2 * Copyright (C) 2011 Gilead Kutnick "Exophase" <exophase@gmail.com>
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.
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.
21 #include "vector_ops.h"
23 #include "psx_gpu_simd.h"
26 u32 span_pixel_blocks = 0;
32 u32 sprites_16bpp = 0;
33 u32 sprite_blocks = 0;
34 u32 sprites_untextured = 0;
36 u32 trivial_rejects = 0;
40 u32 texel_blocks_4bpp = 0;
41 u32 texel_blocks_8bpp = 0;
42 u32 texel_blocks_16bpp = 0;
43 u32 texel_blocks_untextured = 0;
45 u32 render_buffer_flushes = 0;
46 u32 state_changes = 0;
47 u32 left_split_triangles = 0;
48 u32 flat_triangles = 0;
49 u32 clipped_triangles = 0;
50 u32 zero_block_spans = 0;
51 u32 texture_cache_loads = 0;
52 u32 false_modulated_blocks = 0;
54 /* double size for enhancement */
55 u32 reciprocal_table[512 * 2];
58 typedef s32 fixed_type;
60 #define EDGE_STEP_BITS 32
63 #define fixed_center(value) \
64 ((((fixed_type)(value)) << FIXED_BITS) + (1 << (FIXED_BITS - 1))) \
66 #define int_to_fixed(value) \
67 (((fixed_type)(value)) << FIXED_BITS) \
69 #define fixed_to_int(value) \
70 ((value) >> FIXED_BITS) \
72 #define fixed_to_double(value) \
73 ((value) / (double)(1 << FIXED_BITS)) \
75 #define double_to_fixed(value) \
76 (fixed_type)(((value) * (double)(1 << FIXED_BITS))) \
78 typedef void (setup_blocks_function_type)(psx_gpu_struct *psx_gpu);
79 typedef void (texture_blocks_function_type)(psx_gpu_struct *psx_gpu);
80 typedef void (shade_blocks_function_type)(psx_gpu_struct *psx_gpu);
81 typedef void (blend_blocks_function_type)(psx_gpu_struct *psx_gpu);
83 typedef void (setup_sprite_function_type)(psx_gpu_struct *psx_gpu, s32 x,
84 s32 y, s32 u, s32 v, s32 width, s32 height, u32 color);
86 struct render_block_handler_struct
89 texture_blocks_function_type *texture_blocks;
90 shade_blocks_function_type *shade_blocks;
91 blend_blocks_function_type *blend_blocks;
96 u32 fixed_reciprocal(u32 denominator, u32 *_shift)
98 u32 shift = __builtin_clz(denominator);
99 u32 denominator_normalized = denominator << shift;
101 double numerator = (1ULL << 62) + denominator_normalized;
104 double denominator_normalized_dp_b;
105 u64 denominator_normalized_dp_u64;
108 double reciprocal_dp;
110 u64 numerator_u64 = (denominator_normalized >> 10) |
111 ((u64)(62 + 1023) << 52);
112 *((u64 *)(&numerator_b)) = numerator_u64;
114 denominator_normalized_dp_u64 =
115 (u64)(denominator_normalized << 21) |
116 ((u64)((denominator_normalized >> 11) + ((1022 + 31) << 20)) << 32);
117 *((u64 *)(&denominator_normalized_dp_b)) = denominator_normalized_dp_u64;
119 // Implement with a DP divide
120 reciprocal_dp = numerator / denominator_normalized_dp_b;
121 reciprocal = reciprocal_dp;
123 if(reciprocal == 0x80000001)
124 reciprocal = 0x80000000;
126 *_shift = 62 - shift;
130 double reciprocal_estimate(double a)
135 q = (int)(a * 512.0);
136 /* a in units of 1/512 rounded down */
137 r = 1.0 / (((double)q + 0.5) / 512.0); /* reciprocal r */
138 s = (int)(256.0 * r + 0.5);
140 /* r in units of 1/256 rounded to nearest */
142 return (double)s / 256.0;
145 u32 reciprocal_estimate_u32(u32 value)
148 volatile double dp_value;
149 volatile u64 *dp_value_ptr = (volatile u64 *)&dp_value;
151 if((value >> 31) == 0)
154 dp_value_u64 = (0x3FEULL << (31 + 21)) | ((u64)(value & 0x7FFFFFFF) << 21);
156 *dp_value_ptr = dp_value_u64;
158 dp_value = reciprocal_estimate(dp_value);
159 dp_value_u64 = *dp_value_ptr;
161 return (0x80000000 | ((dp_value_u64 >> 21) & 0x7FFFFFFF));
164 u32 fixed_reciprocal_nr(u32 value, u32 *_shift)
166 u32 shift = __builtin_clz(value);
167 u32 value_normalized = value << shift;
169 *_shift = 62 - shift;
171 value_normalized -= 2;
173 u32 reciprocal_normalized = reciprocal_estimate_u32(value_normalized) >> 1;
175 u32 temp = -(((u64)value_normalized * (u32)reciprocal_normalized) >> 31);
176 reciprocal_normalized = (((u64)reciprocal_normalized * temp) >> 31);
177 temp = -(((u64)value_normalized * (u32)reciprocal_normalized) >> 31);
178 reciprocal_normalized = (((u64)reciprocal_normalized * temp) >> 31);
179 temp = -(((u64)value_normalized * (u32)reciprocal_normalized) >> 31);
180 reciprocal_normalized = (((u64)reciprocal_normalized * temp) >> 31);
182 return reciprocal_normalized;
188 s32 triangle_signed_area_x2(s32 x0, s32 y0, s32 x1, s32 y1, s32 x2, s32 y2)
190 return ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0));
193 u32 texture_region_mask(s32 x1, s32 y1, s32 x2, s32 y2)
195 s32 coverage_x, coverage_y;
200 coverage_x = x2 >> 6;
201 coverage_y = y2 >> 8;
209 mask_down_right = ~(0xFFFFFFFF << (coverage_x + 1)) & 0xFFFF;
212 mask_down_right |= mask_down_right << 16;
214 coverage_x = x1 >> 6;
216 mask_up_left = 0xFFFF0000 << coverage_x;
218 mask_up_left = 0xFFFF0000;
220 coverage_y = y1 >> 8;
222 mask_up_left |= mask_up_left >> 16;
224 return mask_up_left & mask_down_right;
227 u32 invalidate_texture_cache_region(psx_gpu_struct *psx_gpu, u32 x1, u32 y1,
230 u32 mask = texture_region_mask(x1, y1, x2, y2);
232 psx_gpu->dirty_textures_4bpp_mask |= mask;
233 psx_gpu->dirty_textures_8bpp_mask |= mask;
234 psx_gpu->dirty_textures_8bpp_alternate_mask |= mask;
239 u32 invalidate_texture_cache_region_viewport(psx_gpu_struct *psx_gpu, u32 x1,
240 u32 y1, u32 x2, u32 y2)
242 u32 mask = texture_region_mask(x1, y1, x2, y2) &
243 psx_gpu->viewport_mask;
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;
252 void update_texture_cache_region(psx_gpu_struct *psx_gpu, u32 x1, u32 y1,
255 u32 mask = texture_region_mask(x1, y1, x2, y2);
257 u8 *texture_page_ptr;
262 psx_gpu->dirty_textures_8bpp_mask |= mask;
263 psx_gpu->dirty_textures_8bpp_alternate_mask |= mask;
265 if ((psx_gpu->dirty_textures_4bpp_mask & mask) == 0 &&
266 (x1 & 3) == 0 && (y1 & 15) == 0 && x2 - x1 < 4 && y2 - y1 < 16)
268 texture_page = ((x1 / 64) & 15) + (y1 / 256) * 16;
269 texture_page_ptr = psx_gpu->texture_4bpp_cache[texture_page];
270 texture_page_ptr += (x1 / 4 & 15) * 16*16 + (y1 / 16 & 15) * 16*16*16;
271 vram_ptr = psx_gpu->vram_ptr + x1 + y1 * 1024;
279 texel_block = *vram_ptr;
281 texture_page_ptr[0] = texel_block & 0xF;
282 texture_page_ptr[1] = (texel_block >> 4) & 0xF;
283 texture_page_ptr[2] = (texel_block >> 8) & 0xF;
284 texture_page_ptr[3] = texel_block >> 12;
287 texture_page_ptr += 4;
301 psx_gpu->dirty_textures_4bpp_mask |= mask;
307 void update_texture_4bpp_cache(psx_gpu_struct *psx_gpu)
309 u32 current_texture_page = psx_gpu->current_texture_page;
310 u8 *texture_page_ptr = psx_gpu->texture_page_base;
311 u16 *vram_ptr = psx_gpu->vram_ptr;
317 vram_ptr += (current_texture_page >> 4) * 256 * 1024;
318 vram_ptr += (current_texture_page & 0xF) * 64;
320 texture_cache_loads++;
327 psx_gpu->dirty_textures_4bpp_mask &= ~(psx_gpu->current_texture_mask);
337 texel_block = *vram_ptr;
339 texture_page_ptr[0] = texel_block & 0xF;
340 texture_page_ptr[1] = (texel_block >> 4) & 0xF;
341 texture_page_ptr[2] = (texel_block >> 8) & 0xF;
342 texture_page_ptr[3] = texel_block >> 12;
345 texture_page_ptr += 4;
359 vram_ptr -= (1024 * 16) - 4;
365 vram_ptr += (16 * 1024) - (4 * 16);
370 void update_texture_8bpp_cache_slice(psx_gpu_struct *psx_gpu,
373 u16 *texture_page_ptr = psx_gpu->texture_page_base;
374 u16 *vram_ptr = psx_gpu->vram_ptr;
381 texture_cache_loads++;
383 vram_ptr += (texture_page >> 4) * 256 * 1024;
384 vram_ptr += (texture_page & 0xF) * 64;
386 if((texture_page ^ psx_gpu->current_texture_page) & 0x1)
387 texture_page_ptr += (8 * 16) * 8;
400 load_128b(texels, vram_ptr);
401 store_128b(texels, texture_page_ptr);
403 texture_page_ptr += 8;
411 vram_ptr -= (1024 * 16);
420 vram_ptr += (16 * 1024);
422 texture_page_ptr += (8 * 16) * 8;
430 void update_texture_8bpp_cache(psx_gpu_struct *psx_gpu)
432 u32 current_texture_page = psx_gpu->current_texture_page;
433 u32 update_textures =
434 psx_gpu->dirty_textures_8bpp_mask & psx_gpu->current_texture_mask;
436 psx_gpu->dirty_textures_8bpp_mask &= ~update_textures;
438 if(update_textures & (1 << current_texture_page))
440 update_texture_8bpp_cache_slice(psx_gpu, current_texture_page);
441 update_textures &= ~(1 << current_texture_page);
446 u32 adjacent_texture_page = ((current_texture_page + 1) & 0xF) |
447 (current_texture_page & 0x10);
449 update_texture_8bpp_cache_slice(psx_gpu, adjacent_texture_page);
453 void flush_render_block_buffer(psx_gpu_struct *psx_gpu)
455 if((psx_gpu->render_mode & RENDER_INTERLACE_ENABLED) &&
456 (psx_gpu->primitive_type == PRIMITIVE_TYPE_SPRITE))
458 u32 num_blocks_dest = 0;
459 block_struct *block_src = psx_gpu->blocks;
460 block_struct *block_dest = psx_gpu->blocks;
462 u16 *vram_ptr = psx_gpu->vram_ptr;
465 if(psx_gpu->render_mode & RENDER_INTERLACE_ODD)
467 for(i = 0; i < psx_gpu->num_blocks; i++)
469 u32 fb_offset = (u32)((u8 *)block_src->fb_ptr - (u8 *)vram_ptr);
470 if(fb_offset & (1 << 11))
472 *block_dest = *block_src;
481 for(i = 0; i < psx_gpu->num_blocks; i++)
483 u32 fb_offset = (u32)((u8 *)block_src->fb_ptr - (u8 *)vram_ptr);
484 if((fb_offset & (1 << 11)) == 0)
486 *block_dest = *block_src;
494 psx_gpu->num_blocks = num_blocks_dest;
497 if(psx_gpu->num_blocks)
499 render_block_handler_struct *render_block_handler =
500 psx_gpu->render_block_handler;
502 render_block_handler->texture_blocks(psx_gpu);
503 render_block_handler->shade_blocks(psx_gpu);
504 render_block_handler->blend_blocks(psx_gpu);
507 span_pixel_blocks += psx_gpu->num_blocks;
508 render_buffer_flushes++;
511 psx_gpu->num_blocks = 0;
518 #define setup_gradient_calculation_input(set, vertex) \
519 /* First type is: uvrg bxxx xxxx */\
520 /* Second type is: yyyy ybyy uvrg */\
521 /* Since x_a and y_c are the same the same variable is used for both. */\
522 x##set##_a_y##set##_c.e[0] = vertex->u; \
523 x##set##_a_y##set##_c.e[1] = vertex->v; \
524 x##set##_a_y##set##_c.e[2] = vertex->r; \
525 x##set##_a_y##set##_c.e[3] = vertex->g; \
526 dup_4x16b(x##set##_b, vertex->x); \
527 dup_4x16b(x##set##_c, vertex->x); \
528 dup_4x16b(y##set##_a, vertex->y); \
529 dup_4x16b(y##set##_b, vertex->y); \
530 x##set##_b.e[0] = vertex->b; \
531 y##set##_b.e[1] = vertex->b \
534 void compute_all_gradients(psx_gpu_struct *psx_gpu, vertex_struct *a,
535 vertex_struct *b, vertex_struct *c)
537 u32 triangle_area = psx_gpu->triangle_area;
538 u32 winding_mask_scalar;
540 u32 triangle_area_shift;
541 u64 triangle_area_reciprocal =
542 fixed_reciprocal(triangle_area, &triangle_area_shift);
543 triangle_area_shift = -(triangle_area_shift - FIXED_BITS);
545 // ((x1 - x0) * (y2 - y1)) - ((x2 - x1) * (y1 - y0)) =
546 // ( d0 * d1 ) - ( d2 * d3 ) =
547 // ( m0 ) - ( m1 ) = gradient
549 // This is split to do 12 elements at a time over three sets: a, b, and c.
