = left_mask_bits_b; \ | |
| 3654 | block->fb_ptr = fb_ptr + 1024 + 8; \ | |
| 3655 | block++; \ | |
| 3656 | \ | |
| 3657 | setup_sprite_tile_fetch_texel_block_8bpp(8); \ | |
| 3658 | zip_8x16b(vector_cast(vec_8x16u, texels_wide), texels, texels); \ | |
| 3659 | block->r = texels_wide.low; \ | |
| 3660 | block->draw_mask_bits = right_mask_bits_a; \ | |
| 3661 | block->fb_ptr = fb_ptr + 16; \ | |
| 3662 | block++; \ | |
| 3663 | \ | |
| 3664 | block->r = texels_wide.low; \ | |
| 3665 | block->draw_mask_bits = right_mask_bits_a; \ | |
| 3666 | block->fb_ptr = fb_ptr + 1024 + 16; \ | |
| 3667 | block++; \ | |
| 3668 | \ | |
| 3669 | block->r = texels_wide.high; \ | |
| 3670 | block->draw_mask_bits = right_mask_bits_b; \ | |
| 3671 | block->fb_ptr = fb_ptr + 24; \ | |
| 3672 | block++; \ | |
| 3673 | \ | |
| 3674 | block->r = texels_wide.high; \ | |
| 3675 | block->draw_mask_bits = right_mask_bits_b; \ | |
| 3676 | block->fb_ptr = fb_ptr + 24 + 1024; \ | |
| 3677 | block++; \ | |
| 3678 | \ | |
| 3679 | fb_ptr += 2048; \ | |
| 3680 | texture_offset += 0x10; \ | |
| 3681 | sub_tile_height--; \ | |
| 3682 | } \ | |
| 3683 | texture_offset += 0xF00; \ | |
| 3684 | psx_gpu->num_blocks = num_blocks; \ | |
| 3685 | } \ | |
| 3686 | ||
| 3687 | #define setup_sprite_tile_half_8bpp_4x(edge) \ | |
| 3688 | { \ | |
| 3689 | setup_sprite_tile_add_blocks(sub_tile_height * 4); \ | |
| 3690 | vec_16x8u texels_wide; \ | |
| 3691 | u32 edge##_mask_bits_a = edge##_mask_bits & 0xFF; \ | |
| 3692 | u32 edge##_mask_bits_b = edge##_mask_bits >> 8; \ | |
| 3693 | \ | |
| 3694 | while(sub_tile_height) \ | |
| 3695 | { \ | |
| 3696 | setup_sprite_tile_fetch_texel_block_8bpp(0); \ | |
| 3697 | zip_8x16b(vector_cast(vec_8x16u, texels_wide), texels, texels); \ | |
| 3698 | block->r = texels_wide.low; \ | |
| 3699 | block->draw_mask_bits = edge##_mask_bits_a; \ | |
| 3700 | block->fb_ptr = fb_ptr; \ | |
| 3701 | block++; \ | |
| 3702 | \ | |
| 3703 | block->r = texels_wide.low; \ | |
| 3704 | block->draw_mask_bits = edge##_mask_bits_a; \ | |
| 3705 | block->fb_ptr = fb_ptr + 1024; \ | |
| 3706 | block++; \ | |
| 3707 | \ | |
| 3708 | block->r = texels_wide.high; \ | |
| 3709 | block->draw_mask_bits = edge##_mask_bits_b; \ | |
| 3710 | block->fb_ptr = fb_ptr + 8; \ | |
| 3711 | block++; \ | |
| 3712 | \ | |
| 3713 | block->r = texels_wide.high; \ | |
| 3714 | block->draw_mask_bits = edge##_mask_bits_b; \ | |
| 3715 | block->fb_ptr = fb_ptr + 8 + 1024; \ | |
| 3716 | block++; \ | |
| 3717 | \ | |
| 3718 | fb_ptr += 2048; \ | |
| 3719 | texture_offset += 0x10; \ | |
| 3720 | sub_tile_height--; \ | |
| 3721 | } \ | |
| 3722 | texture_offset += 0xF00; \ | |
| 3723 | psx_gpu->num_blocks = num_blocks; \ | |
| 3724 | } \ | |
| 3725 | ||
| 3726 | ||
| 3727 | #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \ | |
| 3728 | texture_offset = texture_offset_base + 8; \ | |
| 3729 | fb_ptr += 16 \ | |
| 3730 | ||
| 3731 | #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \ | |
| 3732 | texture_offset = texture_offset_base \ | |
| 3733 | ||
| 3734 | #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \ | |
| 3735 | setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \ | |
| 3736 | ||
| 3737 | #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \ | |
| 3738 | texture_offset = texture_offset_base \ | |
| 3739 | ||
| 3740 | #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \ | |
| 3741 | fb_ptr -= 16 \ | |
| 3742 | ||
| 3743 | #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \ | |
| 3744 | ||
| 3745 | #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \ | |
| 3746 | setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \ | |
| 3747 | ||
| 3748 | #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \ | |
| 3749 | ||
| 3750 | ||
| 3751 | #define setup_sprite_offset_u_adjust() \ | |
| 3752 | ||
| 3753 | #define setup_sprite_comapre_left_block_mask() \ | |
| 3754 | ((left_block_mask & 0xFF) == 0xFF) \ | |
| 3755 | ||
| 3756 | #define setup_sprite_comapre_right_block_mask() \ | |
| 3757 | (((right_block_mask >> 8) & 0xFF) == 0xFF) \ | |
| 3758 | ||
| 3759 | ||
| 3760 | #define setup_sprite_offset_u_adjust_4x() \ | |
| 3761 | offset_u *= 2; \ | |
| 3762 | offset_u_right = offset_u_right * 2 + 1 \ | |
| 3763 | ||
| 3764 | #define setup_sprite_comapre_left_block_mask_4x() \ | |
| 3765 | ((left_block_mask & 0xFFFF) == 0xFFFF) \ | |
| 3766 | ||
| 3767 | #define setup_sprite_comapre_right_block_mask_4x() \ | |
| 3768 | (((right_block_mask >> 16) & 0xFFFF) == 0xFFFF) \ | |
| 3769 | ||
| 3770 | ||
| 3771 | #define setup_sprite_tiled_builder(texture_mode, x4mode) \ | |
| 3772 | void setup_sprite_##texture_mode##x4mode(psx_gpu_struct *psx_gpu, s32 x, s32 y,\ | |
| 75e28f62 E |
3773 | s32 u, s32 v, s32 width, s32 height, u32 color) \ |
| 3774 | { \ | |
| 3775 | s32 offset_u = u & 0xF; \ | |
| 3776 | s32 offset_v = v & 0xF; \ | |
| 3777 | \ | |
| 3778 | s32 width_rounded = offset_u + width + 15; \ | |
| 3779 | s32 height_rounded = offset_v + height + 15; \ | |
| 3780 | s32 tile_height = height_rounded / 16; \ | |
| 3781 | s32 tile_width = width_rounded / 16; \ | |
| 3782 | u32 offset_u_right = width_rounded & 0xF; \ | |
| 3783 | \ | |
| 59d15d23 | 3784 | setup_sprite_offset_u_adjust##x4mode(); \ |
| 3785 | \ | |
| 3786 | u32 left_block_mask = ~(0xFFFFFFFF << offset_u); \ | |
| 3787 | u32 right_block_mask = 0xFFFFFFFE << offset_u_right; \ | |
| 75e28f62 E |
3788 | \ |
| 3789 | u32 left_mask_bits; \ | |
| 3790 | u32 right_mask_bits; \ | |
| 3791 | \ | |
| 3792 | u32 sub_tile_height; \ | |
| 3793 | u32 column_data; \ | |
| 3794 | \ | |
| 3795 | u32 texture_mask = (psx_gpu->texture_mask_width & 0xF) | \ | |
| 3796 | ((psx_gpu->texture_mask_height & 0xF) << 4) | \ | |
| 3797 | ((psx_gpu->texture_mask_width >> 4) << 8) | \ | |
| 3798 | ((psx_gpu->texture_mask_height >> 4) << 12); \ | |
| 3799 | u32 texture_offset = ((v & 0xF) << 4) | ((u & 0xF0) << 4) | \ | |
| 3800 | ((v & 0xF0) << 8); \ | |
| 3801 | u32 texture_offset_base = texture_offset; \ | |
| 3802 | u32 control_mask; \ | |
| 3803 | \ | |
| 59d15d23 | 3804 | u16 *fb_ptr = psx_gpu->vram_out_ptr + (y * 1024) + (x - offset_u); \ |
| 75e28f62 E |
3805 | u32 num_blocks = psx_gpu->num_blocks; \ |
| 3806 | block_struct *block = psx_gpu->blocks + num_blocks; \ | |
| 3807 | \ | |
| df740cdc | 3808 | u8 *texture_block_ptr; \ |
| 75e28f62 E |
3809 | vec_8x8u texels; \ |
| 3810 | \ | |
| 59d15d23 | 3811 | setup_sprite_tiled_initialize_##texture_mode##x4mode(); \ |
| 75e28f62 E |
3812 | \ |
| 3813 | control_mask = tile_width == 1; \ | |
| 3814 | control_mask |= (tile_height == 1) << 1; \ | |
| 59d15d23 | 3815 | control_mask |= setup_sprite_comapre_left_block_mask##x4mode() << 2; \ |
| 3816 | control_mask |= setup_sprite_comapre_right_block_mask##x4mode() << 3; \ | |
| 75e28f62 E |
3817 | \ |
| 3818 | sprites_##texture_mode++; \ | |
| 3819 | \ | |
| 3820 | switch(control_mask) \ | |
| 3821 | { \ | |