550 // Technically we only need to do 10 elements (uvrgb_x and uvrgb_y), so
551 // two of the slots are unused.
553 // Inputs are all 16-bit signed. The m0/m1 results are 32-bit signed, as
556 vec_4x16s x0_a_y0_c, x0_b, x0_c;
557 vec_4x16s y0_a, y0_b;
558 vec_4x16s x1_a_y1_c, x1_b, x1_c;
559 vec_4x16s y1_a, y1_b;
560 vec_4x16s x2_a_y2_c, x2_b, x2_c;
561 vec_4x16s y2_a, y2_b;
565 vec_4x32u uvrgb_phase;
567 vec_4x16s d0_a_d3_c, d0_b, d0_c;
568 vec_4x16s d1_a, d1_b, d1_c_d2_a;
569 vec_4x16s d2_b, d2_c;
570 vec_4x16s d3_a, d3_b;
572 vec_4x32s m0_a, m0_b, m0_c;
573 vec_4x32s m1_a, m1_b, m1_c;
575 vec_4x32u gradient_area_a, gradient_area_c;
576 vec_2x32u gradient_area_b;
578 vec_4x32u gradient_area_sign_a, gradient_area_sign_c;
579 vec_2x32u gradient_area_sign_b;
580 vec_4x32u winding_mask;
582 vec_2x64u gradient_wide_a0, gradient_wide_a1;
583 vec_2x64u gradient_wide_c0, gradient_wide_c1;
584 vec_2x64u gradient_wide_b;
586 vec_4x32u gradient_a, gradient_c;
587 vec_2x32u gradient_b;
588 vec_16x8s gradient_shift;
590 setup_gradient_calculation_input(0, a);
591 setup_gradient_calculation_input(1, b);
592 setup_gradient_calculation_input(2, c);
594 dup_4x32b(uvrgb_phase, psx_gpu->uvrgb_phase);
595 shl_long_4x16b(uvrg_base, x0_a_y0_c, 16);
596 shl_long_4x16b(b_base, x0_b, 16);
598 add_4x32b(uvrg_base, uvrg_base, uvrgb_phase);
599 add_4x32b(b_base, b_base, uvrgb_phase);
601 // Can probably pair these, but it'll require careful register allocation
602 sub_4x16b(d0_a_d3_c, x1_a_y1_c, x0_a_y0_c);
603 sub_4x16b(d1_c_d2_a, x2_a_y2_c, x1_a_y1_c);
605 sub_4x16b(d0_b, x1_b, x0_b);
606 sub_4x16b(d0_c, x1_c, x0_c);
608 sub_4x16b(d1_a, y2_a, y1_a);
609 sub_4x16b(d1_b, y2_b, y1_b);
611 sub_4x16b(d2_b, x2_b, x1_b);
612 sub_4x16b(d2_c, x2_c, x1_c);
614 sub_4x16b(d3_a, y1_a, y0_a);
615 sub_4x16b(d3_b, y1_b, y0_b);
617 mul_long_4x16b(m0_a, d0_a_d3_c, d1_a);
618 mul_long_4x16b(m0_b, d0_b, d1_b);
619 mul_long_4x16b(m0_c, d0_c, d1_c_d2_a);
621 mul_long_4x16b(m1_a, d1_c_d2_a, d3_a);
622 mul_long_4x16b(m1_b, d2_b, d3_b);
623 mul_long_4x16b(m1_c, d2_c, d0_a_d3_c);
625 sub_4x32b(gradient_area_a, m0_a, m1_a);
626 sub_2x32b(gradient_area_b, m0_b.low, m1_b.low);
627 sub_4x32b(gradient_area_c, m0_c, m1_c);
629 cmpltz_4x32b(gradient_area_sign_a, gradient_area_a);
630 cmpltz_2x32b(gradient_area_sign_b, gradient_area_b);
631 cmpltz_4x32b(gradient_area_sign_c, gradient_area_c);
633 abs_4x32b(gradient_area_a, gradient_area_a);
634 abs_2x32b(gradient_area_b, gradient_area_b);
635 abs_4x32b(gradient_area_c, gradient_area_c);
637 winding_mask_scalar = -psx_gpu->triangle_winding;
639 dup_4x32b(winding_mask, winding_mask_scalar);
640 eor_4x32b(gradient_area_sign_a, gradient_area_sign_a, winding_mask);
641 eor_2x32b(gradient_area_sign_b, gradient_area_sign_b, winding_mask);
642 eor_4x32b(gradient_area_sign_c, gradient_area_sign_c, winding_mask);
644 mul_scalar_long_2x32b(gradient_wide_a0,
645 vector_cast(vec_2x32s, gradient_area_a.low),
646 (s64)triangle_area_reciprocal);
647 mul_scalar_long_2x32b(gradient_wide_a1,
648 vector_cast(vec_2x32s, gradient_area_a.high),
649 (s64)triangle_area_reciprocal);
650 mul_scalar_long_2x32b(gradient_wide_b,
651 vector_cast(vec_2x32s, gradient_area_b),
652 (s64)triangle_area_reciprocal);
653 mul_scalar_long_2x32b(gradient_wide_c0,
654 vector_cast(vec_2x32s, gradient_area_c.low),
655 (s64)triangle_area_reciprocal);
656 mul_scalar_long_2x32b(gradient_wide_c1,
657 vector_cast(vec_2x32s, gradient_area_c.high),
658 (s64)triangle_area_reciprocal);
660 dup_16x8b(gradient_shift, triangle_area_shift);
661 shl_reg_2x64b(gradient_wide_a0, gradient_wide_a0,
662 vector_cast(vec_2x64u, gradient_shift));
663 shl_reg_2x64b(gradient_wide_a1, gradient_wide_a1,
664 vector_cast(vec_2x64u, gradient_shift));
665 shl_reg_2x64b(gradient_wide_b, gradient_wide_b,
666 vector_cast(vec_2x64u, gradient_shift));
667 shl_reg_2x64b(gradient_wide_c0, gradient_wide_c0,
668 vector_cast(vec_2x64u, gradient_shift));
669 shl_reg_2x64b(gradient_wide_c1, gradient_wide_c1,
670 vector_cast(vec_2x64u, gradient_shift));
672 mov_narrow_2x64b(gradient_a.low, gradient_wide_a0);
673 mov_narrow_2x64b(gradient_a.high, gradient_wide_a1);
674 mov_narrow_2x64b(gradient_b, gradient_wide_b);
675 mov_narrow_2x64b(gradient_c.low, gradient_wide_c0);
676 mov_narrow_2x64b(gradient_c.high, gradient_wide_c1);
678 shl_4x32b(gradient_a, gradient_a, 4);
679 shl_2x32b(gradient_b, gradient_b, 4);
680 shl_4x32b(gradient_c, gradient_c, 4);
682 eor_4x32b(gradient_a, gradient_a, gradient_area_sign_a);
683 eor_2x32b(gradient_b, gradient_b, gradient_area_sign_b);
684 eor_4x32b(gradient_c, gradient_c, gradient_area_sign_c);
686 sub_4x32b(gradient_a, gradient_a, gradient_area_sign_a);
687 sub_2x32b(gradient_b, gradient_b, gradient_area_sign_b);
688 sub_4x32b(gradient_c, gradient_c, gradient_area_sign_c);
690 u32 left_adjust = a->x;
691 mls_scalar_4x32b(uvrg_base, gradient_a, left_adjust);
692 mls_scalar_2x32b(b_base.low, gradient_b, left_adjust);
702 eor_4x32b(zero, zero, zero);
703 add_4x32b(uvrg_dx2, gradient_a, gradient_a);
704 add_2x32b(b_dx2, gradient_b, gradient_b);
705 add_4x32b(uvrg_dx3, gradient_a, uvrg_dx2);
706 add_2x32b(b_dx3, gradient_b, b_dx2);
708 // Can be done with vst4, assuming that the zero, dx, dx2, and dx3 are
710 psx_gpu->u_block_span.e[0] = zero.e[0];
711 psx_gpu->u_block_span.e[1] = gradient_a.e[0];
712 psx_gpu->u_block_span.e[2] = uvrg_dx2.e[0];
713 psx_gpu->u_block_span.e[3] = uvrg_dx3.e[0];
715 psx_gpu->v_block_span.e[0] = zero.e[1];
716 psx_gpu->v_block_span.e[1] = gradient_a.e[1];
717 psx_gpu->v_block_span.e[2] = uvrg_dx2.e[1];
718 psx_gpu->v_block_span.e[3] = uvrg_dx3.e[1];
720 psx_gpu->r_block_span.e[0] = zero.e[2];
721 psx_gpu->r_block_span.e[1] = gradient_a.e[2];
722 psx_gpu->r_block_span.e[2] = uvrg_dx2.e[2];
723 psx_gpu->r_block_span.e[3] = uvrg_dx3.e[2];
725 psx_gpu->g_block_span.e[0] = zero.e[3];
726 psx_gpu->g_block_span.e[1] = gradient_a.e[3];
727 psx_gpu->g_block_span.e[2] = uvrg_dx2.e[3];
728 psx_gpu->g_block_span.e[3] = uvrg_dx3.e[3];
730 psx_gpu->b_block_span.e[0] = zero.e[0];
731 psx_gpu->b_block_span.e[1] = gradient_b.e[0];
732 psx_gpu->b_block_span.e[2] = b_dx2.e[0];
733 psx_gpu->b_block_span.e[3] = b_dx3.e[0];
735 psx_gpu->uvrg = uvrg_base;
736 psx_gpu->b = b_base.e[0];
738 psx_gpu->uvrg_dx = gradient_a;
739 psx_gpu->uvrg_dy = gradient_c;
740 psx_gpu->b_dy = gradient_b.e[1];
744 #define vector_check(_a, _b) \
745 if(memcmp(&_a, &_b, sizeof(_b))) \
747 if(sizeof(_b) == 8) \
749 printf("mismatch on %s vs %s: (%x %x) vs (%x %x)\n", \
750 #_a, #_b, _a.e[0], _a.e[1], _b.e[0], _b.e[1]); \
754 printf("mismatch on %s vs %s: (%x %x %x %x) vs (%x %x %x %x)\n", \
755 #_a, #_b, _a.e[0], _a.e[1], _a.e[2], _a.e[3], _b.e[0], _b.e[1], \
760 #define scalar_check(_a, _b) \
762 printf("mismatch on %s %s: %x vs %x\n", #_a, #_b, _a, _b) \
766 #define setup_spans_debug_check(span_edge_data_element) \
768 u32 _num_spans = &span_edge_data_element - psx_gpu->span_edge_data; \
769 if (_num_spans > MAX_SPANS) \
771 if (_num_spans < psx_gpu->num_spans) \
773 if(span_edge_data_element.num_blocks > MAX_BLOCKS_PER_ROW) \
775 if(span_edge_data_element.y > 2048) \
781 #define setup_spans_debug_check(span_edge_data_element) \
785 #define setup_spans_prologue_alternate_yes() \
786 vec_2x64s alternate_x; \
787 vec_2x64s alternate_dx_dy; \
788 vec_4x32s alternate_x_32; \
789 vec_2x32s alternate_x_16; \
791 vec_4x16u alternate_select; \
792 vec_4x16s y_mid_point; \
796 s32 edge_dx_dy_alt; \
799 #define setup_spans_prologue_alternate_no() \
801 #define setup_spans_prologue(alternate_active) \
802 edge_data_struct *span_edge_data; \
803 vec_4x32u *span_uvrg_offset; \
804 u32 *span_b_offset; \
808 vec_2x64s edges_xy; \
809 vec_2x32s edges_dx_dy; \
810 vec_2x32u edge_shifts; \
812 vec_2x64s left_x, right_x; \
813 vec_2x64s left_dx_dy, right_dx_dy; \
814 vec_4x32s left_x_32, right_x_32; \
815 vec_8x16s left_right_x_16; \
817 vec_8x16s left_edge; \
818 vec_8x16s right_edge; \
819 vec_4x16u span_shift; \
823 vec_4x16u c_0xFFFE; \
826 vec_2x32s x_starts; \
835 vec_4x32u uvrg = psx_gpu->uvrg; \
836 vec_4x32u uvrg_dy = psx_gpu->uvrg_dy; \
837 u32 b = psx_gpu->b; \
838 u32 b_dy = psx_gpu->b_dy; \
840 dup_2x32b(c_0x01, 0x01); \
841 setup_spans_prologue_alternate_##alternate_active() \
843 #define setup_spans_prologue_b() \
844 span_edge_data = psx_gpu->span_edge_data; \
845 span_uvrg_offset = psx_gpu->span_uvrg_offset; \
846 span_b_offset = psx_gpu->span_b_offset; \
848 vec_8x16u c_0x0001; \
850 dup_8x16b(c_0x0001, 0x0001); \
851 dup_8x16b(left_edge, psx_gpu->viewport_start_x); \
852 dup_8x16b(right_edge, psx_gpu->viewport_end_x); \
853 add_8x16b(right_edge, right_edge, c_0x0001); \
854 dup_4x16b(c_0x04, 0x04); \
855 dup_4x16b(c_0x07, 0x07); \
856 dup_4x16b(c_0xFFFE, 0xFFFE); \
859 #define compute_edge_delta_x2() \
862 vec_2x32s height_reciprocals; \
863 vec_2x32s heights_b; \
866 u32 edge_shift = reciprocal_table[height]; \
868 dup_2x32b(heights, height); \
869 sub_2x32b(widths, x_ends, x_starts); \
871 dup_2x32b(edge_shifts, edge_shift); \
872 sub_2x32b(heights_b, heights, c_0x01); \
873 shr_2x32b(height_reciprocals, edge_shifts, 10); \
875 mla_2x32b(heights_b, x_starts, heights); \
876 bic_immediate_4x16b(vector_cast(vec_4x16u, edge_shifts), 0xE0); \
877 mul_2x32b(edges_dx_dy, widths, height_reciprocals); \
878 mul_long_2x32b(edges_xy, heights_b, height_reciprocals); \
881 #define compute_edge_delta_x3(start_c, height_a, height_b) \
884 vec_2x32s height_reciprocals; \
885 vec_2x32s heights_b; \
890 u32 height_reciprocal_alt; \
892 heights.e[0] = height_a; \
893 heights.e[1] = height_b; \
895 edge_shifts.e[0] = reciprocal_table[height_a]; \
896 edge_shifts.e[1] = reciprocal_table[height_b]; \
897 edge_shift_alt = reciprocal_table[height_minor_b]; \
899 sub_2x32b(widths, x_ends, x_starts); \
900 width_alt = x_c - start_c; \
902 shr_2x32b(height_reciprocals, edge_shifts, 10); \
903 height_reciprocal_alt = edge_shift_alt >> 10; \
905 bic_immediate_4x16b(vector_cast(vec_4x16u, edge_shifts), 0xE0); \
906 edge_shift_alt &= 0x1F; \
908 sub_2x32b(heights_b, heights, c_0x01); \
909 height_b_alt = height_minor_b - 1; \
911 mla_2x32b(heights_b, x_starts, heights); \
912 height_b_alt += height_minor_b * start_c; \
914 mul_long_2x32b(edges_xy, heights_b, height_reciprocals); \
915 edge_alt = (s64)height_b_alt * height_reciprocal_alt; \
917 mul_2x32b(edges_dx_dy, widths, height_reciprocals); \
918 edge_dx_dy_alt = width_alt * height_reciprocal_alt; \
922 #define setup_spans_adjust_y_up() \
923 sub_4x32b(y_x4, y_x4, c_0x04) \
925 #define setup_spans_adjust_y_down() \
926 add_4x32b(y_x4, y_x4, c_0x04) \
928 #define setup_spans_adjust_interpolants_up() \
929 sub_4x32b(uvrg, uvrg, uvrg_dy); \
932 #define setup_spans_adjust_interpolants_down() \
933 add_4x32b(uvrg, uvrg, uvrg_dy); \
937 #define setup_spans_clip_interpolants_increment() \
938 mla_scalar_4x32b(uvrg, uvrg_dy, clip); \
941 #define setup_spans_clip_interpolants_decrement() \
942 mls_scalar_4x32b(uvrg, uvrg_dy, clip); \
945 #define setup_spans_clip_alternate_yes() \
946 edge_alt += edge_dx_dy_alt * (s64)(clip) \
948 #define setup_spans_clip_alternate_no() \
950 #define setup_spans_clip(direction, alternate_active) \
952 clipped_triangles++; \
953 mla_scalar_long_2x32b(edges_xy, edges_dx_dy, (s64)clip); \
954 setup_spans_clip_alternate_##alternate_active(); \
955 setup_spans_clip_interpolants_##direction(); \
959 #define setup_spans_adjust_edges_alternate_no(left_index, right_index) \
961 vec_2x64u edge_shifts_64; \
962 vec_2x64s edges_dx_dy_64; \
964 mov_wide_2x32b(edge_shifts_64, edge_shifts); \
965 shl_variable_2x64b(edges_xy, edges_xy, edge_shifts_64); \
967 mov_wide_2x32b(edges_dx_dy_64, edges_dx_dy); \
968 shl_variable_2x64b(edges_dx_dy_64, edges_dx_dy_64, edge_shifts_64); \
970 left_x.e[0] = edges_xy.e[left_index]; \
971 right_x.e[0] = edges_xy.e[right_index]; \
973 left_dx_dy.e[0] = edges_dx_dy_64.e[left_index]; \
974 left_dx_dy.e[1] = edges_dx_dy_64.e[left_index]; \
975 right_dx_dy.e[0] = edges_dx_dy_64.e[right_index]; \
976 right_dx_dy.e[1] = edges_dx_dy_64.e[right_index]; \
978 add_1x64b(left_x.high, left_x.low, left_dx_dy.low); \
979 add_1x64b(right_x.high, right_x.low, right_dx_dy.low); \
981 add_2x64b(left_dx_dy, left_dx_dy, left_dx_dy); \
982 add_2x64b(right_dx_dy, right_dx_dy, right_dx_dy); \
985 #define setup_spans_adjust_edges_alternate_yes(left_index, right_index) \
987 setup_spans_adjust_edges_alternate_no(left_index, right_index); \
988 s64 edge_dx_dy_alt_64; \
990 dup_4x16b(y_mid_point, y_b); \
992 edge_alt <<= edge_shift_alt; \
993 edge_dx_dy_alt_64 = (s64)edge_dx_dy_alt << edge_shift_alt; \
995 alternate_x.e[0] = edge_alt; \
996 alternate_dx_dy.e[0] = edge_dx_dy_alt_64; \
997 alternate_dx_dy.e[1] = edge_dx_dy_alt_64; \
999 add_1x64b(alternate_x.high, alternate_x.low, alternate_dx_dy.low); \
1000 add_2x64b(alternate_dx_dy, alternate_dx_dy, alternate_dx_dy); \
1004 #define setup_spans_y_select_up() \
1005 cmplt_4x16b(alternate_select, y_x4, y_mid_point) \
1007 #define setup_spans_y_select_down() \
1008 cmpgt_4x16b(alternate_select, y_x4, y_mid_point) \
1010 #define setup_spans_y_select_alternate_yes(direction) \
1011 setup_spans_y_select_##direction() \
1013 #define setup_spans_y_select_alternate_no(direction) \
1015 #define setup_spans_alternate_select_left() \
1016 bit_4x16b(left_right_x_16.low, alternate_x_16, alternate_select) \
1018 #define setup_spans_alternate_select_right() \
1019 bit_4x16b(left_right_x_16.high, alternate_x_16, alternate_select) \
1021 #define setup_spans_alternate_select_none() \
1023 #define setup_spans_increment_alternate_yes() \
1024 shr_narrow_2x64b(alternate_x_32.low, alternate_x, 32); \
1025 add_2x64b(alternate_x, alternate_x, alternate_dx_dy); \
1026 shr_narrow_2x64b(alternate_x_32.high, alternate_x, 32); \
1027 add_2x64b(alternate_x, alternate_x, alternate_dx_dy); \
1028 mov_narrow_4x32b(alternate_x_16, alternate_x_32) \
1030 #define setup_spans_increment_alternate_no() \
1032 #define setup_spans_set_x4(alternate, direction, alternate_active) \
1034 span_uvrg_offset[0] = uvrg; \
1035 span_b_offset[0] = b; \
1036 setup_spans_adjust_interpolants_##direction(); \
1038 span_uvrg_offset[1] = uvrg; \
1039 span_b_offset[1] = b; \
1040 setup_spans_adjust_interpolants_##direction(); \
1042 span_uvrg_offset[2] = uvrg; \
1043 span_b_offset[2] = b; \
1044 setup_spans_adjust_interpolants_##direction(); \
1046 span_uvrg_offset[3] = uvrg; \
1047 span_b_offset[3] = b; \
1048 setup_spans_adjust_interpolants_##direction(); \
1050 span_uvrg_offset += 4; \
1051 span_b_offset += 4; \
1053 shr_narrow_2x64b(left_x_32.low, left_x, 32); \
1054 shr_narrow_2x64b(right_x_32.low, right_x, 32); \
1056 add_2x64b(left_x, left_x, left_dx_dy); \
1057 add_2x64b(right_x, right_x, right_dx_dy); \
1059 shr_narrow_2x64b(left_x_32.high, left_x, 32); \
1060 shr_narrow_2x64b(right_x_32.high, right_x, 32); \
1062 add_2x64b(left_x, left_x, left_dx_dy); \
1063 add_2x64b(right_x, right_x, right_dx_dy); \
1065 mov_narrow_4x32b(left_right_x_16.low, left_x_32); \
1066 mov_narrow_4x32b(left_right_x_16.high, right_x_32); \
1068 setup_spans_increment_alternate_##alternate_active(); \
1069 setup_spans_y_select_alternate_##alternate_active(direction); \
1070 setup_spans_alternate_select_##alternate(); \
1072 max_8x16b(left_right_x_16, left_right_x_16, left_edge); \
1073 min_8x16b(left_right_x_16, left_right_x_16, right_edge); \
1075 sub_4x16b(left_right_x_16.high, left_right_x_16.high, left_right_x_16.low); \
1076 add_4x16b(left_right_x_16.high, left_right_x_16.high, c_0x07); \
1077 and_4x16b(span_shift, left_right_x_16.high, c_0x07); \
1078 shl_variable_4x16b(span_shift, c_0xFFFE, span_shift); \
1079 shr_4x16b(left_right_x_16.high, left_right_x_16.high, 3); \
1082 for(i = 0; i < 4; i++) \
1084 span_edge_data[i].left_x = left_right_x_16.low.e[i]; \
1085 span_edge_data[i].num_blocks = left_right_x_16.high.e[i]; \
1086 span_edge_data[i].right_mask = span_shift.e[i]; \
1087 span_edge_data[i].y = y_x4.e[i]; \
1088 setup_spans_debug_check(span_edge_data[i]); \
1091 span_edge_data += 4; \
1093 setup_spans_adjust_y_##direction(); \
1097 #define setup_spans_alternate_adjust_yes() \
1098 edge_alt -= edge_dx_dy_alt * (s64)height_minor_a \
1100 #define setup_spans_alternate_adjust_no() \
1103 #define setup_spans_down(left_index, right_index, alternate, alternate_active) \
1104 setup_spans_alternate_adjust_##alternate_active(); \
1105 if(y_c > psx_gpu->viewport_end_y) \
1106 height -= y_c - psx_gpu->viewport_end_y - 1; \
1108 clip = psx_gpu->viewport_start_y - y_a; \
1113 setup_spans_clip(increment, alternate_active); \
1116 setup_spans_prologue_b(); \
1121 y_x4.e[1] = y_a + 1; \
1122 y_x4.e[2] = y_a + 2; \
1123 y_x4.e[3] = y_a + 3; \
1124 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
1127 psx_gpu->num_spans = height; \
1130 setup_spans_set_x4(alternate, down, alternate_active); \
1132 } while(height > 0); \
1136 #define setup_spans_alternate_pre_increment_yes() \
1137 edge_alt += edge_dx_dy_alt \
1139 #define setup_spans_alternate_pre_increment_no() \
1141 #define setup_spans_up_decrement_height_yes() \
1144 #define setup_spans_up_decrement_height_no() \
1147 #define setup_spans_up(left_index, right_index, alternate, alternate_active) \
1148 setup_spans_alternate_adjust_##alternate_active(); \
1151 if(y_c < psx_gpu->viewport_start_y) \
1152 height -= psx_gpu->viewport_start_y - y_c; \
1154 setup_spans_up_decrement_height_##alternate_active(); \
1156 clip = y_a - psx_gpu->viewport_end_y; \
1161 setup_spans_clip(decrement, alternate_active); \
1164 setup_spans_prologue_b(); \
1169 y_x4.e[1] = y_a - 1; \
1170 y_x4.e[2] = y_a - 2; \
1171 y_x4.e[3] = y_a - 3; \
1172 add_wide_2x32b(edges_xy, edges_xy, edges_dx_dy); \
1173 setup_spans_alternate_pre_increment_##alternate_active(); \
1174 setup_spans_adjust_edges_alternate_##alternate_active(left_index, \
1176 setup_spans_adjust_interpolants_up(); \
1178 psx_gpu->num_spans = height; \
1181 setup_spans_set_x4(alternate, up, alternate_active); \
1186 #define index_left 0
1187 #define index_right 1
1189 #define setup_spans_up_up(minor, major) \
1190 setup_spans_prologue(yes); \
1191 s32 height_minor_a = y_a - y_b; \
1192 s32 height_minor_b = y_b - y_c; \
1193 s32 height = y_a - y_c; \
1195 dup_2x32b(x_starts, x_a); \
1196 x_ends.e[0] = x_c; \
1197 x_ends.e[1] = x_b; \
1199 compute_edge_delta_x3(x_b, height, height_minor_a); \
1200 setup_spans_up(index_##major, index_##minor, minor, yes) \
1205 void setup_spans_up_left(psx_gpu_struct *psx_gpu, vertex_struct *v_a,
1206 vertex_struct *v_b, vertex_struct *v_c)
1208 setup_spans_up_up(left, right);
1211 void setup_spans_up_right(psx_gpu_struct *psx_gpu, vertex_struct *v_a,
1212 vertex_struct *v_b, vertex_struct *v_c)
1214 setup_spans_up_up(right, left);
1217 #define setup_spans_down_down(minor, major) \
1218 setup_spans_prologue(yes); \
1219 s32 height_minor_a = y_b - y_a; \
1220 s32 height_minor_b = y_c - y_b; \
1221 s32 height = y_c - y_a; \
1223 dup_2x32b(x_starts, x_a); \
1224 x_ends.e[0] = x_c; \
1225 x_ends.e[1] = x_b; \
1227 compute_edge_delta_x3(x_b, height, height_minor_a); \
1228 setup_spans_down(index_##major, index_##minor, minor, yes) \
1230 void setup_spans_down_left(psx_gpu_struct *psx_gpu, vertex_struct *v_a,
1231 vertex_struct *v_b, vertex_struct *v_c)
1233 setup_spans_down_down(left, right);
1236 void setup_spans_down_right(psx_gpu_struct *psx_gpu, vertex_struct *v_a,
1237 vertex_struct *v_b, vertex_struct *v_c)
1239 setup_spans_down_down(right, left);
1242 #define setup_spans_up_flat() \
1243 s32 height = y_a - y_c; \