| 3822 | default: \ | |
| 3823 | case 0x0: \ | |
| 59d15d23 | 3824 | setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \ |
| 3825 | x4mode); \ | |
| 75e28f62 E |
3826 | break; \ |
| 3827 | \ | |
| 3828 | case 0x1: \ | |
| 59d15d23 | 3829 | setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \ |
| 3830 | x4mode); \ | |
| 75e28f62 E |
3831 | break; \ |
| 3832 | \ | |
| 3833 | case 0x2: \ | |
| 59d15d23 | 3834 | setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \ |
| 3835 | x4mode); \ | |
| 75e28f62 E |
3836 | break; \ |
| 3837 | \ | |
| 3838 | case 0x3: \ | |
| 59d15d23 | 3839 | setup_sprite_tile_column_width_single(texture_mode, single, full, none, \ |
| 3840 | x4mode); \ | |
| 75e28f62 E |
3841 | break; \ |
| 3842 | \ | |
| 3843 | case 0x4: \ | |
| 59d15d23 | 3844 | setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \ |
| 3845 | x4mode); \ | |
| 75e28f62 E |
3846 | break; \ |
| 3847 | \ | |
| 3848 | case 0x5: \ | |
| 59d15d23 | 3849 | setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \ |
| 3850 | x4mode); \ | |
| 75e28f62 E |
3851 | break; \ |
| 3852 | \ | |
| 3853 | case 0x6: \ | |
| 59d15d23 | 3854 | setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \ |
| 3855 | x4mode); \ | |
| 75e28f62 E |
3856 | break; \ |
| 3857 | \ | |
| 3858 | case 0x7: \ | |
| 59d15d23 | 3859 | setup_sprite_tile_column_width_single(texture_mode, single, half, right, \ |
| 3860 | x4mode); \ | |
| 75e28f62 E |
3861 | break; \ |
| 3862 | \ | |
| 3863 | case 0x8: \ | |
| 59d15d23 | 3864 | setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \ |
| 3865 | x4mode); \ | |
| 75e28f62 E |
3866 | break; \ |
| 3867 | \ | |
| 3868 | case 0x9: \ | |
| 59d15d23 | 3869 | setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \ |
| 3870 | x4mode); \ | |
| 75e28f62 E |
3871 | break; \ |
| 3872 | \ | |
| 3873 | case 0xA: \ | |
| 59d15d23 | 3874 | setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \ |
| 3875 | x4mode); \ | |
| 75e28f62 E |
3876 | break; \ |
| 3877 | \ | |
| 3878 | case 0xB: \ | |
| 59d15d23 | 3879 | setup_sprite_tile_column_width_single(texture_mode, single, half, left, \ |
| 3880 | x4mode); \ | |
| 75e28f62 E |
3881 | break; \ |
| 3882 | \ | |
| 3883 | case 0xC: \ | |
| 59d15d23 | 3884 | setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \ |
| 3885 | x4mode); \ | |
| 75e28f62 E |
3886 | break; \ |
| 3887 | \ | |
| 3888 | case 0xE: \ | |
| 59d15d23 | 3889 | setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \ |
| 3890 | x4mode); \ | |
| 75e28f62 E |
3891 | break; \ |
| 3892 | } \ | |
| 3893 | } \ | |
| 3894 | ||
| 2bbbb7af | 3895 | #ifndef NEON_BUILD |
| 59d15d23 | 3896 | setup_sprite_tiled_builder(4bpp,); |
| 3897 | setup_sprite_tiled_builder(8bpp,); | |
| 3898 | ||
| 3899 | setup_sprite_tiled_builder(4bpp,_4x); | |
| 3900 | setup_sprite_tiled_builder(8bpp,_4x); | |
| a2cb152a | 3901 | #endif |
| 3902 | ||
| 3903 | #if !defined(NEON_BUILD) || defined(SIMD_BUILD) | |
| 75e28f62 E |
3904 | |
| 3905 | void setup_sprite_16bpp(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u, | |
| 3906 | s32 v, s32 width, s32 height, u32 color) | |
| 3907 | { | |
| 3908 | u32 left_offset = u & 0x7; | |
| 3909 | u32 width_rounded = width + left_offset + 7; | |
| 3910 | ||
| fc6cef7d | 3911 | u16 *fb_ptr = psx_gpu->vram_out_ptr + (y * 1024) + (s32)(x - left_offset); |
| 75e28f62 E |
3912 | u32 right_width = width_rounded & 0x7; |
| 3913 | u32 block_width = width_rounded / 8; | |
| 3914 | u32 fb_ptr_pitch = (1024 + 8) - (block_width * 8); | |
| 3915 | ||
| 3916 | u32 left_mask_bits = ~(0xFF << left_offset); | |
| 3917 | u32 right_mask_bits = 0xFE << right_width; | |
| 3918 | ||
| 3919 | u32 texture_offset_base = u + (v * 1024); | |
| 3920 | u32 texture_mask = | |
| 3921 | psx_gpu->texture_mask_width | (psx_gpu->texture_mask_height * 1024); | |
| 3922 | ||
| 3923 | u32 blocks_remaining; | |
| 3924 | u32 num_blocks = psx_gpu->num_blocks; | |
| 3925 | block_struct *block = psx_gpu->blocks + num_blocks; | |
| 3926 | ||
| 3927 | u16 *texture_page_ptr = psx_gpu->texture_page_ptr; | |
| 3928 | u16 *texture_block_ptr; | |
| 3929 | ||
| 3930 | texture_offset_base &= ~0x7; | |
| 3931 | ||
| a2cb152a | 3932 | stats_add(sprites_16bpp, 1); |
| 75e28f62 E |
3933 | |
| 3934 | if(block_width == 1) | |
| 3935 | { | |
| 3936 | u32 mask_bits = left_mask_bits | right_mask_bits; | |
| 3937 | ||
| 3938 | while(height) | |
| 3939 | { | |
| 3940 | num_blocks++; | |
| 3941 | sprite_blocks++; | |
| 3942 | ||
| 3943 | if(num_blocks > MAX_BLOCKS) | |
| 3944 | { | |
| 3945 | flush_render_block_buffer(psx_gpu); | |
| 3946 | num_blocks = 1; | |
| 3947 | block = psx_gpu->blocks; | |
| 3948 | } | |
| 3949 | ||
| 3950 | texture_block_ptr = | |
| 3951 | texture_page_ptr + (texture_offset_base & texture_mask); | |
| 3952 | ||
| a2cb152a | 3953 | block->texels = *(vec_8x16u *)texture_block_ptr; |
| 75e28f62 E |
3954 | block->draw_mask_bits = mask_bits; |
| 3955 | block->fb_ptr = fb_ptr; | |
| 3956 | ||
| 3957 | block++; | |
| 3958 | ||
| 3959 | texture_offset_base += 1024; | |
| 3960 | fb_ptr += 1024; | |
| 3961 | ||
| 3962 | height--; | |
| 3963 | psx_gpu->num_blocks = num_blocks; | |
| 3964 | } | |
| 3965 | } | |
| 3966 | else | |
| 3967 | { | |
| 3968 | u32 texture_offset; | |
| 3969 | ||
| 3970 | while(height) | |
| 3971 | { | |
| 3972 | blocks_remaining = block_width - 2; | |
| 3973 | num_blocks += block_width; | |
| 3974 | sprite_blocks += block_width; | |
| 3975 | ||
| 3976 | if(num_blocks > MAX_BLOCKS) | |
| 3977 | { | |
| 3978 | flush_render_block_buffer(psx_gpu); | |
| 3979 | num_blocks = block_width; | |
| 3980 | block = psx_gpu->blocks; | |
| 3981 | } | |
| 3982 | ||
| 3983 | texture_offset = texture_offset_base; | |
| 3984 | texture_offset_base += 1024; | |
| 3985 | ||
| 3986 | texture_block_ptr = texture_page_ptr + (texture_offset & texture_mask); | |
| a2cb152a | 3987 | block->texels = *(vec_8x16u *)texture_block_ptr; |
| 75e28f62 E |
3988 | |
| 3989 | block->draw_mask_bits = left_mask_bits; | |
| 3990 | block->fb_ptr = fb_ptr; | |
| 3991 | ||
| 3992 | texture_offset += 8; | |
| 3993 | fb_ptr += 8; | |
| 3994 | block++; | |
| 3995 | ||
| 3996 | while(blocks_remaining) | |
| 3997 | { | |
| 3998 | texture_block_ptr = texture_page_ptr + (texture_offset & texture_mask); | |
| a2cb152a | 3999 | block->texels = *(vec_8x16u *)texture_block_ptr; |
| 75e28f62 E |
4000 | |
| 4001 | block->draw_mask_bits = 0; | |
| 4002 | block->fb_ptr = fb_ptr; | |
| 4003 | ||
| 4004 | texture_offset += 8; | |
| 4005 | fb_ptr += 8; | |
| 4006 | block++; | |
| 4007 | ||
| 4008 | blocks_remaining--; | |
| 4009 | } | |
| 4010 | ||
| 4011 | texture_block_ptr = texture_page_ptr + (texture_offset & texture_mask); | |
| a2cb152a | 4012 | block->texels = *(vec_8x16u *)texture_block_ptr; |
| 75e28f62 E |
4013 | |
| 4014 | block->draw_mask_bits = right_mask_bits; | |
| 4015 | block->fb_ptr = fb_ptr; | |
| 4016 | ||
| 4017 | fb_ptr += fb_ptr_pitch; | |
| 4018 | block++; | |