1246 compute_edge_delta_x2(); \
1247 setup_spans_up(index_left, index_right, none, no) \
1249 void setup_spans_up_a(psx_gpu_struct *psx_gpu, vertex_struct *v_a,
1250 vertex_struct *v_b, vertex_struct *v_c)
1252 setup_spans_prologue(no);
1253 x_starts.e[0] = x_a;
1254 x_starts.e[1] = x_b;
1255 dup_2x32b(x_ends, x_c);
1257 setup_spans_up_flat();
1260 void setup_spans_up_b(psx_gpu_struct *psx_gpu, vertex_struct *v_a,
1261 vertex_struct *v_b, vertex_struct *v_c)
1263 setup_spans_prologue(no);
1264 dup_2x32b(x_starts, x_a);
1268 setup_spans_up_flat();
1271 #define setup_spans_down_flat() \
1272 s32 height = y_c - y_a; \
1275 compute_edge_delta_x2(); \
1276 setup_spans_down(index_left, index_right, none, no) \
1278 void setup_spans_down_a(psx_gpu_struct *psx_gpu, vertex_struct *v_a,
1279 vertex_struct *v_b, vertex_struct *v_c)
1281 setup_spans_prologue(no);
1282 x_starts.e[0] = x_a;
1283 x_starts.e[1] = x_b;
1284 dup_2x32b(x_ends, x_c);
1286 setup_spans_down_flat();
1289 void setup_spans_down_b(psx_gpu_struct *psx_gpu, vertex_struct *v_a,
1290 vertex_struct *v_b, vertex_struct *v_c)
1292 setup_spans_prologue(no);
1293 dup_2x32b(x_starts, x_a);
1297 setup_spans_down_flat();
1300 void setup_spans_up_down(psx_gpu_struct *psx_gpu, vertex_struct *v_a,
1301 vertex_struct *v_b, vertex_struct *v_c)
1303 setup_spans_prologue(no);
1311 s32 height_minor_a = y_a - y_b;
1312 s32 height_minor_b = y_c - y_a;
1313 s32 height_major = y_c - y_b;
1315 vec_2x64s edges_xy_b;
1316 vec_2x32s edges_dx_dy_b;
1317 vec_2x32u edge_shifts_b;
1319 vec_2x32s height_increment;
1321 x_starts.e[0] = x_a;
1322 x_starts.e[1] = x_c;
1323 dup_2x32b(x_ends, x_b);
1325 compute_edge_delta_x3(x_a, height_minor_a, height_major);
1327 height_increment.e[0] = 0;
1328 height_increment.e[1] = height_minor_b;
1330 mla_long_2x32b(edges_xy, edges_dx_dy, height_increment);
1332 edges_xy_b.e[0] = edge_alt;
1333 edges_xy_b.e[1] = edges_xy.e[1];
1335 edge_shifts_b = edge_shifts;
1336 edge_shifts_b.e[0] = edge_shift_alt;
1338 neg_2x32b(edges_dx_dy_b, edges_dx_dy);
1339 edges_dx_dy_b.e[0] = edge_dx_dy_alt;
1343 if(y_b < psx_gpu->viewport_start_y)
1344 height_minor_a -= psx_gpu->viewport_start_y - y_b;
1346 clip = y_a - psx_gpu->viewport_end_y;
1349 height_minor_a -= clip;
1351 setup_spans_clip(decrement, no);
1354 setup_spans_prologue_b();
1356 if(height_minor_a > 0)
1359 y_x4.e[1] = y_a - 1;
1360 y_x4.e[2] = y_a - 2;
1361 y_x4.e[3] = y_a - 3;
1362 add_wide_2x32b(edges_xy, edges_xy, edges_dx_dy);
1363 setup_spans_adjust_edges_alternate_no(index_left, index_right);
1364 setup_spans_adjust_interpolants_up();
1366 psx_gpu->num_spans = height_minor_a;
1367 while(height_minor_a > 0)
1369 setup_spans_set_x4(none, up, no);
1370 height_minor_a -= 4;
1373 span_edge_data += height_minor_a;
1374 span_uvrg_offset += height_minor_a;
1375 span_b_offset += height_minor_a;
1378 edges_xy = edges_xy_b;
1379 edges_dx_dy = edges_dx_dy_b;
1380 edge_shifts = edge_shifts_b;
1382 uvrg = psx_gpu->uvrg;
1387 if(y_c > psx_gpu->viewport_end_y)
1388 height_minor_b -= y_c - psx_gpu->viewport_end_y - 1;
1390 clip = psx_gpu->viewport_start_y - y_a;
1393 height_minor_b -= clip;
1395 setup_spans_clip(increment, no);
1398 if(height_minor_b > 0)
1401 y_x4.e[1] = y_a + 1;
1402 y_x4.e[2] = y_a + 2;
1403 y_x4.e[3] = y_a + 3;
1404 setup_spans_adjust_edges_alternate_no(index_left, index_right);
1406 // FIXME: overflow corner case
1407 if(psx_gpu->num_spans + height_minor_b == MAX_SPANS)
1408 height_minor_b &= ~3;
1410 psx_gpu->num_spans += height_minor_b;
1411 while(height_minor_b > 0)
1413 setup_spans_set_x4(none, down, no);
1414 height_minor_b -= 4;
1418 left_split_triangles++;
1424 #define dither_table_entry_normal(value) \
1428 #define setup_blocks_load_msb_mask_indirect() \
1430 #define setup_blocks_load_msb_mask_direct() \
1431 vec_8x16u msb_mask; \
1432 dup_8x16b(msb_mask, psx_gpu->mask_msb); \
1435 #define setup_blocks_variables_shaded_textured(target) \
1436 vec_4x32u u_block; \
1437 vec_4x32u v_block; \
1438 vec_4x32u r_block; \
1439 vec_4x32u g_block; \
1440 vec_4x32u b_block; \
1441 vec_4x32u uvrg_dx = psx_gpu->uvrg_dx; \
1442 vec_4x32u uvrg_dx4; \
1443 vec_4x32u uvrg_dx8; \
1445 u32 b_dx = psx_gpu->b_block_span.e[1]; \
1446 u32 b_dx4 = b_dx << 2; \
1447 u32 b_dx8 = b_dx << 3; \
1450 vec_16x8u texture_mask; \
1451 shl_4x32b(uvrg_dx4, uvrg_dx, 2); \
1452 shl_4x32b(uvrg_dx8, uvrg_dx, 3); \
1453 dup_8x8b(texture_mask.low, psx_gpu->texture_mask_width); \
1454 dup_8x8b(texture_mask.high, psx_gpu->texture_mask_height) \
1456 #define setup_blocks_variables_shaded_untextured(target) \
1457 vec_4x32u r_block; \
1458 vec_4x32u g_block; \
1459 vec_4x32u b_block; \
1461 vec_4x32u rgb_dx4; \
1462 vec_4x32u rgb_dx8; \
1465 vec_8x8u d64_0x07; \
1470 dup_8x8b(d64_0x07, 0x07); \
1471 dup_8x8b(d64_1, 1); \
1472 dup_8x8b(d64_4, 4); \
1473 dup_8x8b(d64_128, 128); \
1475 rgb_dx.low = psx_gpu->uvrg_dx.high; \
1476 rgb_dx.e[2] = psx_gpu->b_block_span.e[1]; \
1477 shl_4x32b(rgb_dx4, rgb_dx, 2); \
1478 shl_4x32b(rgb_dx8, rgb_dx, 3) \
1480 #define setup_blocks_variables_unshaded_textured(target) \
1481 vec_4x32u u_block; \
1482 vec_4x32u v_block; \
1483 vec_2x32u uv_dx = psx_gpu->uvrg_dx.low; \
1486 vec_2x32u uv = psx_gpu->uvrg.low; \
1488 vec_16x8u texture_mask; \
1489 shl_2x32b(uv_dx4, uv_dx, 2); \
1490 shl_2x32b(uv_dx8, uv_dx, 3); \
1491 dup_8x8b(texture_mask.low, psx_gpu->texture_mask_width); \
1492 dup_8x8b(texture_mask.high, psx_gpu->texture_mask_height) \
1495 #define setup_blocks_variables_unshaded_untextured_direct() \
1496 or_8x16b(colors, colors, msb_mask) \
1498 #define setup_blocks_variables_unshaded_untextured_indirect() \
1500 #define setup_blocks_variables_unshaded_untextured(target) \
1501 u32 color = psx_gpu->triangle_color; \
1504 u32 color_r = color & 0xFF; \
1505 u32 color_g = (color >> 8) & 0xFF; \
1506 u32 color_b = (color >> 16) & 0xFF; \
1508 color = (color_r >> 3) | ((color_g >> 3) << 5) | \
1509 ((color_b >> 3) << 10); \
1510 dup_8x16b(colors, color); \
1511 setup_blocks_variables_unshaded_untextured_##target() \
1513 #define setup_blocks_span_initialize_dithered_textured() \
1514 vec_8x16u dither_offsets; \
1515 shl_long_8x8b(dither_offsets, dither_offsets_short, 4) \
1517 #define setup_blocks_span_initialize_dithered_untextured() \
1518 vec_8x8u dither_offsets; \
1519 add_8x8b(dither_offsets, dither_offsets_short, d64_4) \
1521 #define setup_blocks_span_initialize_dithered(texturing) \
1522 u32 dither_row = psx_gpu->dither_table[y & 0x3]; \
1523 u32 dither_shift = (span_edge_data->left_x & 0x3) * 8; \
1524 vec_8x8s dither_offsets_short; \
1527 (dither_row >> dither_shift) | (dither_row << (32 - dither_shift)); \
1528 dup_2x32b(vector_cast(vec_2x32u, dither_offsets_short), dither_row); \
1529 setup_blocks_span_initialize_dithered_##texturing() \
1531 #define setup_blocks_span_initialize_undithered(texturing) \
1534 #define setup_blocks_span_initialize_shaded_textured() \
1536 vec_4x32u block_span; \
1537 u32 offset = span_edge_data->left_x; \
1539 uvrg = *span_uvrg_offset; \
1540 mla_scalar_4x32b(uvrg, uvrg_dx, offset); \
1541 b = *span_b_offset; \
1542 b += b_dx * offset; \
1544 dup_4x32b(u_block, uvrg.e[0]); \
1545 dup_4x32b(v_block, uvrg.e[1]); \
1546 dup_4x32b(r_block, uvrg.e[2]); \
1547 dup_4x32b(g_block, uvrg.e[3]); \
1548 dup_4x32b(b_block, b); \
1550 block_span = psx_gpu->u_block_span; \
1551 add_4x32b(u_block, u_block, block_span); \
1552 block_span = psx_gpu->v_block_span; \
1553 add_4x32b(v_block, v_block, block_span); \
1554 block_span = psx_gpu->r_block_span; \
1555 add_4x32b(r_block, r_block, block_span); \
1556 block_span = psx_gpu->g_block_span; \
1557 add_4x32b(g_block, g_block, block_span); \
1558 block_span = psx_gpu->b_block_span; \
1559 add_4x32b(b_block, b_block, block_span); \
1562 #define setup_blocks_span_initialize_shaded_untextured() \
1564 vec_4x32u block_span; \
1565 u32 offset = span_edge_data->left_x; \
1567 rgb.low = span_uvrg_offset->high; \
1568 rgb.high.e[0] = *span_b_offset; \
1569 mla_scalar_4x32b(rgb, rgb_dx, offset); \
1571 dup_4x32b(r_block, rgb.e[0]); \
1572 dup_4x32b(g_block, rgb.e[1]); \
1573 dup_4x32b(b_block, rgb.e[2]); \
1575 block_span = psx_gpu->r_block_span; \
1576 add_4x32b(r_block, r_block, block_span); \
1577 block_span = psx_gpu->g_block_span; \
1578 add_4x32b(g_block, g_block, block_span); \
1579 block_span = psx_gpu->b_block_span; \
1580 add_4x32b(b_block, b_block, block_span); \
1583 #define setup_blocks_span_initialize_unshaded_textured() \
1585 vec_4x32u block_span; \
1586 u32 offset = span_edge_data->left_x; \
1588 uv = span_uvrg_offset->low; \
1589 mla_scalar_2x32b(uv, uv_dx, offset); \
1591 dup_4x32b(u_block, uv.e[0]); \
1592 dup_4x32b(v_block, uv.e[1]); \
1594 block_span = psx_gpu->u_block_span; \
1595 add_4x32b(u_block, u_block, block_span); \
1596 block_span = psx_gpu->v_block_span; \
1597 add_4x32b(v_block, v_block, block_span); \
1600 #define setup_blocks_span_initialize_unshaded_untextured() \
1603 #define setup_blocks_texture_swizzled() \
1605 vec_8x8u u_saved = u; \
1606 sli_8x8b(u, v, 4); \
1607 sri_8x8b(v, u_saved, 4); \
1610 #define setup_blocks_texture_unswizzled() \
1612 #define setup_blocks_store_shaded_textured(swizzling, dithering, target, \
1615 vec_8x16u u_whole; \
1616 vec_8x16u v_whole; \
1617 vec_8x16u r_whole; \
1618 vec_8x16u g_whole; \
1619 vec_8x16u b_whole; \
1631 shr_narrow_4x32b(u_whole.low, u_block, 16); \
1632 shr_narrow_4x32b(v_whole.low, v_block, 16); \
1633 shr_narrow_4x32b(r_whole.low, r_block, 16); \
1634 shr_narrow_4x32b(g_whole.low, g_block, 16); \
1635 shr_narrow_4x32b(b_whole.low, b_block, 16); \
1637 dup_4x32b(dx4, uvrg_dx4.e[0]); \
1638 add_high_narrow_4x32b(u_whole.high, u_block, dx4); \
1639 dup_4x32b(dx4, uvrg_dx4.e[1]); \
1640 add_high_narrow_4x32b(v_whole.high, v_block, dx4); \
1641 dup_4x32b(dx4, uvrg_dx4.e[2]); \
1642 add_high_narrow_4x32b(r_whole.high, r_block, dx4); \
1643 dup_4x32b(dx4, uvrg_dx4.e[3]); \
1644 add_high_narrow_4x32b(g_whole.high, g_block, dx4); \
1645 dup_4x32b(dx4, b_dx4); \
1646 add_high_narrow_4x32b(b_whole.high, b_block, dx4); \
1648 mov_narrow_8x16b(u, u_whole); \
1649 mov_narrow_8x16b(v, v_whole); \
1650 mov_narrow_8x16b(r, r_whole); \
1651 mov_narrow_8x16b(g, g_whole); \
1652 mov_narrow_8x16b(b, b_whole); \
1654 dup_4x32b(dx8, uvrg_dx8.e[0]); \
1655 add_4x32b(u_block, u_block, dx8); \
1656 dup_4x32b(dx8, uvrg_dx8.e[1]); \
1657 add_4x32b(v_block, v_block, dx8); \
1658 dup_4x32b(dx8, uvrg_dx8.e[2]); \
1659 add_4x32b(r_block, r_block, dx8); \
1660 dup_4x32b(dx8, uvrg_dx8.e[3]); \
1661 add_4x32b(g_block, g_block, dx8); \
1662 dup_4x32b(dx8, b_dx8); \
1663 add_4x32b(b_block, b_block, dx8); \
1665 and_8x8b(u, u, texture_mask.low); \
1666 and_8x8b(v, v, texture_mask.high); \
1667 setup_blocks_texture_##swizzling(); \
1669 zip_8x16b(uv, u, v); \
1674 block->dither_offsets = vector_cast(vec_8x16u, dither_offsets); \
1675 block->fb_ptr = fb_ptr; \
1678 #define setup_blocks_store_unshaded_textured(swizzling, dithering, target, \
1681 vec_8x16u u_whole; \
1682 vec_8x16u v_whole; \
1691 shr_narrow_4x32b(u_whole.low, u_block, 16); \
1692 shr_narrow_4x32b(v_whole.low, v_block, 16); \
1694 dup_4x32b(dx4, uv_dx4.e[0]); \
1695 add_high_narrow_4x32b(u_whole.high, u_block, dx4); \
1696 dup_4x32b(dx4, uv_dx4.e[1]); \
1697 add_high_narrow_4x32b(v_whole.high, v_block, dx4); \
1699 mov_narrow_8x16b(u, u_whole); \
1700 mov_narrow_8x16b(v, v_whole); \
1702 dup_4x32b(dx8, uv_dx8.e[0]); \
1703 add_4x32b(u_block, u_block, dx8); \
1704 dup_4x32b(dx8, uv_dx8.e[1]); \
1705 add_4x32b(v_block, v_block, dx8); \
1707 and_8x8b(u, u, texture_mask.low); \
1708 and_8x8b(v, v, texture_mask.high); \
1709 setup_blocks_texture_##swizzling(); \
1711 zip_8x16b(uv, u, v); \
1713 block->dither_offsets = vector_cast(vec_8x16u, dither_offsets); \
1714 block->fb_ptr = fb_ptr; \
1717 #define setup_blocks_store_shaded_untextured_dithered() \
1718 addq_8x8b(r, r, dither_offsets); \
1719 addq_8x8b(g, g, dither_offsets); \
1720 addq_8x8b(b, b, dither_offsets); \
1722 subq_8x8b(r, r, d64_4); \
1723 subq_8x8b(g, g, d64_4); \
1724 subq_8x8b(b, b, d64_4) \
1726 #define setup_blocks_store_shaded_untextured_undithered() \
1729 #define setup_blocks_store_untextured_pixels_indirect_full(_pixels) \
1730 block->pixels = _pixels; \
1731 block->fb_ptr = fb_ptr \
1733 #define setup_blocks_store_untextured_pixels_indirect_edge(_pixels) \
1734 block->pixels = _pixels; \
1735 block->fb_ptr = fb_ptr \
1737 #define setup_blocks_store_shaded_untextured_seed_pixels_indirect() \
1738 mul_long_8x8b(pixels, r, d64_1) \
1741 #define setup_blocks_store_untextured_pixels_direct_full(_pixels) \
1742 store_8x16b(_pixels, fb_ptr) \
1744 #define setup_blocks_store_untextured_pixels_direct_edge(_pixels) \
1746 vec_8x16u fb_pixels; \
1747 vec_8x16u draw_mask; \
1748 vec_8x16u test_mask = psx_gpu->test_mask; \
1750 load_8x16b(fb_pixels, fb_ptr); \
1751 dup_8x16b(draw_mask, span_edge_data->right_mask); \
1752 tst_8x16b(draw_mask, draw_mask, test_mask); \
1753 bif_8x16b(fb_pixels, _pixels, draw_mask); \
1754 store_8x16b(fb_pixels, fb_ptr); \
1757 #define setup_blocks_store_shaded_untextured_seed_pixels_direct() \
1758 pixels = msb_mask; \
1759 mla_long_8x8b(pixels, r, d64_1) \
1762 #define setup_blocks_store_shaded_untextured(swizzling, dithering, target, \
1765 vec_8x16u r_whole; \
1766 vec_8x16u g_whole; \
1767 vec_8x16u b_whole; \
1778 shr_narrow_4x32b(r_whole.low, r_block, 16); \
1779 shr_narrow_4x32b(g_whole.low, g_block, 16); \
1780 shr_narrow_4x32b(b_whole.low, b_block, 16); \
1782 dup_4x32b(dx4, rgb_dx4.e[0]); \
1783 add_high_narrow_4x32b(r_whole.high, r_block, dx4); \
1784 dup_4x32b(dx4, rgb_dx4.e[1]); \
1785 add_high_narrow_4x32b(g_whole.high, g_block, dx4); \
1786 dup_4x32b(dx4, rgb_dx4.e[2]); \
1787 add_high_narrow_4x32b(b_whole.high, b_block, dx4); \
1789 mov_narrow_8x16b(r, r_whole); \
1790 mov_narrow_8x16b(g, g_whole); \
1791 mov_narrow_8x16b(b, b_whole); \
1793 dup_4x32b(dx8, rgb_dx8.e[0]); \
1794 add_4x32b(r_block, r_block, dx8); \
1795 dup_4x32b(dx8, rgb_dx8.e[1]); \
1796 add_4x32b(g_block, g_block, dx8); \
1797 dup_4x32b(dx8, rgb_dx8.e[2]); \
1798 add_4x32b(b_block, b_block, dx8); \
1800 setup_blocks_store_shaded_untextured_##dithering(); \
1802 shr_8x8b(r, r, 3); \
1803 bic_8x8b(g, g, d64_0x07); \
1804 bic_8x8b(b, b, d64_0x07); \
1806 setup_blocks_store_shaded_untextured_seed_pixels_##target(); \
1807 mla_long_8x8b(pixels, g, d64_4); \
1808 mla_long_8x8b(pixels, b, d64_128) \
1810 setup_blocks_store_untextured_pixels_##target##_##edge_type(pixels); \
1813 #define setup_blocks_store_unshaded_untextured(swizzling, dithering, target, \
1815 setup_blocks_store_untextured_pixels_##target##_##edge_type(colors) \
1818 #define setup_blocks_store_draw_mask_textured_indirect(_block, bits) \
1819 (_block)->draw_mask_bits = bits \
1821 #define setup_blocks_store_draw_mask_untextured_indirect(_block, bits) \
1823 vec_8x16u bits_mask; \
1824 vec_8x16u test_mask = psx_gpu->test_mask; \
1825 dup_8x16b(bits_mask, bits); \
1826 tst_8x16b(bits_mask, bits_mask, test_mask); \
1827 (_block)->draw_mask = bits_mask; \
1830 #define setup_blocks_store_draw_mask_untextured_direct(_block, bits) \
1833 #define setup_blocks_add_blocks_indirect() \
1834 num_blocks += span_num_blocks; \
1836 if(num_blocks > MAX_BLOCKS) \
1838 psx_gpu->num_blocks = num_blocks - span_num_blocks; \
1839 flush_render_block_buffer(psx_gpu); \
1840 num_blocks = span_num_blocks; \
1841 block = psx_gpu->blocks; \
1844 #define setup_blocks_add_blocks_direct() \
1845 texel_blocks_untextured += span_num_blocks; \
1846 span_pixel_blocks += span_num_blocks \
1849 #define setup_blocks_builder(shading, texturing, dithering, sw, target) \
1850 void setup_blocks_##shading##_##texturing##_##dithering##_##sw##_##target( \
1851 psx_gpu_struct *psx_gpu) \
1853 setup_blocks_load_msb_mask_##target(); \
1854 setup_blocks_variables_##shading##_##texturing(target); \
1856 edge_data_struct *span_edge_data = psx_gpu->span_edge_data; \
1857 vec_4x32u *span_uvrg_offset = psx_gpu->span_uvrg_offset; \
1858 u32 *span_b_offset = psx_gpu->span_b_offset; \
1860 block_struct *block = psx_gpu->blocks + psx_gpu->num_blocks; \
1862 u32 num_spans = psx_gpu->num_spans; \
1867 u32 num_blocks = psx_gpu->num_blocks; \
1868 u32 span_num_blocks; \
1872 span_num_blocks = span_edge_data->num_blocks; \
1873 if(span_num_blocks) \
1875 y = span_edge_data->y; \
1876 fb_ptr = psx_gpu->vram_out_ptr + span_edge_data->left_x + (y * 1024); \
1878 setup_blocks_span_initialize_##shading##_##texturing(); \
1879 setup_blocks_span_initialize_##dithering(texturing); \
1881 setup_blocks_add_blocks_##target(); \
1883 s32 pixel_span = span_num_blocks * 8; \
1884 pixel_span -= __builtin_popcount(span_edge_data->right_mask & 0xFF); \
1885 span_pixels += pixel_span; \
1887 span_num_blocks--; \
1888 while(span_num_blocks) \
1890 setup_blocks_store_##shading##_##texturing(sw, dithering, target, \
1892 setup_blocks_store_draw_mask_##texturing##_##target(block, 0x00); \
1896 span_num_blocks--; \
1899 setup_blocks_store_##shading##_##texturing(sw, dithering, target, edge); \
1900 setup_blocks_store_draw_mask_##texturing##_##target(block, \
1901 span_edge_data->right_mask); \
1907 zero_block_spans++; \
1912 span_uvrg_offset++; \
1916 psx_gpu->num_blocks = num_blocks; \
1920 //setup_blocks_builder(unshaded, untextured, undithered, unswizzled, direct);
1924 setup_blocks_builder(shaded, textured, dithered, swizzled, indirect);
1925 setup_blocks_builder(shaded, textured, dithered, unswizzled, indirect);
1927 setup_blocks_builder(unshaded, textured, dithered, unswizzled, indirect);
1928 setup_blocks_builder(unshaded, textured, dithered, swizzled, indirect);
1930 setup_blocks_builder(shaded, untextured, undithered, unswizzled, indirect);
1931 setup_blocks_builder(shaded, untextured, dithered, unswizzled, indirect);
1932 setup_blocks_builder(shaded, untextured, undithered, unswizzled, direct);
1933 setup_blocks_builder(shaded, untextured, dithered, unswizzled, direct);
1935 setup_blocks_builder(unshaded, untextured, undithered, unswizzled, indirect);
1936 setup_blocks_builder(unshaded, untextured, undithered, unswizzled, direct);
1938 void texture_blocks_untextured(psx_gpu_struct *psx_gpu)
1940 if(psx_gpu->primitive_type != PRIMITIVE_TYPE_SPRITE)
1941 texel_blocks_untextured += psx_gpu->num_blocks;
1944 void texture_blocks_4bpp(psx_gpu_struct *psx_gpu)
1946 block_struct *block = psx_gpu->blocks;
1947 u32 num_blocks = psx_gpu->num_blocks;
1948 texel_blocks_4bpp += num_blocks;
1950 vec_8x8u texels_low;
1951 vec_8x8u texels_high;
1958 vec_16x8u clut_high;
1960 u8 *texture_ptr_8bpp = psx_gpu->texture_page_ptr;
1961 u16 *clut_ptr = psx_gpu->clut_ptr;
1963 // Can be done with one deinterleaving load on NEON
1964 load_8x16b(clut_a, clut_ptr);
1965 load_8x16b(clut_b, clut_ptr + 8);
1966 unzip_16x8b(clut_low, clut_high, clut_a, clut_b);
1968 if(psx_gpu->current_texture_mask & psx_gpu->dirty_textures_4bpp_mask)
1969 update_texture_4bpp_cache(psx_gpu);
1973 texels.e[0] = texture_ptr_8bpp[block->uv.e[0]];
1974 texels.e[1] = texture_ptr_8bpp[block->uv.e[1]];
1975 texels.e[2] = texture_ptr_8bpp[block->uv.e[2]];
1976 texels.e[3] = texture_ptr_8bpp[block->uv.e[3]];
1977 texels.e[4] = texture_ptr_8bpp[block->uv.e[4]];
1978 texels.e[5] = texture_ptr_8bpp[block->uv.e[5]];
1979 texels.e[6] = texture_ptr_8bpp[block->uv.e[6]];
1980 texels.e[7] = texture_ptr_8bpp[block->uv.e[7]];
1982 tbl_16(texels_low, texels, clut_low);
1983 tbl_16(texels_high, texels, clut_high);
1985 // Can be done with an interleaving store on NEON