| 4019 | ||
| 4020 | height--; | |
| 4021 | psx_gpu->num_blocks = num_blocks; | |
| 4022 | } | |
| 4023 | } | |
| 4024 | } | |
| 4025 | ||
| a2cb152a | 4026 | #endif |
| 4027 | ||
| 4028 | #ifndef NEON_BUILD | |
| 4029 | ||
| 2d658c89 | 4030 | void setup_sprite_untextured_512(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u, |
| 75e28f62 E |
4031 | s32 v, s32 width, s32 height, u32 color) |
| 4032 | { | |
| 4033 | u32 right_width = ((width - 1) & 0x7) + 1; | |
| 4034 | u32 right_mask_bits = (0xFF << right_width); | |
| fc6cef7d | 4035 | u16 *fb_ptr = psx_gpu->vram_out_ptr + (y * 1024) + x; |
| 75e28f62 E |
4036 | u32 block_width = (width + 7) / 8; |
| 4037 | u32 fb_ptr_pitch = 1024 - ((block_width - 1) * 8); | |
| 4038 | u32 blocks_remaining; | |
| 4039 | u32 num_blocks = psx_gpu->num_blocks; | |
| 4040 | block_struct *block = psx_gpu->blocks + num_blocks; | |
| 4041 | ||
| 4042 | u32 color_r = color & 0xFF; | |
| 4043 | u32 color_g = (color >> 8) & 0xFF; | |
| 4044 | u32 color_b = (color >> 16) & 0xFF; | |
| 4045 | vec_8x16u colors; | |
| 4046 | vec_8x16u right_mask; | |
| 4047 | vec_8x16u test_mask = psx_gpu->test_mask; | |
| 4048 | vec_8x16u zero_mask; | |
| 4049 | ||
| 4050 | sprites_untextured++; | |
| 4051 | ||
| 4052 | color = (color_r >> 3) | ((color_g >> 3) << 5) | ((color_b >> 3) << 10); | |
| 4053 | ||
| 4054 | dup_8x16b(colors, color); | |
| 4055 | dup_8x16b(zero_mask, 0x00); | |
| 4056 | dup_8x16b(right_mask, right_mask_bits); | |
| 4057 | tst_8x16b(right_mask, right_mask, test_mask); | |
| 4058 | ||
| 4059 | while(height) | |
| 4060 | { | |
| 4061 | blocks_remaining = block_width - 1; | |
| 4062 | num_blocks += block_width; | |
| 3867c6ef E |
4063 | |
| 4064 | #ifdef PROFILE | |
| 75e28f62 | 4065 | sprite_blocks += block_width; |
| 3867c6ef | 4066 | #endif |
| 75e28f62 E |
4067 | |
| 4068 | if(num_blocks > MAX_BLOCKS) | |
| 4069 | { | |
| 4070 | flush_render_block_buffer(psx_gpu); | |
| 4071 | num_blocks = block_width; | |
| 4072 | block = psx_gpu->blocks; | |
| 4073 | } | |
| 4074 | ||
| 4075 | while(blocks_remaining) | |
| 4076 | { | |
| 4077 | block->pixels = colors; | |
| 4078 | block->draw_mask = zero_mask; | |
| 4079 | block->fb_ptr = fb_ptr; | |
| 4080 | ||
| 4081 | fb_ptr += 8; | |
| 4082 | block++; | |
| 4083 | blocks_remaining--; | |
| 4084 | } | |
| 4085 | ||
| 4086 | block->pixels = colors; | |
| 4087 | block->draw_mask = right_mask; | |
| 4088 | block->fb_ptr = fb_ptr; | |
| 4089 | ||
| 4090 | block++; | |
| 4091 | fb_ptr += fb_ptr_pitch; | |
| 4092 | ||
| 4093 | height--; | |
| 4094 | psx_gpu->num_blocks = num_blocks; | |
| 4095 | } | |
| 4096 | } | |
| 4097 | ||
| f0931e56 | 4098 | #endif |
| 4099 | ||
| 2d658c89 | 4100 | static void __attribute__((noinline)) |
| 4101 | setup_sprite_untextured_simple(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u, | |
| 4102 | s32 v, s32 width, s32 height, u32 color) | |
| f0931e56 | 4103 | { |
| 4104 | u32 r = color & 0xFF; | |
| 4105 | u32 g = (color >> 8) & 0xFF; | |
| 4106 | u32 b = (color >> 16) & 0xFF; | |
| 4107 | u32 color_16bpp = (r >> 3) | ((g >> 3) << 5) | ((b >> 3) << 10) | | |
| 4108 | psx_gpu->mask_msb; | |
| 4109 | u32 color_32bpp = color_16bpp | (color_16bpp << 16); | |
| 4110 | ||
| 4111 | u16 *vram_ptr16 = psx_gpu->vram_out_ptr + x + (y * 1024); | |
| 4112 | u32 *vram_ptr; | |
| 4113 | ||
| 4114 | u32 num_width; | |
| 4115 | ||
| 2d658c89 | 4116 | if(psx_gpu->num_blocks) |
| f0931e56 | 4117 | { |
| 4118 | flush_render_block_buffer(psx_gpu); | |
| 4119 | } | |
| 4120 | ||
| 4121 | while(height) | |
| 4122 | { | |
| 4123 | num_width = width; | |
| 4124 | ||
| 4125 | vram_ptr = (void *)vram_ptr16; | |
| 77e1e479 | 4126 | if((uintptr_t)vram_ptr16 & 2) |
| f0931e56 | 4127 | { |
| 4128 | *vram_ptr16 = color_32bpp; | |
| 4129 | vram_ptr = (void *)(vram_ptr16 + 1); | |
| 4130 | num_width--; | |
| 4131 | } | |
| 4132 | ||
| 4133 | while(num_width >= 4 * 2) | |
| 4134 | { | |
| 4135 | vram_ptr[0] = color_32bpp; | |
| 4136 | vram_ptr[1] = color_32bpp; | |
| 4137 | vram_ptr[2] = color_32bpp; | |
| 4138 | vram_ptr[3] = color_32bpp; | |
| 4139 | ||
| 4140 | vram_ptr += 4; | |
| 4141 | num_width -= 4 * 2; | |
| 4142 | } | |
| 4143 | ||
| 4144 | while(num_width >= 2) | |
| 4145 | { | |
| 4146 | *vram_ptr++ = color_32bpp; | |
| 4147 | num_width -= 2; | |
| 4148 | } | |
| 4149 | ||
| 4150 | if(num_width > 0) | |
| 4151 | { | |
| 4152 | *(u16 *)vram_ptr = color_32bpp; | |
| 4153 | } | |
| 4154 | ||
| 4155 | vram_ptr16 += 1024; | |
| 4156 | height--; | |
| 4157 | } | |
| 2d658c89 | 4158 | } |
| 4159 | ||
| 4160 | void setup_sprite_untextured_512(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u, | |
| 4161 | s32 v, s32 width, s32 height, u32 color); | |
| 4162 | ||
| 4163 | void setup_sprite_untextured(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u, | |
| 4164 | s32 v, s32 width, s32 height, u32 color) | |
| 4165 | { | |
| 4166 | if((psx_gpu->render_state & (RENDER_STATE_MASK_EVALUATE | | |
| 4167 | RENDER_FLAGS_MODULATE_TEXELS | RENDER_FLAGS_BLEND)) == 0 && | |
| 4168 | (psx_gpu->render_mode & RENDER_INTERLACE_ENABLED) == 0) | |
| 4169 | { | |
| 4170 | setup_sprite_untextured_simple(psx_gpu, x, y, u, v, width, height, color); | |
| 4171 | return; | |
| 4172 | } | |
| 4173 | ||
| 4174 | while (width > 0) | |
| 4175 | { | |
| 4176 | s32 w1 = width > 512 ? 512 : width; | |
| 4177 | setup_sprite_untextured_512(psx_gpu, x, y, 0, 0, w1, height, color); | |
| 4178 | x += 512; | |
| 4179 | width -= 512; | |
| 4180 | } | |
| f0931e56 | 4181 | } |
| 75e28f62 E |
4182 | |
| 4183 | ||
| 4184 | #define setup_sprite_blocks_switch_textured(texture_mode) \ | |
| 4185 | setup_sprite_##texture_mode \ | |
| 4186 | ||
| 4187 | #define setup_sprite_blocks_switch_untextured(texture_mode) \ | |
| 4188 | setup_sprite_untextured \ | |
| 4189 | ||
| 4190 | #define setup_sprite_blocks_switch(texturing, texture_mode) \ | |
| 4191 | setup_sprite_blocks_switch_##texturing(texture_mode) \ | |
| 4192 | ||
| 4193 | ||
| 4194 | #define texture_sprite_blocks_switch_4bpp() \ | |
| 4195 | texture_blocks_untextured \ | |
| 4196 | ||
| 4197 | #define texture_sprite_blocks_switch_8bpp() \ | |
| 4198 | texture_sprite_blocks_8bpp \ | |
| 4199 | ||
| 4200 | #define texture_sprite_blocks_switch_16bpp() \ | |
| 4201 | texture_blocks_untextured \ | |
| 4202 | ||
| 4203 | #define texture_sprite_blocks_switch_untextured(texture_mode) \ | |
| 4204 | texture_blocks_untextured \ | |
| 4205 | ||
| 4206 | #define texture_sprite_blocks_switch_textured(texture_mode) \ | |
| 4207 | texture_sprite_blocks_switch_##texture_mode() \ | |
| 4208 | ||
| 4209 | #define render_sprite_blocks_switch_block_modulation(texture_mode, blend_mode, \ | |
| 4210 | mask_evaluate, shading, dithering, texturing, blending, modulation) \ | |
| 4211 | { \ | |
| 4212 | setup_sprite_blocks_switch(texturing, texture_mode), \ | |
| 4213 | texture_sprite_blocks_switch_##texturing(texture_mode), \ | |