1986 zip_8x16b(pixels, texels_low, texels_high);
1988 block->texels = pixels;
1995 void texture_blocks_8bpp(psx_gpu_struct *psx_gpu)
1997 block_struct *block = psx_gpu->blocks;
1998 u32 num_blocks = psx_gpu->num_blocks;
2000 texel_blocks_8bpp += num_blocks;
2002 if(psx_gpu->current_texture_mask & psx_gpu->dirty_textures_8bpp_mask)
2003 update_texture_8bpp_cache(psx_gpu);
2006 u8 *texture_ptr_8bpp = psx_gpu->texture_page_ptr;
2014 for(i = 0; i < 8; i++)
2016 offset = block->uv.e[i];
2018 texel = texture_ptr_8bpp[offset];
2019 texels.e[i] = psx_gpu->clut_ptr[texel];
2022 block->texels = texels;
2029 void texture_blocks_16bpp(psx_gpu_struct *psx_gpu)
2031 block_struct *block = psx_gpu->blocks;
2032 u32 num_blocks = psx_gpu->num_blocks;
2034 texel_blocks_16bpp += num_blocks;
2038 u16 *texture_ptr_16bpp = psx_gpu->texture_page_ptr;
2044 for(i = 0; i < 8; i++)
2046 offset = block->uv.e[i];
2047 offset += ((offset & 0xFF00) * 3);
2049 texels.e[i] = texture_ptr_16bpp[offset];
2052 block->texels = texels;
2062 #define shade_blocks_load_msb_mask_indirect() \
2064 #define shade_blocks_load_msb_mask_direct() \
2065 vec_8x16u msb_mask; \
2066 dup_8x16b(msb_mask, psx_gpu->mask_msb); \
2068 #define shade_blocks_store_indirect(_draw_mask, _pixels) \
2069 block->draw_mask = _draw_mask; \
2070 block->pixels = _pixels \
2072 #define shade_blocks_store_direct(_draw_mask, _pixels) \
2074 vec_8x16u fb_pixels; \
2075 or_8x16b(_pixels, _pixels, msb_mask); \
2076 load_8x16b(fb_pixels, block->fb_ptr); \
2077 bif_8x16b(fb_pixels, _pixels, _draw_mask); \
2078 store_8x16b(fb_pixels, block->fb_ptr); \
2082 #define shade_blocks_textured_false_modulated_check_dithered(target) \
2083 if(psx_gpu->triangle_color == 0x808080) \
2085 false_modulated_blocks += num_blocks; \
2088 #define shade_blocks_textured_false_modulated_check_undithered(target) \
2089 if(psx_gpu->triangle_color == 0x808080) \
2092 shade_blocks_textured_unmodulated_##target(psx_gpu); \
2093 false_modulated_blocks += num_blocks; \
2098 #define shade_blocks_textured_modulated_shaded_primitive_load(dithering, \
2101 #define shade_blocks_textured_modulated_unshaded_primitive_load(dithering, \
2104 u32 color = psx_gpu->triangle_color; \
2105 dup_8x8b(colors_r, color); \
2106 dup_8x8b(colors_g, color >> 8); \
2107 dup_8x8b(colors_b, color >> 16); \
2108 shade_blocks_textured_false_modulated_check_##dithering(target); \
2111 #define shade_blocks_textured_modulated_shaded_block_load() \
2112 colors_r = block->r; \
2113 colors_g = block->g; \
2114 colors_b = block->b \
2116 #define shade_blocks_textured_modulated_unshaded_block_load() \
2118 #define shade_blocks_textured_modulate_dithered(component) \
2119 pixels_##component = block->dither_offsets; \
2120 mla_long_8x8b(pixels_##component, texels_##component, colors_##component) \
2122 #define shade_blocks_textured_modulate_undithered(component) \
2123 mul_long_8x8b(pixels_##component, texels_##component, colors_##component) \
2125 #define shade_blocks_textured_modulated_builder(shading, dithering, target) \
2126 void shade_blocks_##shading##_textured_modulated_##dithering##_##target( \
2127 psx_gpu_struct *psx_gpu) \
2129 block_struct *block = psx_gpu->blocks; \
2130 u32 num_blocks = psx_gpu->num_blocks; \
2133 vec_8x8u texels_r; \
2134 vec_8x8u texels_g; \
2135 vec_8x8u texels_b; \
2137 vec_8x8u colors_r; \
2138 vec_8x8u colors_g; \
2139 vec_8x8u colors_b; \
2141 vec_8x8u pixels_r_low; \
2142 vec_8x8u pixels_g_low; \
2143 vec_8x8u pixels_b_low; \
2146 vec_8x16u pixels_r; \
2147 vec_8x16u pixels_g; \
2148 vec_8x16u pixels_b; \
2150 vec_8x16u draw_mask; \
2151 vec_8x16u zero_mask; \
2153 vec_8x8u d64_0x07; \
2154 vec_8x8u d64_0x1F; \
2159 vec_8x16u d128_0x8000; \
2161 vec_8x16u test_mask = psx_gpu->test_mask; \
2162 u32 draw_mask_bits; \
2163 shade_blocks_load_msb_mask_##target(); \
2165 dup_8x8b(d64_0x07, 0x07); \
2166 dup_8x8b(d64_0x1F, 0x1F); \
2167 dup_8x8b(d64_1, 1); \
2168 dup_8x8b(d64_4, 4); \
2169 dup_8x8b(d64_128, 128); \
2171 dup_8x16b(d128_0x8000, 0x8000); \
2173 shade_blocks_textured_modulated_##shading##_primitive_load(dithering, \
2178 draw_mask_bits = block->draw_mask_bits; \
2179 dup_8x16b(draw_mask, draw_mask_bits); \
2180 tst_8x16b(draw_mask, draw_mask, test_mask); \
2182 shade_blocks_textured_modulated_##shading##_block_load(); \
2184 texels = block->texels; \
2186 mov_narrow_8x16b(texels_r, texels); \
2187 shr_narrow_8x16b(texels_g, texels, 5); \
2188 shr_narrow_8x16b(texels_b, texels, 7); \
2190 and_8x8b(texels_r, texels_r, d64_0x1F); \
2191 and_8x8b(texels_g, texels_g, d64_0x1F); \
2192 shr_8x8b(texels_b, texels_b, 3); \
2194 shade_blocks_textured_modulate_##dithering(r); \
2195 shade_blocks_textured_modulate_##dithering(g); \
2196 shade_blocks_textured_modulate_##dithering(b); \
2198 cmpeqz_8x16b(zero_mask, texels); \
2199 and_8x16b(pixels, texels, d128_0x8000); \
2201 shrq_narrow_signed_8x16b(pixels_r_low, pixels_r, 4); \
2202 shrq_narrow_signed_8x16b(pixels_g_low, pixels_g, 4); \
2203 shrq_narrow_signed_8x16b(pixels_b_low, pixels_b, 4); \
2205 or_8x16b(zero_mask, draw_mask, zero_mask); \
2207 shr_8x8b(pixels_r_low, pixels_r_low, 3); \
2208 bic_8x8b(pixels_g_low, pixels_g_low, d64_0x07); \
2209 bic_8x8b(pixels_b_low, pixels_b_low, d64_0x07); \
2211 mla_long_8x8b(pixels, pixels_r_low, d64_1); \
2212 mla_long_8x8b(pixels, pixels_g_low, d64_4); \
2213 mla_long_8x8b(pixels, pixels_b_low, d64_128); \
2215 shade_blocks_store_##target(zero_mask, pixels); \
2224 shade_blocks_textured_modulated_builder(shaded, dithered, direct);
2225 shade_blocks_textured_modulated_builder(shaded, undithered, direct);
2226 shade_blocks_textured_modulated_builder(unshaded, dithered, direct);
2227 shade_blocks_textured_modulated_builder(unshaded, undithered, direct);
2229 shade_blocks_textured_modulated_builder(shaded, dithered, indirect);
2230 shade_blocks_textured_modulated_builder(shaded, undithered, indirect);
2231 shade_blocks_textured_modulated_builder(unshaded, dithered, indirect);
2232 shade_blocks_textured_modulated_builder(unshaded, undithered, indirect);
2237 #define shade_blocks_textured_unmodulated_builder(target) \
2238 void shade_blocks_textured_unmodulated_##target(psx_gpu_struct *psx_gpu) \
2240 block_struct *block = psx_gpu->blocks; \
2241 u32 num_blocks = psx_gpu->num_blocks; \
2242 vec_8x16u draw_mask; \
2243 vec_8x16u test_mask = psx_gpu->test_mask; \
2244 u32 draw_mask_bits; \
2247 shade_blocks_load_msb_mask_##target(); \
2251 vec_8x16u zero_mask; \
2253 draw_mask_bits = block->draw_mask_bits; \
2254 dup_8x16b(draw_mask, draw_mask_bits); \
2255 tst_8x16b(draw_mask, draw_mask, test_mask); \
2257 pixels = block->texels; \
2259 cmpeqz_8x16b(zero_mask, pixels); \
2260 or_8x16b(zero_mask, draw_mask, zero_mask); \
2262 shade_blocks_store_##target(zero_mask, pixels); \
2269 #define shade_blocks_textured_unmodulated_dithered_builder(target) \
2270 void shade_blocks_textured_unmodulated_dithered_##target(psx_gpu_struct \
2273 block_struct *block = psx_gpu->blocks; \
2274 u32 num_blocks = psx_gpu->num_blocks; \
2275 vec_8x16u draw_mask; \
2276 vec_8x16u test_mask = psx_gpu->test_mask; \
2277 u32 draw_mask_bits; \
2280 shade_blocks_load_msb_mask_##target(); \
2284 vec_8x16u zero_mask; \
2286 draw_mask_bits = block->draw_mask_bits; \
2287 dup_8x16b(draw_mask, draw_mask_bits); \
2288 tst_8x16b(draw_mask, draw_mask, test_mask); \
2290 pixels = block->texels; \
2292 cmpeqz_8x16b(zero_mask, pixels); \
2293 or_8x16b(zero_mask, draw_mask, zero_mask); \
2295 shade_blocks_store_##target(zero_mask, pixels); \
2304 shade_blocks_textured_unmodulated_builder(indirect)
2305 shade_blocks_textured_unmodulated_builder(direct)
2307 void shade_blocks_unshaded_untextured_indirect(psx_gpu_struct *psx_gpu)
2311 void shade_blocks_unshaded_untextured_direct(psx_gpu_struct *psx_gpu)
2313 block_struct *block = psx_gpu->blocks;
2314 u32 num_blocks = psx_gpu->num_blocks;
2316 vec_8x16u pixels = block->pixels;
2317 shade_blocks_load_msb_mask_direct();
2321 shade_blocks_store_direct(block->draw_mask, pixels);
2330 void shade_blocks_shaded_untextured(psx_gpu_struct *psx_gpu)
2335 #define blend_blocks_mask_evaluate_on() \
2336 vec_8x16u mask_pixels; \
2337 cmpltz_8x16b(mask_pixels, framebuffer_pixels); \
2338 or_8x16b(draw_mask, draw_mask, mask_pixels) \
2340 #define blend_blocks_mask_evaluate_off() \
2342 #define blend_blocks_average() \
2344 vec_8x16u pixels_no_msb; \
2345 vec_8x16u fb_pixels_no_msb; \
2347 vec_8x16u d128_0x0421; \
2348 vec_8x16u d128_0x8000; \
2350 dup_8x16b(d128_0x0421, 0x0421); \
2351 dup_8x16b(d128_0x8000, 0x8000); \
2353 eor_8x16b(blend_pixels, pixels, framebuffer_pixels); \
2354 bic_8x16b(pixels_no_msb, pixels, d128_0x8000); \
2355 and_8x16b(blend_pixels, blend_pixels, d128_0x0421); \
2356 sub_8x16b(blend_pixels, pixels_no_msb, blend_pixels); \
2357 bic_8x16b(fb_pixels_no_msb, framebuffer_pixels, d128_0x8000); \
2358 average_8x16b(blend_pixels, fb_pixels_no_msb, blend_pixels); \
2361 #define blend_blocks_add() \
2363 vec_8x16u pixels_rb, pixels_g; \
2364 vec_8x16u fb_rb, fb_g; \
2366 vec_8x16u d128_0x7C1F; \
2367 vec_8x16u d128_0x03E0; \
2369 dup_8x16b(d128_0x7C1F, 0x7C1F); \
2370 dup_8x16b(d128_0x03E0, 0x03E0); \
2372 and_8x16b(pixels_rb, pixels, d128_0x7C1F); \
2373 and_8x16b(pixels_g, pixels, d128_0x03E0); \
2375 and_8x16b(fb_rb, framebuffer_pixels, d128_0x7C1F); \
2376 and_8x16b(fb_g, framebuffer_pixels, d128_0x03E0); \
2378 add_8x16b(fb_rb, fb_rb, pixels_rb); \
2379 add_8x16b(fb_g, fb_g, pixels_g); \
2381 min_16x8b(vector_cast(vec_16x8u, fb_rb), vector_cast(vec_16x8u, fb_rb), \
2382 vector_cast(vec_16x8u, d128_0x7C1F)); \
2383 min_8x16b(fb_g, fb_g, d128_0x03E0); \
2385 or_8x16b(blend_pixels, fb_rb, fb_g); \
2388 #define blend_blocks_subtract() \
2390 vec_8x16u pixels_rb, pixels_g; \
2391 vec_8x16u fb_rb, fb_g; \
2393 vec_8x16u d128_0x7C1F; \
2394 vec_8x16u d128_0x03E0; \
2396 dup_8x16b(d128_0x7C1F, 0x7C1F); \
2397 dup_8x16b(d128_0x03E0, 0x03E0); \
2399 and_8x16b(pixels_rb, pixels, d128_0x7C1F); \
2400 and_8x16b(pixels_g, pixels, d128_0x03E0); \
2402 and_8x16b(fb_rb, framebuffer_pixels, d128_0x7C1F); \
2403 and_8x16b(fb_g, framebuffer_pixels, d128_0x03E0); \
2405 subs_16x8b(vector_cast(vec_16x8u, fb_rb), \
2406 vector_cast(vec_16x8u, fb_rb), vector_cast(vec_16x8u, pixels_rb)); \
2407 subs_8x16b(fb_g, fb_g, pixels_g); \
2409 or_8x16b(blend_pixels, fb_rb, fb_g); \
2412 #define blend_blocks_add_fourth() \
2414 vec_8x16u pixels_rb, pixels_g; \
2415 vec_8x16u pixels_fourth; \
2416 vec_8x16u fb_rb, fb_g; \
2418 vec_8x16u d128_0x7C1F; \
2419 vec_8x16u d128_0x1C07; \
2420 vec_8x16u d128_0x03E0; \
2421 vec_8x16u d128_0x00E0; \
2423 dup_8x16b(d128_0x7C1F, 0x7C1F); \
2424 dup_8x16b(d128_0x1C07, 0x1C07); \
2425 dup_8x16b(d128_0x03E0, 0x03E0); \
2426 dup_8x16b(d128_0x00E0, 0x00E0); \
2428 shr_8x16b(pixels_fourth, vector_cast(vec_8x16s, pixels), 2); \
2430 and_8x16b(fb_rb, framebuffer_pixels, d128_0x7C1F); \
2431 and_8x16b(fb_g, framebuffer_pixels, d128_0x03E0); \
2433 and_8x16b(pixels_rb, pixels_fourth, d128_0x1C07); \
2434 and_8x16b(pixels_g, pixels_fourth, d128_0x00E0); \
2436 add_8x16b(fb_rb, fb_rb, pixels_rb); \
2437 add_8x16b(fb_g, fb_g, pixels_g); \
2439 min_16x8b(vector_cast(vec_16x8u, fb_rb), vector_cast(vec_16x8u, fb_rb), \
2440 vector_cast(vec_16x8u, d128_0x7C1F)); \
2441 min_8x16b(fb_g, fb_g, d128_0x03E0); \
2443 or_8x16b(blend_pixels, fb_rb, fb_g); \
2446 #define blend_blocks_blended_combine_textured() \
2448 vec_8x16u blend_mask; \
2449 cmpltz_8x16b(blend_mask, pixels); \
2451 or_immediate_8x16b(blend_pixels, blend_pixels, 0x8000); \
2452 bif_8x16b(blend_pixels, pixels, blend_mask); \
2455 #define blend_blocks_blended_combine_untextured() \
2458 #define blend_blocks_body_blend(blend_mode, texturing) \
2460 blend_blocks_##blend_mode(); \
2461 blend_blocks_blended_combine_##texturing(); \
2464 #define blend_blocks_body_average(texturing) \
2465 blend_blocks_body_blend(average, texturing) \
2467 #define blend_blocks_body_add(texturing) \
2468 blend_blocks_body_blend(add, texturing) \
2470 #define blend_blocks_body_subtract(texturing) \
2471 blend_blocks_body_blend(subtract, texturing) \
2473 #define blend_blocks_body_add_fourth(texturing) \
2474 blend_blocks_body_blend(add_fourth, texturing) \
2476 #define blend_blocks_body_unblended(texturing) \
2477 blend_pixels = pixels \
2480 #define blend_blocks_builder(texturing, blend_mode, mask_evaluate) \
2482 blend_blocks_##texturing##_##blend_mode##_##mask_evaluate(psx_gpu_struct \
2485 block_struct *block = psx_gpu->blocks; \
2486 u32 num_blocks = psx_gpu->num_blocks; \
2487 vec_8x16u draw_mask; \
2489 vec_8x16u blend_pixels; \
2490 vec_8x16u framebuffer_pixels; \
2491 vec_8x16u msb_mask; \
2495 dup_8x16b(msb_mask, psx_gpu->mask_msb); \
2499 pixels = block->pixels; \
2500 draw_mask = block->draw_mask; \
2501 fb_ptr = block->fb_ptr; \
2503 load_8x16b(framebuffer_pixels, fb_ptr); \
2505 blend_blocks_mask_evaluate_##mask_evaluate(); \
2506 blend_blocks_body_##blend_mode(texturing); \
2508 or_8x16b(blend_pixels, blend_pixels, msb_mask); \
2509 bif_8x16b(framebuffer_pixels, blend_pixels, draw_mask); \
2510 store_8x16b(framebuffer_pixels, fb_ptr); \
2520 void blend_blocks_textured_unblended_off(psx_gpu_struct *psx_gpu)
2524 blend_blocks_builder(textured, average, off);
2525 blend_blocks_builder(textured, average, on);
2526 blend_blocks_builder(textured, add, off);
2527 blend_blocks_builder(textured, add, on);
2528 blend_blocks_builder(textured, subtract, off);
2529 blend_blocks_builder(textured, subtract, on);
2530 blend_blocks_builder(textured, add_fourth, off);
2531 blend_blocks_builder(textured, add_fourth, on);
2533 blend_blocks_builder(untextured, average, off);
2534 blend_blocks_builder(untextured, average, on);
2535 blend_blocks_builder(untextured, add, off);
2536 blend_blocks_builder(untextured, add, on);
2537 blend_blocks_builder(untextured, subtract, off);
2538 blend_blocks_builder(untextured, subtract, on);
2539 blend_blocks_builder(untextured, add_fourth, off);
2540 blend_blocks_builder(untextured, add_fourth, on);
2542 blend_blocks_builder(textured, unblended, on);
2547 #define vertex_swap(_a, _b) \
2549 vertex_struct *temp_vertex = _a; \
2552 triangle_winding ^= 1; \
2556 // Setup blocks parametric-variables:
2557 // SHADE TEXTURE_MAP SWIZZLING
2564 // 8 inputs, 6 combinations
2566 #define setup_blocks_switch_untextured_unshaded(dithering, target) \
2567 setup_blocks_unshaded_untextured_undithered_unswizzled_##target \
2569 #define setup_blocks_switch_untextured_shaded(dithering, target) \
2570 setup_blocks_shaded_untextured_##dithering##_unswizzled_##target \
2572 #define setup_blocks_switch_untextured(shading, texture_mode, dithering, \
2574 setup_blocks_switch_untextured_##shading(dithering, target) \
2576 #define setup_blocks_switch_texture_mode_4bpp(shading) \
2577 setup_blocks_##shading##_textured_dithered_swizzled_indirect \
2579 #define setup_blocks_switch_texture_mode_8bpp(shading) \
2580 setup_blocks_##shading##_textured_dithered_swizzled_indirect \
2582 #define setup_blocks_switch_texture_mode_16bpp(shading) \
2583 setup_blocks_##shading##_textured_dithered_unswizzled_indirect \
2585 #define setup_blocks_switch_textured(shading, texture_mode, dithering, target) \
2586 setup_blocks_switch_texture_mode_##texture_mode(shading) \
2588 #define setup_blocks_switch_blended(shading, texturing, texture_mode, \
2589 dithering, mask_evaluate) \
2590 setup_blocks_switch_##texturing(shading, texture_mode, dithering, indirect) \
2592 #define setup_blocks_switch_unblended_on(shading, texturing, texture_mode, \
2594 setup_blocks_switch_##texturing(shading, texture_mode, dithering, indirect) \
2596 #define setup_blocks_switch_unblended_off(shading, texturing, texture_mode, \
2598 setup_blocks_switch_##texturing(shading, texture_mode, dithering, direct) \
2600 #define setup_blocks_switch_unblended(shading, texturing, texture_mode, \
2601 dithering, mask_evaluate) \
2602 setup_blocks_switch_unblended_##mask_evaluate(shading, texturing, \
2603 texture_mode, dithering) \
2605 #define setup_blocks_switch(shading, texturing, texture_mode, dithering, \
2606 blending, mask_evaluate) \
2607 setup_blocks_switch_##blending(shading, texturing, texture_mode, \
2608 dithering, mask_evaluate) \
2613 #define texture_blocks_switch_untextured(texture_mode) \
2614 texture_blocks_untextured \
2616 #define texture_blocks_switch_textured(texture_mode) \
2617 texture_blocks_##texture_mode \
2619 #define texture_blocks_switch(texturing, texture_mode) \
2620 texture_blocks_switch_##texturing(texture_mode) \
2623 // Shade blocks parametric-variables:
2624 // SHADE TEXTURE_MAP MODULATE_TEXELS dither_mode
2633 // 16 inputs, 8 combinations
2635 #define shade_blocks_switch_unshaded_untextured(modulation, dithering, target) \
2636 shade_blocks_unshaded_untextured_##target \
2638 #define shade_blocks_switch_unshaded_textured_unmodulated(dithering, target) \
2639 shade_blocks_textured_unmodulated_##target \
2641 #define shade_blocks_switch_unshaded_textured_modulated(dithering, target) \
2642 shade_blocks_unshaded_textured_modulated_##dithering##_##target \
2644 #define shade_blocks_switch_unshaded_textured(modulation, dithering, target) \
2645 shade_blocks_switch_unshaded_textured_##modulation(dithering, target) \
2647 #define shade_blocks_switch_unshaded(texturing, modulation, dithering, target) \
2648 shade_blocks_switch_unshaded_##texturing(modulation, dithering, target) \
2650 #define shade_blocks_switch_shaded_untextured(modulation, dithering, target) \
2651 shade_blocks_shaded_untextured \
2653 #define shade_blocks_switch_shaded_textured_unmodulated(dithering, target) \
2654 shade_blocks_textured_unmodulated_##target \
2656 #define shade_blocks_switch_shaded_textured_modulated(dithering, target) \
2657 shade_blocks_shaded_textured_modulated_##dithering##_##target \
2659 #define shade_blocks_switch_shaded_textured(modulation, dithering, target) \
2660 shade_blocks_switch_shaded_textured_##modulation(dithering, target) \
2662 #define shade_blocks_switch_shaded(texturing, modulation, dithering, target) \
2663 shade_blocks_switch_shaded_##texturing(modulation, dithering, target) \
2665 #define shade_blocks_switch_mask_off(shading, texturing, modulation, \
2667 shade_blocks_switch_##shading(texturing, modulation, dithering, direct) \
2669 #define shade_blocks_switch_mask_on(shading, texturing, modulation, \
2671 shade_blocks_switch_##shading(texturing, modulation, dithering, indirect) \
2673 #define shade_blocks_switch_blended(shading, texturing, modulation, dithering, \
2675 shade_blocks_switch_##shading(texturing, modulation, dithering, indirect) \
2677 #define shade_blocks_switch_unblended(shading, texturing, modulation, \
2678 dithering, mask_evaluate) \
2679 shade_blocks_switch_mask_##mask_evaluate(shading, texturing, modulation, \
2682 #define shade_blocks_switch(shading, texturing, modulation, dithering, \
2683 blending, mask_evaluate) \
2684 shade_blocks_switch_##blending(shading, texturing, modulation, dithering, \
2688 // Blend blocks parametric-variables:
2689 // TEXTURE_MAP BLEND BM_A BM_B mask_evaluate
2708 // 32 inputs, 18 combinations
2710 #define blend_blocks_switch_unblended(texturing, blend_mode, mask_evaluate) \
2711 blend_blocks_textured_unblended_##mask_evaluate \
2713 #define blend_blocks_switch_blended(texturing, blend_mode, mask_evaluate) \
2714 blend_blocks_##texturing##_##blend_mode##_##mask_evaluate \
2716 #define blend_blocks_switch(texturing, blending, blend_mode, mask_evaluate) \
2717 blend_blocks_switch_##blending(texturing, blend_mode, mask_evaluate) \