| 4214 | shade_blocks_switch(unshaded, texturing, modulation, undithered, blending, \ | |
| 4215 | mask_evaluate), \ | |
| 4216 | blend_blocks_switch(texturing, blending, blend_mode, mask_evaluate) \ | |
| 4217 | } \ | |
| 4218 | ||
| 4219 | #define render_sprite_blocks_switch_block_blending(texture_mode, blend_mode, \ | |
| 4220 | mask_evaluate, shading, dithering, texturing, blending) \ | |
| 4221 | render_sprite_blocks_switch_block_modulation(texture_mode, blend_mode, \ | |
| 4222 | mask_evaluate, shading, dithering, texturing, blending, modulated), \ | |
| 4223 | render_sprite_blocks_switch_block_modulation(texture_mode, blend_mode, \ | |
| 4224 | mask_evaluate, shading, dithering, texturing, blending, unmodulated) \ | |
| 4225 | ||
| 4226 | #define render_sprite_blocks_switch_block_texturing(texture_mode, blend_mode, \ | |
| 4227 | mask_evaluate, shading, dithering, texturing) \ | |
| 4228 | render_sprite_blocks_switch_block_blending(texture_mode, blend_mode, \ | |
| 4229 | mask_evaluate, shading, dithering, texturing, unblended), \ | |
| 4230 | render_sprite_blocks_switch_block_blending(texture_mode, blend_mode, \ | |
| 4231 | mask_evaluate, shading, dithering, texturing, blended) \ | |
| 4232 | ||
| 4233 | #define render_sprite_blocks_switch_block_dithering(texture_mode, blend_mode, \ | |
| 4234 | mask_evaluate, shading, dithering) \ | |
| 4235 | render_sprite_blocks_switch_block_texturing(texture_mode, blend_mode, \ | |
| 4236 | mask_evaluate, shading, dithering, untextured), \ | |
| 4237 | render_sprite_blocks_switch_block_texturing(texture_mode, blend_mode, \ | |
| 4238 | mask_evaluate, shading, dithering, textured) \ | |
| 4239 | ||
| 4240 | #define render_sprite_blocks_switch_block_shading(texture_mode, blend_mode, \ | |
| 4241 | mask_evaluate, shading) \ | |
| 4242 | render_sprite_blocks_switch_block_dithering(texture_mode, blend_mode, \ | |
| 4243 | mask_evaluate, shading, undithered), \ | |
| 4244 | render_sprite_blocks_switch_block_dithering(texture_mode, blend_mode, \ | |
| 4245 | mask_evaluate, shading, dithered) \ | |
| 4246 | ||
| 4247 | #define render_sprite_blocks_switch_block_mask_evaluate(texture_mode, \ | |
| 4248 | blend_mode, mask_evaluate) \ | |
| 4249 | render_sprite_blocks_switch_block_shading(texture_mode, blend_mode, \ | |
| 4250 | mask_evaluate, unshaded), \ | |
| 4251 | render_sprite_blocks_switch_block_shading(texture_mode, blend_mode, \ | |
| 4252 | mask_evaluate, shaded) \ | |
| 4253 | ||
| 4254 | #define render_sprite_blocks_switch_block_blend_mode(texture_mode, blend_mode) \ | |
| 4255 | render_sprite_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, \ | |
| 4256 | off), \ | |
| 4257 | render_sprite_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, \ | |
| 4258 | on) \ | |
| 4259 | ||
| 4260 | #define render_sprite_blocks_switch_block_texture_mode(texture_mode) \ | |
| 4261 | render_sprite_blocks_switch_block_blend_mode(texture_mode, average), \ | |
| 4262 | render_sprite_blocks_switch_block_blend_mode(texture_mode, add), \ | |
| 4263 | render_sprite_blocks_switch_block_blend_mode(texture_mode, subtract), \ | |
| 4264 | render_sprite_blocks_switch_block_blend_mode(texture_mode, add_fourth) \ | |
| 4265 | ||
| 4266 | #define render_sprite_blocks_switch_block() \ | |
| 4267 | render_sprite_blocks_switch_block_texture_mode(4bpp), \ | |
| 4268 | render_sprite_blocks_switch_block_texture_mode(8bpp), \ | |
| 4269 | render_sprite_blocks_switch_block_texture_mode(16bpp), \ | |
| b7f5c059 | 4270 | render_sprite_blocks_switch_block_texture_mode(16bpp) \ |
| 75e28f62 E |
4271 | |
| 4272 | ||
| 4273 | render_block_handler_struct render_sprite_block_handlers[] = | |
| 4274 | { | |
| 4275 | render_sprite_blocks_switch_block() | |
| 4276 | }; | |
| 4277 | ||
| 4278 | ||
| 4279 | void render_sprite(psx_gpu_struct *psx_gpu, s32 x, s32 y, u32 u, u32 v, | |
| 1cec4719 | 4280 | s32 *width, s32 *height, u32 flags, u32 color) |
| 75e28f62 | 4281 | { |
| 1cec4719 | 4282 | s32 x_right = x + *width - 1; |
| 4283 | s32 y_bottom = y + *height - 1; | |
| 75e28f62 | 4284 | |
| 3867c6ef E |
4285 | #ifdef PROFILE |
| 4286 | sprites++; | |
| 4287 | #endif | |
| 4288 | ||
| 75e28f62 E |
4289 | if(invalidate_texture_cache_region_viewport(psx_gpu, x, y, x_right, |
| 4290 | y_bottom) == 0) | |
| 4291 | { | |
| 1cec4719 | 4292 | *width = *height = 0; |
| 75e28f62 E |
4293 | return; |
| 4294 | } | |
| 4295 | ||
| 4296 | if(x < psx_gpu->viewport_start_x) | |
| 4297 | { | |
| 4298 | u32 clip = psx_gpu->viewport_start_x - x; | |
| 4299 | x += clip; | |
| 4300 | u += clip; | |
| 1cec4719 | 4301 | *width -= clip; |
| 75e28f62 E |
4302 | } |
| 4303 | ||
| 4304 | if(y < psx_gpu->viewport_start_y) | |
| 4305 | { | |
| 4306 | s32 clip = psx_gpu->viewport_start_y - y; | |
| 4307 | y += clip; | |
| 4308 | v += clip; | |
| 1cec4719 | 4309 | *height -= clip; |
| 75e28f62 E |
4310 | } |
| 4311 | ||
| 4312 | if(x_right > psx_gpu->viewport_end_x) | |
| 1cec4719 | 4313 | *width -= x_right - psx_gpu->viewport_end_x; |
| 75e28f62 E |
4314 | |
| 4315 | if(y_bottom > psx_gpu->viewport_end_y) | |
| 1cec4719 | 4316 | *height -= y_bottom - psx_gpu->viewport_end_y; |
| 75e28f62 | 4317 | |
| 1cec4719 | 4318 | if((*width <= 0) || (*height <= 0)) |
| 4319 | { | |
| 4320 | *width = *height = 0; | |
| 75e28f62 | 4321 | return; |
| 1cec4719 | 4322 | } |
| 75e28f62 | 4323 | |
| 3867c6ef | 4324 | #ifdef PROFILE |
| 1cec4719 | 4325 | span_pixels += *width * *height; |
| 4326 | spans += *height; | |
| 3867c6ef | 4327 | #endif |
| 75e28f62 E |
4328 | |
| 4329 | u32 render_state = flags & | |
| 4330 | (RENDER_FLAGS_MODULATE_TEXELS | RENDER_FLAGS_BLEND | | |
| 4331 | RENDER_FLAGS_TEXTURE_MAP); | |
| 4332 | render_state |= | |
| 4333 | (psx_gpu->render_state_base & ~RENDER_STATE_DITHER); | |
| 1f88961f | 4334 | |
| 75e28f62 E |
4335 | if((psx_gpu->render_state != render_state) || |
| 4336 | (psx_gpu->primitive_type != PRIMITIVE_TYPE_SPRITE)) | |
| 4337 | { | |
| 4338 | psx_gpu->render_state = render_state; | |
| 4339 | flush_render_block_buffer(psx_gpu); | |
| 3867c6ef | 4340 | #ifdef PROFILE |
| 75e28f62 | 4341 | state_changes++; |
| 3867c6ef | 4342 | #endif |
| 75e28f62 E |
4343 | } |
| 4344 | ||
| 4345 | psx_gpu->primitive_type = PRIMITIVE_TYPE_SPRITE; | |
| 4346 | ||
| 4347 | color &= 0xFFFFFF; | |
| 4348 | ||
| 4349 | if(psx_gpu->triangle_color != color) | |
| 4350 | { | |
| 4351 | flush_render_block_buffer(psx_gpu); | |
| 4352 | psx_gpu->triangle_color = color; | |
| 4353 | } | |
| 4354 | ||
| 4355 | if(color == 0x808080) | |
| 4356 | render_state |= RENDER_FLAGS_MODULATE_TEXELS; | |
| 4357 | ||
| 4358 | render_block_handler_struct *render_block_handler = | |
| 4359 | &(render_sprite_block_handlers[render_state]); | |
| 4360 | psx_gpu->render_block_handler = render_block_handler; | |
| 4361 | ||
| 4362 | ((setup_sprite_function_type *)render_block_handler->setup_blocks) | |
| 1cec4719 | 4363 | (psx_gpu, x, y, u, v, *width, *height, color); |
| 75e28f62 E |
4364 | } |
| 4365 | ||
| 4366 | #define draw_pixel_line_mask_evaluate_yes() \ | |
| 2177c1ea | 4367 | if((*vram_ptr & 0x8000) == 0) \ |
| 75e28f62 E |
4368 | |
| 4369 | #define draw_pixel_line_mask_evaluate_no() \ | |
| 4370 | ||
| 4371 | ||
| 4372 | #define draw_pixel_line_shaded() \ | |
| 4373 | { \ | |
| 4374 | color_r = fixed_to_int(current_r); \ | |
| 4375 | color_g = fixed_to_int(current_g); \ | |
| 4376 | color_b = fixed_to_int(current_b); \ | |
| 4377 | \ | |
| 4378 | current_r += gradient_r; \ | |
| 4379 | current_g += gradient_g; \ | |
| 4380 | current_b += gradient_b; \ | |
| 4381 | } \ | |
| 4382 | ||
| 4383 | #define draw_pixel_line_unshaded() \ | |
| 4384 | { \ | |
| 4385 | color_r = color & 0xFF; \ | |
| 4386 | color_g = (color >> 8) & 0xFF; \ | |
| 4387 | color_b = (color >> 16) & 0xFF; \ | |
| 4388 | } \ | |
| 4389 | ||
| 4390 | ||
| 4391 | #define draw_pixel_line_dithered(_x, _y) \ | |
| 4392 | { \ | |
| 4393 | u32 dither_xor = _x ^ _y; \ | |
| 4394 | s32 dither_offset = (dither_xor >> 1) & 0x1; \ | |
| 4395 | dither_offset |= (_y & 0x1) << 1; \ | |
| 4396 | dither_offset |= (dither_xor & 0x1) << 2; \ | |
| 4397 | dither_offset -= 4; \ | |
| 4398 | \ | |
| 4399 | color_r += dither_offset; \ | |
| 4400 | color_g += dither_offset; \ | |
| 4401 | color_b += dither_offset; \ | |
| 4402 | \ | |
| 4403 | if(color_r < 0) \ | |
| 4404 | color_r = 0; \ | |
| 4405 | \ | |
| 4406 | if(color_g < 0) \ | |
| 4407 | color_g = 0; \ | |
| 4408 | \ | |
| 4409 | if(color_b < 0) \ | |
| 4410 | color_b = 0; \ | |
| 4411 | \ | |
| 4412 | if(color_r > 255) \ | |
| 4413 | color_r = 255; \ | |
| 4414 | \ | |
| 4415 | if(color_g > 255) \ | |
| 4416 | color_g = 255; \ | |
| 4417 | \ | |
| 4418 | if(color_b > 255) \ | |
| 4419 | color_b = 255; \ | |
| 4420 | } \ | |
| 4421 | ||
| 4422 | #define draw_pixel_line_undithered(_x, _y) \ | |
| 4423 | ||
| 4424 | ||
| 4425 | #define draw_pixel_line_average() \ | |
| 4426 | color_r = (color_r + fb_r) / 2; \ | |
| 4427 | color_g = (color_g + fb_g) / 2; \ | |
| 4428 | color_b = (color_b + fb_b) / 2 \ | |
| 4429 | ||
| 4430 | #define draw_pixel_line_add() \ | |
| 4431 | color_r += fb_r; \ | |
| 4432 | color_g += fb_g; \ | |
| 4433 | color_b += fb_b; \ | |
| 4434 | \ | |
| 4435 | if(color_r > 31) \ | |
| 4436 | color_r = 31; \ | |
| 4437 | \ | |
| 4438 | if(color_g > 31) \ | |
| 4439 | color_g = 31; \ | |
| 4440 | \ | |
| 4441 | if(color_b > 31) \ | |
| 4442 | color_b = 31 \ | |
| 4443 | \ | |
| 4444 | ||
| 4445 | #define draw_pixel_line_subtract() \ | |
| 4446 | color_r = fb_r - color_r; \ | |
| 4447 | color_g = fb_g - color_g; \ | |
| 4448 | color_b = fb_b - color_b; \ | |
| 4449 | \ | |
| 4450 | if(color_r < 0) \ | |
| 4451 | color_r = 0; \ | |
| 4452 | \ | |
| 4453 | if(color_g < 0) \ | |
| 4454 | color_g = 0; \ | |
| 4455 | \ | |
| 4456 | if(color_b < 0) \ | |
| 4457 | color_b = 0 \ | |
| 4458 | ||
| 4459 | #define draw_pixel_line_add_fourth() \ | |
| 4460 | color_r = fb_r + (color_r / 4); \ | |
| 4461 | color_g = fb_g + (color_g / 4); \ | |
| 4462 | color_b = fb_b + (color_b / 4); \ | |
| 4463 | \ | |
| 4464 | if(color_r > 31) \ | |
| 4465 | color_r = 31; \ | |
| 4466 | \ | |
| 4467 | if(color_g > 31) \ | |
| 4468 | color_g = 31; \ | |
| 4469 | \ | |
| 4470 | if(color_b > 31) \ | |
| 4471 | color_b = 31 \ | |
| 4472 | ||
| 4473 | ||
| 4474 | #define draw_pixel_line_blended(blend_mode) \ | |
| 4475 | s32 fb_pixel = *vram_ptr; \ | |
| 4476 | s32 fb_r = fb_pixel & 0x1F; \ | |
| 4477 | s32 fb_g = (fb_pixel >> 5) & 0x1F; \ | |
| 4478 | s32 fb_b = (fb_pixel >> 10) & 0x1F; \ | |
| 4479 | \ | |
| 4480 | draw_pixel_line_##blend_mode() \ | |
| 4481 | ||
| 4482 | #define draw_pixel_line_unblended(blend_mode) \ | |
| 4483 | ||
| 4484 | ||
| 4485 | #define draw_pixel_line(_x, _y, shading, blending, dithering, mask_evaluate, \ | |
| 4486 | blend_mode) \ | |
| 4487 | if((_x >= psx_gpu->viewport_start_x) && (_y >= psx_gpu->viewport_start_y) && \ | |
| 4488 | (_x <= psx_gpu->viewport_end_x) && (_y <= psx_gpu->viewport_end_y)) \ | |
| 4489 | { \ | |
| 4490 | draw_pixel_line_mask_evaluate_##mask_evaluate() \ | |
| 4491 | { \ | |
| 4492 | draw_pixel_line_##shading(); \ | |
| 4493 | draw_pixel_line_##dithering(_x, _y); \ | |
| 4494 | \ | |
| 4495 | color_r >>= 3; \ | |
| 4496 | color_g >>= 3; \ | |
| 4497 | color_b >>= 3; \ | |
| 4498 | \ | |
| 4499 | draw_pixel_line_##blending(blend_mode); \ | |
| 4500 | \ | |
| 4501 | *vram_ptr = color_r | (color_g << 5) | (color_b << 10) | \ | |
| 4502 | psx_gpu->mask_msb; \ | |
| 4503 | } \ | |
| 4504 | } \ | |
| 4505 | ||
| 4506 | #define update_increment(value) \ | |
| 4507 | value++ \ | |
| 4508 | ||
| 4509 | #define update_decrement(value) \ | |
| 4510 | value-- \ | |
| 4511 | ||
| 4512 | #define update_vram_row_increment(value) \ | |
| 4513 | vram_ptr += 1024 \ | |
| 4514 | ||
| 4515 | #define update_vram_row_decrement(value) \ | |
| 4516 | vram_ptr -= 1024 \ | |
| 4517 | ||
| 4518 | #define compare_increment(a, b) \ | |
| 4519 | (a <= b) \ | |
| 4520 | ||
| 4521 | #define compare_decrement(a, b) \ | |
| 4522 | (a >= b) \ | |
| 4523 | ||
| 4524 | #define set_line_gradients(minor) \ | |
| 4525 | { \ | |
| 4526 | s32 gradient_divisor = delta_##minor; \ | |
| e86b6fec E |
4527 | if(gradient_divisor != 0) \ |
| 4528 | { \ | |
| 4529 | gradient_r = int_to_fixed(vertex_b->r - vertex_a->r) / gradient_divisor; \ | |
| 4530 | gradient_g = int_to_fixed(vertex_b->g - vertex_a->g) / gradient_divisor; \ | |
| 4531 | gradient_b = int_to_fixed(vertex_b->b - vertex_a->b) / gradient_divisor; \ | |
| 4532 | } \ | |
| 4533 | else \ | |
| 4534 | { \ | |
| 4535 | gradient_r = 0; \ | |
| 4536 | gradient_g = 0; \ | |
| 4537 | gradient_b = 0; \ | |
| 4538 | } \ | |
| 75e28f62 E |
4539 | current_r = fixed_center(vertex_a->r); \ |
| 4540 | current_g = fixed_center(vertex_a->g); \ | |
| 4541 | current_b = fixed_center(vertex_a->b); \ | |
| 4542 | } | |
| 4543 | ||
| 4544 | #define draw_line_span_horizontal(direction, shading, blending, dithering, \ | |
| 4545 | mask_evaluate, blend_mode) \ | |
| 4546 | do \ | |
| 4547 | { \ | |
| 4548 | error_step = delta_y * 2; \ | |
| 4549 | error_wrap = delta_x * 2; \ | |
| 4550 | error = delta_x; \ | |
| 4551 | \ | |
| 4552 | current_y = y_a; \ | |
| 4553 | set_line_gradients(x); \ | |
| 4554 | \ | |
| 4555 | for(current_x = x_a; current_x <= x_b; current_x++) \ | |
| 4556 | { \ | |
| 4557 | draw_pixel_line(current_x, current_y, shading, blending, dithering, \ | |
| 4558 | mask_evaluate, blend_mode); \ | |
| 4559 | error += error_step; \ | |
| 4560 | vram_ptr++; \ | |
| 4561 | \ | |
| 4562 | if(error >= error_wrap) \ | |
| 4563 | { \ | |
| 4564 | update_##direction(current_y); \ | |
| 4565 | update_vram_row_##direction(); \ | |
| 4566 | error -= error_wrap; \ | |
| 4567 | } \ | |
| 4568 | } \ | |
| 4569 | } while(0) \ | |
| 4570 | ||
| 4571 | #define draw_line_span_vertical(direction, shading, blending, dithering, \ | |
| 4572 | mask_evaluate, blend_mode) \ | |
| 4573 | do \ | |
| 4574 | { \ | |
| 4575 | error_step = delta_x * 2; \ | |
| 4576 | error_wrap = delta_y * 2; \ | |
| 4577 | error = delta_y; \ | |
| 4578 | \ | |
| 4579 | current_x = x_a; \ | |
| 4580 | set_line_gradients(y); \ | |
| 4581 | \ | |
| 4582 | for(current_y = y_a; compare_##direction(current_y, y_b); \ | |
| 4583 | update_##direction(current_y)) \ | |
| 4584 | { \ | |
| 4585 | draw_pixel_line(current_x, current_y, shading, blending, dithering, \ | |
| 4586 | mask_evaluate, blend_mode); \ | |
| 4587 | error += error_step; \ | |
| 4588 | update_vram_row_##direction(); \ | |
| 4589 | \ | |
| 4590 | if(error > error_wrap) \ | |
| 4591 | { \ | |
| 4592 | vram_ptr++; \ | |
| 4593 | current_x++; \ | |
| 4594 | error -= error_wrap; \ | |
| 4595 | } \ | |
| 4596 | } \ | |
| 4597 | } while(0) \ | |
| 4598 | ||
| 4599 | ||
| 4600 | #define render_line_body(shading, blending, dithering, mask_evaluate, \ | |
| 4601 | blend_mode) \ | |
| 4602 | if(delta_y < 0) \ | |
| 4603 | { \ | |
| 4604 | delta_y *= -1; \ | |
| 4605 | \ | |
| 75e28f62 E |
4606 | if(delta_x > delta_y) \ |
| 4607 | { \ | |
| 4608 | draw_line_span_horizontal(decrement, shading, blending, dithering, \ | |
| 4609 | mask_evaluate, blend_mode); \ | |
| 4610 | } \ | |
| 4611 | else \ | |
| 4612 | { \ | |
| 4613 | draw_line_span_vertical(decrement, shading, blending, dithering, \ | |
| 4614 | mask_evaluate, blend_mode); \ | |
| 4615 | } \ | |
| 4616 | } \ | |
| 4617 | else \ | |
| 4618 | { \ | |
| 75e28f62 E |
4619 | if(delta_x > delta_y) \ |
| 4620 | { \ | |
| 4621 | draw_line_span_horizontal(increment, shading, blending, dithering, \ | |
| 4622 | mask_evaluate, blend_mode); \ | |
| 4623 | } \ | |
| 4624 | else \ | |
| 4625 | { \ | |
| 4626 | draw_line_span_vertical(increment, shading, blending, dithering, \ | |
| 4627 | mask_evaluate, blend_mode); \ | |
| 4628 | } \ | |
| 4629 | } \ | |
| 4630 | ||
| 4631 | ||
| 4632 | void render_line(psx_gpu_struct *psx_gpu, vertex_struct *vertexes, u32 flags, | |
| 3b3dee71 | 4633 | u32 color, int double_resolution) |
| 75e28f62 E |
4634 | { |
| 4635 | s32 color_r, color_g, color_b; | |
| 4636 | u32 triangle_winding = 0; | |
| 4637 | ||
| 4638 | fixed_type gradient_r = 0; | |
| 4639 | fixed_type gradient_g = 0; | |
| 4640 | fixed_type gradient_b = 0; | |
| 4641 | fixed_type current_r = 0; | |
| 4642 | fixed_type current_g = 0; | |
| 4643 | fixed_type current_b = 0; | |
| 4644 | ||
| 4645 | s32 y_a, y_b; | |
| 4646 | s32 x_a, x_b; | |
| 4647 | ||
| 4648 | s32 delta_x, delta_y; | |
| 4649 | ||
| 4650 | s32 current_x; | |
| 4651 | s32 current_y; | |
| 4652 | ||
| 4653 | u32 error_step; | |
| 4654 | u32 error; | |
| 4655 | u32 error_wrap; | |
| 4656 | ||
| 4657 | u16 *vram_ptr; | |
| 4658 | ||
| 4659 | flush_render_block_buffer(psx_gpu); | |
| 4660 | psx_gpu->primitive_type = PRIMITIVE_TYPE_LINE; | |
| 4661 | ||
| 4662 | vertex_struct *vertex_a = &(vertexes[0]); | |
| 4663 | vertex_struct *vertex_b = &(vertexes[1]); | |
| 4664 | ||
| 4665 | u32 control_mask; | |
| 4666 | ||
| 3867c6ef | 4667 | #ifdef PROFILE |
| 75e28f62 | 4668 | lines++; |
| 3867c6ef | 4669 | #endif |
| 75e28f62 E |
4670 | |
| 4671 | if(vertex_a->x >= vertex_b->x) | |
| 4672 | { | |
| 4673 | vertex_swap(vertex_a, vertex_b); | |
| c57af5e6 | 4674 | (void)triangle_winding; |
| 75e28f62 E |
4675 | } |
| 4676 | ||
| 4677 | x_a = vertex_a->x; | |
| 4678 | x_b = vertex_b->x; | |
| 4679 | ||
| 4680 | y_a = vertex_a->y; | |
| 4681 | y_b = vertex_b->y; | |
| 4682 | ||
| 4683 | delta_x = x_b - x_a; | |
| 4684 | delta_y = y_b - y_a; | |
| 4685 | ||
| 3b3dee71 | 4686 | if(delta_x >= 1024 || delta_y >= 512 || delta_y <= -512) |
| 75e28f62 E |
4687 | return; |
| 4688 | ||
| 3b3dee71 | 4689 | if(double_resolution) |
| 4690 | { | |
| 4691 | x_a *= 2; | |
| 4692 | x_b *= 2; | |
| 4693 | y_a *= 2; | |
| 4694 | y_b *= 2; | |
| 4695 | delta_x *= 2; | |
| 4696 | delta_y *= 2; | |
| 4697 | } | |
| 4698 | ||
| 75e28f62 E |
4699 | flags &= ~RENDER_FLAGS_TEXTURE_MAP; |
| 4700 | ||
| c1817bd9 | 4701 | vram_ptr = psx_gpu->vram_out_ptr + (y_a * 1024) + x_a; |
| 75e28f62 E |
4702 | |
| 4703 | control_mask = 0x0; | |
| 4704 | ||
| 4705 | if(flags & RENDER_FLAGS_SHADE) | |
| 4706 | control_mask |= 0x1; | |
| 4707 | ||
| 4708 | if(flags & RENDER_FLAGS_BLEND) | |
| 4709 | { | |
| 4710 | control_mask |= 0x2; | |
| 4711 | control_mask |= ((psx_gpu->render_state_base >> 6) & 0x3) << 4; | |
| 4712 | } | |
| 4713 | ||
| 4714 | if(psx_gpu->render_state_base & RENDER_STATE_DITHER) | |
| 4715 | control_mask |= 0x4; | |
| 4716 | ||
| 4717 | if(psx_gpu->render_state_base & RENDER_STATE_MASK_EVALUATE) | |
| 4718 | control_mask |= 0x8; | |
| 4719 | ||
| 4720 | switch(control_mask) | |
| 4721 | { | |
| 4722 | case 0x0: | |
| 4723 | render_line_body(unshaded, unblended, undithered, no, none); | |
| 4724 | break; | |
| 4725 | ||
| 4726 | case 0x1: | |
| 4727 | render_line_body(shaded, unblended, undithered, no, none); | |
| 4728 | break; | |
| 4729 | ||
| 4730 | case 0x2: | |
| 4731 | render_line_body(unshaded, blended, undithered, no, average); | |
| 4732 | break; | |
| 4733 | ||
| 4734 | case 0x3: | |
| 4735 | render_line_body(shaded, blended, undithered, no, average); | |
| 4736 | break; | |
| 4737 | ||
| 4738 | case 0x4: | |
| 4739 | render_line_body(unshaded, unblended, dithered, no, none); | |
| 4740 | break; | |
| 4741 | ||
| 4742 | case 0x5: | |
| 4743 | render_line_body(shaded, unblended, dithered, no, none); | |
| 4744 | break; | |
| 4745 | ||
| 4746 | case 0x6: | |
| 4747 | render_line_body(unshaded, blended, dithered, no, average); | |
| 4748 | break; | |
| 4749 | ||
| 4750 | case 0x7: | |
| 4751 | render_line_body(shaded, blended, dithered, no, average); | |
| 4752 | break; | |
| 4753 | ||
| 4754 | case 0x8: | |
| 4755 | render_line_body(unshaded, unblended, undithered, yes, none); | |
| 4756 | break; | |
| 4757 | ||
| 4758 | case 0x9: | |
| 4759 | render_line_body(shaded, unblended, undithered, yes, none); | |
| 4760 | break; | |
| 4761 | ||
| 4762 | case 0xA: | |
| 4763 | render_line_body(unshaded, blended, undithered, yes, average); | |
| 4764 | break; | |
| 4765 | ||
| 4766 | case 0xB: | |
| 4767 | render_line_body(shaded, blended, undithered, yes, average); | |
| 4768 | break; | |
| 4769 | ||
| 4770 | case 0xC: | |
| 4771 | render_line_body(unshaded, unblended, dithered, yes, none); | |
| 4772 | break; | |
| 4773 | ||
| 4774 | case 0xD: | |
| 4775 | render_line_body(shaded, unblended, dithered, yes, none); | |
| 4776 | break; | |
| 4777 | ||
| 4778 | case 0xE: | |
| 4779 | render_line_body(unshaded, blended, dithered, yes, average); | |
| 4780 | break; | |
| 4781 | ||
| 4782 | case 0xF: | |
| 4783 | render_line_body(shaded, blended, dithered, yes, average); | |
| 4784 | break; | |
| 4785 | ||
| 4786 | case 0x12: | |
| 4787 | render_line_body(unshaded, blended, undithered, no, add); | |
| 4788 | break; | |
| 4789 | ||
| 4790 | case 0x13: | |
| 4791 | render_line_body(shaded, blended, undithered, no, add); | |
| 4792 | break; | |
| 4793 | ||
| 4794 | case 0x16: | |
| 4795 | render_line_body(unshaded, blended, dithered, no, add); | |
| 4796 | break; | |
| 4797 | ||
| 4798 | case 0x17: | |
| 4799 | render_line_body(shaded, blended, dithered, no, add); | |
| 4800 | break; | |
| 4801 | ||
| 4802 | case 0x1A: | |
| 4803 | render_line_body(unshaded, blended, undithered, yes, add); | |
| 4804 | break; | |
| 4805 | ||
| 4806 | case 0x1B: | |
| 4807 | render_line_body(shaded, blended, undithered, yes, add); | |
| 4808 | break; | |
| 4809 | ||
| 4810 | case 0x1E: | |
| 4811 | render_line_body(unshaded, blended, dithered, yes, add); | |
| 4812 | break; | |
| 4813 | ||
| 4814 | case 0x1F: | |
| 4815 | render_line_body(shaded, blended, dithered, yes, add); | |
| 4816 | break; | |
| 4817 | ||
| 4818 | case 0x22: | |
| 4819 | render_line_body(unshaded, blended, undithered, no, subtract); | |
| 4820 | break; | |
| 4821 | ||
| 4822 | case 0x23: | |
| 4823 | render_line_body(shaded, blended, undithered, no, subtract); | |
| 4824 | break; | |
| 4825 | ||
| 4826 | case 0x26: | |
| 4827 | render_line_body(unshaded, blended, dithered, no, subtract); | |
| 4828 | break; | |
| 4829 | ||
| 4830 | case 0x27: | |
| 4831 | render_line_body(shaded, blended, dithered, no, subtract); | |
| 4832 | break; | |
| 4833 | ||
| 4834 | case 0x2A: | |
| 4835 | render_line_body(unshaded, blended, undithered, yes, subtract); | |
| 4836 | break; | |
| 4837 | ||
| 4838 | case 0x2B: | |
| 4839 | render_line_body(shaded, blended, undithered, yes, subtract); | |
| 4840 | break; | |
| 4841 | ||
| 4842 | case 0x2E: | |
| 4843 | render_line_body(unshaded, blended, dithered, yes, subtract); | |
| 4844 | break; | |
| 4845 | ||
| 4846 | case 0x2F: | |
| 4847 | render_line_body(shaded, blended, dithered, yes, subtract); | |
| 4848 | break; | |
| 4849 | ||
| 4850 | case 0x32: | |
| 4851 | render_line_body(unshaded, blended, undithered, no, add_fourth); | |
| 4852 | break; | |
| 4853 | ||
| 4854 | case 0x33: | |
| 4855 | render_line_body(shaded, blended, undithered, no, add_fourth); | |
| 4856 | break; | |
| 4857 | ||
| 4858 | case 0x36: | |
| 4859 | render_line_body(unshaded, blended, dithered, no, add_fourth); | |
| 4860 | break; | |
| 4861 | ||
| 4862 | case 0x37: | |
| 4863 | render_line_body(shaded, blended, dithered, no, add_fourth); | |
| 4864 | break; | |
| 4865 | ||
| 4866 | case 0x3A: | |
| 4867 | render_line_body(unshaded, blended, undithered, yes, add_fourth); | |
| 4868 | break; | |
| 4869 | ||
| 4870 | case 0x3B: | |
| 4871 | render_line_body(shaded, blended, undithered, yes, add_fourth); | |
| 4872 | break; | |
| 4873 | ||
| 4874 | case 0x3E: | |
| 4875 | render_line_body(unshaded, blended, dithered, yes, add_fourth); | |
| 4876 | break; | |
| 4877 | ||
| 4878 | case 0x3F: | |
| 4879 | render_line_body(shaded, blended, dithered, yes, add_fourth); | |
| 4880 | break; | |
| 4881 | } | |
| 4882 | } | |
| 4883 | ||
| 4884 | ||
| 4885 | void render_block_fill(psx_gpu_struct *psx_gpu, u32 color, u32 x, u32 y, | |
| 4886 | u32 width, u32 height) | |
| 4887 | { | |
| 1f88961f E |
4888 | if((width == 0) || (height == 0)) |
| 4889 | return; | |
| 4890 | ||
| 75e28f62 E |
4891 | invalidate_texture_cache_region(psx_gpu, x, y, x + width - 1, y + height - 1); |
| 4892 | ||
| 75e28f62 E |
4893 | u32 r = color & 0xFF; |
| 4894 | u32 g = (color >> 8) & 0xFF; | |
| 4895 | u32 b = (color >> 16) & 0xFF; | |
| 03eb3b69 | 4896 | u32 color_16bpp = (r >> 3) | ((g >> 3) << 5) | ((b >> 3) << 10); |
| 6c4a10c4 E |
4897 | u32 color_32bpp = color_16bpp | (color_16bpp << 16); |
| 4898 | ||
| c1817bd9 | 4899 | u32 *vram_ptr = (u32 *)(psx_gpu->vram_out_ptr + x + (y * 1024)); |
| 6c4a10c4 E |
4900 | |
| 4901 | u32 pitch = 512 - (width / 2); | |
| 4902 | u32 num_width; | |
| 75e28f62 | 4903 | |
| f1359c57 | 4904 | if(psx_gpu->render_mode & RENDER_INTERLACE_ENABLED) |
| 6c4a10c4 E |
4905 | { |
| 4906 | pitch += 512; | |
| 4907 | height /= 2; | |
| 4908 | ||
| f1359c57 | 4909 | if(psx_gpu->render_mode & RENDER_INTERLACE_ODD) |
| 6c4a10c4 E |
4910 | vram_ptr += 512; |
| 4911 | } | |
| 75e28f62 | 4912 | |
| 6c4a10c4 | 4913 | while(height) |
| 75e28f62 | 4914 | { |
| 6c4a10c4 E |
4915 | num_width = width; |
| 4916 | while(num_width) | |
| 75e28f62 | 4917 | { |
| 6c4a10c4 E |
4918 | vram_ptr[0] = color_32bpp; |
| 4919 | vram_ptr[1] = color_32bpp; | |
| 4920 | vram_ptr[2] = color_32bpp; | |
| 4921 | vram_ptr[3] = color_32bpp; | |
| 4922 | vram_ptr[4] = color_32bpp; | |
| 4923 | vram_ptr[5] = color_32bpp; | |
| 4924 | vram_ptr[6] = color_32bpp; | |
| 4925 | vram_ptr[7] = color_32bpp; | |
| 4926 | ||
| 4927 | vram_ptr += 8; | |
| 4928 | num_width -= 16; | |
| 75e28f62 E |
4929 | } |
| 4930 | ||
| 6c4a10c4 E |
4931 | vram_ptr += pitch; |
| 4932 | height--; | |
| 75e28f62 | 4933 | } |
| 75e28f62 E |
4934 | } |
| 4935 | ||
| f1359c57 | 4936 | void render_block_fill_enh(psx_gpu_struct *psx_gpu, u32 color, u32 x, u32 y, |
| 4937 | u32 width, u32 height) | |
| 4938 | { | |
| 4939 | if((width == 0) || (height == 0)) | |
| 4940 | return; | |
| 4941 | ||
| e929dec5 | 4942 | if(width > 1024) |
| 4943 | width = 1024; | |
| 4944 | ||
| f1359c57 | 4945 | u32 r = color & 0xFF; |
| 4946 | u32 g = (color >> 8) & 0xFF; | |
| 4947 | u32 b = (color >> 16) & 0xFF; | |
| 03eb3b69 | 4948 | u32 color_16bpp = (r >> 3) | ((g >> 3) << 5) | ((b >> 3) << 10); |
| f1359c57 | 4949 | u32 color_32bpp = color_16bpp | (color_16bpp << 16); |
| 4950 | ||
| e929dec5 | 4951 | u32 *vram_ptr = (u32 *)(psx_gpu->vram_out_ptr + x + (y * 1024)); |
| f1359c57 | 4952 | |
| e929dec5 | 4953 | u32 pitch = 1024 / 2 - (width / 2); |
| f1359c57 | 4954 | u32 num_width; |
| 4955 | ||
| 4956 | while(height) | |
| 4957 | { | |
| 4958 | num_width = width; | |
| 4959 | while(num_width) | |
| 4960 | { | |
| 4961 | vram_ptr[0] = color_32bpp; | |
| 4962 | vram_ptr[1] = color_32bpp; | |
| 4963 | vram_ptr[2] = color_32bpp; | |
| 4964 | vram_ptr[3] = color_32bpp; | |
| 4965 | vram_ptr[4] = color_32bpp; | |
| 4966 | vram_ptr[5] = color_32bpp; | |
| 4967 | vram_ptr[6] = color_32bpp; | |
| 4968 | vram_ptr[7] = color_32bpp; | |
| 4969 | ||
| 4970 | vram_ptr += 8; | |
| 4971 | num_width -= 16; | |
| 4972 | } | |
| 4973 | ||
| 4974 | vram_ptr += pitch; | |
| 4975 | height--; | |
| 4976 | } | |
| 4977 | } | |
| 4978 | ||
| 72583812 | 4979 | #ifndef PCSX |
| 75e28f62 E |
4980 | void render_block_copy(psx_gpu_struct *psx_gpu, u16 *source, u32 x, u32 y, |
| 4981 | u32 width, u32 height, u32 pitch) | |
| 4982 | { | |
| 4983 | u16 *vram_ptr = psx_gpu->vram_ptr + x + (y * 1024); | |
| 4984 | u32 draw_x, draw_y; | |
| 87c45ad1 | 4985 | u32 mask_msb = psx_gpu->mask_msb; |
| 75e28f62 | 4986 | |
| 1f88961f E |
4987 | if((width == 0) || (height == 0)) |
| 4988 | return; | |
| 4989 | ||
| cec398c0 | 4990 | flush_render_block_buffer(psx_gpu); |
| 75e28f62 E |
4991 | invalidate_texture_cache_region(psx_gpu, x, y, x + width - 1, y + height - 1); |
| 4992 | ||
| 75e28f62 E |
4993 | for(draw_y = 0; draw_y < height; draw_y++) |
| 4994 | { | |
| 4995 | for(draw_x = 0; draw_x < width; draw_x++) | |
| 4996 | { | |
| 24b9bacc | 4997 | vram_ptr[draw_x] = source[draw_x] | mask_msb; |
| 75e28f62 E |
4998 | } |
| 4999 | ||
| 5000 | source += pitch; | |
| 5001 | vram_ptr += 1024; | |
| 5002 | } | |
| 5003 | } | |
| 5004 | ||
| 5005 | void render_block_move(psx_gpu_struct *psx_gpu, u32 source_x, u32 source_y, | |
| 5006 | u32 dest_x, u32 dest_y, u32 width, u32 height) | |
| 5007 | { | |
| 5008 | render_block_copy(psx_gpu, psx_gpu->vram_ptr + source_x + (source_y * 1024), | |
| 5009 | dest_x, dest_y, width, height, 1024); | |
| 5010 | } | |
| 72583812 | 5011 | #endif |
| 75e28f62 E |
5012 | |
| 5013 | void initialize_reciprocal_table(void) | |
| 5014 | { | |
| 5015 | u32 height; | |
| 5016 | u32 height_normalized; | |
| 5017 | u32 height_reciprocal; | |
| 5018 | s32 shift; | |
| 5019 | ||
| c1817bd9 | 5020 | for(height = 1; height < sizeof(reciprocal_table) |
| 5021 | / sizeof(reciprocal_table[0]); height++) | |
| 75e28f62 E |
5022 | { |
| 5023 | shift = __builtin_clz(height); | |
| 5024 | height_normalized = height << shift; | |
| c111e8f8 | 5025 | height_reciprocal = ((1ULL << 51) + (height_normalized - 1)) / |
| 75e28f62 E |
5026 | height_normalized; |
| 5027 | ||
| c111e8f8 | 5028 | shift = 32 - (51 - shift); |
| 75e28f62 | 5029 | |
| 7d5140f5 | 5030 | reciprocal_table[height] = (height_reciprocal << 10) | shift; |
| 75e28f62 E |
5031 | } |
| 5032 | } | |
| 5033 | ||
| 5034 | ||
| 5035 | #define dither_table_row(a, b, c, d) \ | |
| f707f14b | 5036 | ((a & 0xFF) | ((b & 0xFF) << 8) | ((c & 0xFF) << 16) | ((u32)(d & 0xFF) << 24)) \ |
| 75e28f62 | 5037 | |
| e8c0e0bb | 5038 | void initialize_psx_gpu(psx_gpu_struct *psx_gpu, u16 *vram) |
| 75e28f62 E |
5039 | { |
| 5040 | vec_8x16u test_mask = | |
| 5041 | { { { 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80 } } }; | |
| 5042 | ||
| 5043 | psx_gpu->test_mask = test_mask; | |
| 5044 | ||
| 75e28f62 E |
5045 | psx_gpu->dirty_textures_4bpp_mask = 0xFFFFFFFF; |
| 5046 | psx_gpu->dirty_textures_8bpp_mask = 0xFFFFFFFF; | |
| 5047 | psx_gpu->dirty_textures_8bpp_alternate_mask = 0xFFFFFFFF; | |
| 5048 | psx_gpu->viewport_mask = 0; | |
| 5049 | psx_gpu->current_texture_page = 0; | |
| 5050 | psx_gpu->current_texture_mask = 0; | |
| 5051 | psx_gpu->last_8bpp_texture_page = 0; | |
| 5052 | ||
| 5053 | psx_gpu->clut_settings = 0; | |
| 5054 | psx_gpu->texture_settings = 0; | |
| 5055 | psx_gpu->render_state = 0; | |
| 5056 | psx_gpu->render_state_base = 0; | |
| 5057 | psx_gpu->num_blocks = 0; | |
| c6063f89 | 5058 | psx_gpu->uvrgb_phase = 0x8000; |
| 75e28f62 | 5059 | |
| e8c0e0bb | 5060 | psx_gpu->vram_ptr = vram; |
| c1817bd9 | 5061 | psx_gpu->vram_out_ptr = vram; |
| 75e28f62 | 5062 | |
| 3867c6ef | 5063 | psx_gpu->texture_page_base = psx_gpu->vram_ptr; |
| 75e28f62 E |
5064 | psx_gpu->texture_page_ptr = psx_gpu->vram_ptr; |
| 5065 | psx_gpu->clut_ptr = psx_gpu->vram_ptr; | |
| 5066 | ||
| 5fe1a2b1 | 5067 | psx_gpu->viewport_start_x = psx_gpu->viewport_start_y = 0; |
| 5068 | psx_gpu->viewport_end_x = psx_gpu->viewport_end_y = 0; | |
| 75e28f62 E |
5069 | psx_gpu->mask_msb = 0; |
| 5070 | ||
| f9248bbf E |
5071 | psx_gpu->texture_window_x = 0; |
| 5072 | psx_gpu->texture_window_y = 0; | |
| 5073 | psx_gpu->texture_mask_width = 0xFF; | |
| 5074 | psx_gpu->texture_mask_height = 0xFF; | |
| 5075 | ||
| f1359c57 | 5076 | psx_gpu->render_mode = 0; |
| 69b09c0d | 5077 | |
| 75e28f62 E |
5078 | memset(psx_gpu->vram_ptr, 0, sizeof(u16) * 1024 * 512); |
| 5079 | ||
| 5080 | initialize_reciprocal_table(); | |
| ed0fd81d | 5081 | psx_gpu->reciprocal_table_ptr = reciprocal_table; |
| 75e28f62 E |
5082 | |
| 5083 | // 00 01 10 11 | |
| 5084 | // 00 0 4 1 5 | |
| 5085 | // 01 6 2 7 3 | |
| 5086 | // 10 1 5 0 4 | |
| 5087 | // 11 7 3 6 2 | |
| 5088 | // (minus ones(4) * 4) | |
| 5089 | ||
| 5090 | // d0: (1 3 5 7): x1 ^ y1 | |
| 5091 | // d1: (2 3 6 7): y0 | |
| 5092 | // d2: (4 5 6 7): x0 ^ y0 | |
| 5093 | ||
| 75e28f62 E |
5094 | psx_gpu->dither_table[0] = dither_table_row(-4, 0, -3, 1); |
| 5095 | psx_gpu->dither_table[1] = dither_table_row(2, -2, 3, -1); | |
| 5096 | psx_gpu->dither_table[2] = dither_table_row(-3, 1, -4, 0); | |
| 5097 | psx_gpu->dither_table[3] = dither_table_row(3, -1, 2, -2); | |
| df05223d | 5098 | psx_gpu->allow_dithering = 1; |
| 75e28f62 E |
5099 | |
| 5100 | psx_gpu->primitive_type = PRIMITIVE_TYPE_UNKNOWN; | |
| e929dec5 | 5101 | |
| 0b4038f8 | 5102 | psx_gpu->saved_hres = 256; |
| ee060c58 | 5103 | psx_gpu->hacks_active = 0; |
| 2da2fc76 | 5104 | |
| ee060c58 | 5105 | // check some offsets, asm relies on these |
| 5106 | psx_gpu->reserved_a[(offsetof(psx_gpu_struct, test_mask) == 0) - 1] = 0; | |
| 2da2fc76 | 5107 | psx_gpu->reserved_a[(offsetof(psx_gpu_struct, blocks) == psx_gpu_blocks_offset) - 1] = 0; |
| 75e28f62 E |
5108 | } |
| 5109 | ||
| 5110 | u64 get_us(void) | |
| 5111 | { | |
| 5112 | struct timeval tv; | |
| 5113 | gettimeofday(&tv, NULL); | |
| 5114 | ||
| 5115 | return (tv.tv_sec * 1000000ULL) + tv.tv_usec; | |
| 5116 | } | |
| 5117 | ||
| 37725e8c | 5118 | #if 0 //def NEON_BUILD |
| 75e28f62 E |
5119 | |
| 5120 | u32 get_counter() | |
| 5121 | { | |
| 5122 | u32 counter; | |
| 5123 | __asm__ volatile("mrc p15, 0, %0, c9, c13, 0" : "=r"(counter)); | |
| 5124 | ||
| 5125 | return counter; | |
| 5126 | } | |
| 5127 | ||
| 5128 | void init_counter(void) | |
| 5129 | { | |
| 5130 | u32 value; | |
| 5131 | asm volatile("mrc p15, 0, %0, c9, c12, 0" : "=r"(value)); | |
| 5132 | value |= 5; // master enable, ccnt reset | |
| 5133 | value &= ~8; // ccnt divider 0 | |
| 5134 | asm volatile("mcr p15, 0, %0, c9, c12, 0" :: "r"(value)); | |
| 5135 | // enable cycle counter | |
| 5136 | asm volatile("mcr p15, 0, %0, c9, c12, 1" :: "r"(1 << 31)); | |
| 5137 | } | |
| 5138 | ||
| 5139 | void triangle_benchmark(psx_gpu_struct *psx_gpu) | |
| 5140 | { | |
| 5141 | u32 i; | |
| 5142 | ||
| 5143 | u32 ticks; | |
| 5144 | u32 ticks_elapsed; | |
| 5145 | ||
| 5146 | const u32 iterations = 500000; | |
| 5147 | ||
| 5148 | psx_gpu->num_blocks = 64; | |
| 5149 | psx_gpu->clut_ptr = psx_gpu->vram_ptr; | |
| 5150 | ||
| 5151 | for(i = 0; i < 64; i++) | |
| 5152 | { | |
| 5153 | memset(&(psx_gpu->blocks[i].r), 0, 16); | |
| 5154 | } | |
| 5155 | ||
| 5156 | init_counter(); | |
| 5157 | ||
| 5158 | ticks = get_counter(); | |
| 5159 | ||
| 5160 | for(i = 0; i < iterations; i++) | |
| 5161 | { | |
| 5162 | texture_sprite_blocks_8bpp(psx_gpu); | |
| 5163 | } | |
| 5164 | ||
| 5165 | ticks_elapsed = get_counter() - ticks; | |
| 5166 | ||
| 5167 | printf("benchmark: %lf cycles\n", (double)ticks_elapsed / (iterations * 64)); | |
| 5168 | } | |
| 5169 | ||
| 5170 | #endif | |
| 5171 | ||
| fc6cef7d | 5172 | #include "psx_gpu_4x.c" |
| 47c15995 | 5173 | |
| 5174 | // vim:ts=2:sw=2:expandtab |