2720 #define render_blocks_switch_block_modulation(texture_mode, blend_mode, \
2721 mask_evaluate, shading, dithering, texturing, blending, modulation) \
2723 setup_blocks_switch(shading, texturing, texture_mode, dithering, blending, \
2725 texture_blocks_switch(texturing, texture_mode), \
2726 shade_blocks_switch(shading, texturing, modulation, dithering, blending, \
2728 blend_blocks_switch(texturing, blending, blend_mode, mask_evaluate) \
2731 #define render_blocks_switch_block_blending(texture_mode, blend_mode, \
2732 mask_evaluate, shading, dithering, texturing, blending) \
2733 render_blocks_switch_block_modulation(texture_mode, blend_mode, \
2734 mask_evaluate, shading, dithering, texturing, blending, modulated), \
2735 render_blocks_switch_block_modulation(texture_mode, blend_mode, \
2736 mask_evaluate, shading, dithering, texturing, blending, unmodulated) \
2738 #define render_blocks_switch_block_texturing(texture_mode, blend_mode, \
2739 mask_evaluate, shading, dithering, texturing) \
2740 render_blocks_switch_block_blending(texture_mode, blend_mode, \
2741 mask_evaluate, shading, dithering, texturing, unblended), \
2742 render_blocks_switch_block_blending(texture_mode, blend_mode, \
2743 mask_evaluate, shading, dithering, texturing, blended) \
2745 #define render_blocks_switch_block_dithering(texture_mode, blend_mode, \
2746 mask_evaluate, shading, dithering) \
2747 render_blocks_switch_block_texturing(texture_mode, blend_mode, \
2748 mask_evaluate, shading, dithering, untextured), \
2749 render_blocks_switch_block_texturing(texture_mode, blend_mode, \
2750 mask_evaluate, shading, dithering, textured) \
2752 #define render_blocks_switch_block_shading(texture_mode, blend_mode, \
2753 mask_evaluate, shading) \
2754 render_blocks_switch_block_dithering(texture_mode, blend_mode, \
2755 mask_evaluate, shading, undithered), \
2756 render_blocks_switch_block_dithering(texture_mode, blend_mode, \
2757 mask_evaluate, shading, dithered) \
2759 #define render_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, \
2761 render_blocks_switch_block_shading(texture_mode, blend_mode, mask_evaluate, \
2763 render_blocks_switch_block_shading(texture_mode, blend_mode, mask_evaluate, \
2766 #define render_blocks_switch_block_blend_mode(texture_mode, blend_mode) \
2767 render_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, off), \
2768 render_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, on) \
2770 #define render_blocks_switch_block_texture_mode(texture_mode) \
2771 render_blocks_switch_block_blend_mode(texture_mode, average), \
2772 render_blocks_switch_block_blend_mode(texture_mode, add), \
2773 render_blocks_switch_block_blend_mode(texture_mode, subtract), \
2774 render_blocks_switch_block_blend_mode(texture_mode, add_fourth) \
2776 #define render_blocks_switch_block() \
2777 render_blocks_switch_block_texture_mode(4bpp), \
2778 render_blocks_switch_block_texture_mode(8bpp), \
2779 render_blocks_switch_block_texture_mode(16bpp), \
2780 render_blocks_switch_block_texture_mode(16bpp) \
2783 render_block_handler_struct render_triangle_block_handlers[] =
2785 render_blocks_switch_block()
2788 #undef render_blocks_switch_block_modulation
2790 #define render_blocks_switch_block_modulation(texture_mode, blend_mode, \
2791 mask_evaluate, shading, dithering, texturing, blending, modulation) \
2793 "texture mode: " #texture_mode "\n" \
2794 "blend mode: " #blend_mode "\n" \
2795 "mask evaluation: " #mask_evaluate "\n" \
2802 char *render_block_flag_strings[] =
2804 render_blocks_switch_block()
2808 #define triangle_y_direction_up 1
2809 #define triangle_y_direction_flat 2
2810 #define triangle_y_direction_down 0
2812 #define triangle_winding_positive 0
2813 #define triangle_winding_negative 1
2815 #define triangle_set_direction(direction_variable, value) \
2816 u32 direction_variable = (u32)(value) >> 31; \
2818 direction_variable = 2 \
2820 #define triangle_case(direction_a, direction_b, direction_c, winding) \
2821 case (triangle_y_direction_##direction_a | \
2822 (triangle_y_direction_##direction_b << 2) | \
2823 (triangle_y_direction_##direction_c << 4) | \
2824 (triangle_winding_##winding << 6)) \
2826 static int prepare_triangle(psx_gpu_struct *psx_gpu, vertex_struct *vertexes,
2827 vertex_struct *vertexes_out[3])
2829 s32 y_top, y_bottom;
2831 u32 triangle_winding = 0;
2833 vertex_struct *a = &(vertexes[0]);
2834 vertex_struct *b = &(vertexes[1]);
2835 vertex_struct *c = &(vertexes[2]);
2837 triangle_area = triangle_signed_area_x2(a->x, a->y, b->x, b->y, c->x, c->y);
2843 if(triangle_area == 0)
2865 if((y_bottom - y_top) >= 512)
2873 if(triangle_area < 0)
2875 triangle_area = -triangle_area;
2876 triangle_winding ^= 1;
2891 if((c->x - psx_gpu->offset_x) >= 1024 || (c->x - a->x) >= 1024)
2899 if(invalidate_texture_cache_region_viewport(psx_gpu, a->x, y_top, c->x,
2908 psx_gpu->triangle_area = triangle_area;
2909 psx_gpu->triangle_winding = triangle_winding;
2911 vertexes_out[0] = a;
2912 vertexes_out[1] = b;
2913 vertexes_out[2] = c;
2918 static void render_triangle_p(psx_gpu_struct *psx_gpu,
2919 vertex_struct *vertex_ptrs[3], u32 flags)
2921 psx_gpu->num_spans = 0;
2923 vertex_struct *a = vertex_ptrs[0];
2924 vertex_struct *b = vertex_ptrs[1];
2925 vertex_struct *c = vertex_ptrs[2];
2927 s32 y_delta_a = b->y - a->y;
2928 s32 y_delta_b = c->y - b->y;
2929 s32 y_delta_c = c->y - a->y;
2931 triangle_set_direction(y_direction_a, y_delta_a);
2932 triangle_set_direction(y_direction_b, y_delta_b);
2933 triangle_set_direction(y_direction_c, y_delta_c);
2935 compute_all_gradients(psx_gpu, a, b, c);
2937 switch(y_direction_a | (y_direction_b << 2) | (y_direction_c << 4) |
2938 (psx_gpu->triangle_winding << 6))
2940 triangle_case(up, up, up, negative):
2941 triangle_case(up, up, flat, negative):
2942 triangle_case(up, up, down, negative):
2943 setup_spans_up_right(psx_gpu, a, b, c);
2946 triangle_case(flat, up, up, negative):
2947 triangle_case(flat, up, flat, negative):
2948 triangle_case(flat, up, down, negative):
2949 setup_spans_up_a(psx_gpu, a, b, c);
2952 triangle_case(down, up, up, negative):
2953 setup_spans_up_down(psx_gpu, a, c, b);
2956 triangle_case(down, up, flat, negative):
2957 setup_spans_down_a(psx_gpu, a, c, b);
2960 triangle_case(down, up, down, negative):
2961 setup_spans_down_right(psx_gpu, a, c, b);
2964 triangle_case(down, flat, up, negative):
2965 triangle_case(down, flat, flat, negative):
2966 triangle_case(down, flat, down, negative):
2967 setup_spans_down_b(psx_gpu, a, b, c);
2970 triangle_case(down, down, up, negative):
2971 triangle_case(down, down, flat, negative):
2972 triangle_case(down, down, down, negative):
2973 setup_spans_down_left(psx_gpu, a, b, c);
2976 triangle_case(up, up, up, positive):
2977 triangle_case(up, up, flat, positive):
2978 triangle_case(up, up, down, positive):
2979 setup_spans_up_left(psx_gpu, a, b, c);
2982 triangle_case(up, flat, up, positive):
2983 triangle_case(up, flat, flat, positive):
2984 triangle_case(up, flat, down, positive):
2985 setup_spans_up_b(psx_gpu, a, b, c);
2988 triangle_case(up, down, up, positive):
2989 setup_spans_up_right(psx_gpu, a, c, b);
2992 triangle_case(up, down, flat, positive):
2993 setup_spans_up_a(psx_gpu, a, c, b);
2996 triangle_case(up, down, down, positive):
2997 setup_spans_up_down(psx_gpu, a, b, c);
3000 triangle_case(flat, down, up, positive):
3001 triangle_case(flat, down, flat, positive):
3002 triangle_case(flat, down, down, positive):
3003 setup_spans_down_a(psx_gpu, a, b, c);
3006 triangle_case(down, down, up, positive):
3007 triangle_case(down, down, flat, positive):
3008 triangle_case(down, down, down, positive):
3009 setup_spans_down_right(psx_gpu, a, b, c);
3014 spans += psx_gpu->num_spans;
3017 if(unlikely(psx_gpu->render_mode & RENDER_INTERLACE_ENABLED))
3021 if(psx_gpu->render_mode & RENDER_INTERLACE_ODD)
3023 for(i = 0; i < psx_gpu->num_spans; i++)
3025 if((psx_gpu->span_edge_data[i].y & 1) == 0)
3026 psx_gpu->span_edge_data[i].num_blocks = 0;
3031 for(i = 0; i < psx_gpu->num_spans; i++)
3033 if(psx_gpu->span_edge_data[i].y & 1)
3034 psx_gpu->span_edge_data[i].num_blocks = 0;
3039 u32 render_state = flags &
3040 (RENDER_FLAGS_MODULATE_TEXELS | RENDER_FLAGS_BLEND |
3041 RENDER_FLAGS_TEXTURE_MAP | RENDER_FLAGS_SHADE);
3042 render_state |= psx_gpu->render_state_base;
3044 if((psx_gpu->render_state != render_state) ||
3045 (psx_gpu->primitive_type != PRIMITIVE_TYPE_TRIANGLE))
3047 psx_gpu->render_state = render_state;
3048 flush_render_block_buffer(psx_gpu);
3054 psx_gpu->primitive_type = PRIMITIVE_TYPE_TRIANGLE;
3056 psx_gpu->render_block_handler =
3057 &(render_triangle_block_handlers[render_state]);
3058 ((setup_blocks_function_type *)psx_gpu->render_block_handler->setup_blocks)
3062 void render_triangle(psx_gpu_struct *psx_gpu, vertex_struct *vertexes,
3065 vertex_struct *vertex_ptrs[3];
3066 if (prepare_triangle(psx_gpu, vertexes, vertex_ptrs))
3067 render_triangle_p(psx_gpu, vertex_ptrs, flags);
3072 void texture_sprite_blocks_8bpp(psx_gpu_struct *psx_gpu)
3074 block_struct *block = psx_gpu->blocks;
3075 u32 num_blocks = psx_gpu->num_blocks;
3078 vec_8x8u texel_indexes;
3080 u16 *clut_ptr = psx_gpu->clut_ptr;
3085 texel_indexes = block->r;
3087 for(i = 0; i < 8; i++)
3089 texels.e[i] = clut_ptr[texel_indexes.e[i]];
3092 block->texels = texels;
3102 #define setup_sprite_tiled_initialize_4bpp_clut() \
3103 u16 *clut_ptr = psx_gpu->clut_ptr; \
3104 vec_8x16u clut_a, clut_b; \
3105 vec_16x8u clut_low, clut_high; \
3107 load_8x16b(clut_a, clut_ptr); \
3108 load_8x16b(clut_b, clut_ptr + 8); \
3109 unzip_16x8b(clut_low, clut_high, clut_a, clut_b) \
3111 #define setup_sprite_tiled_initialize_4bpp() \
3112 setup_sprite_tiled_initialize_4bpp_clut(); \
3114 if(psx_gpu->current_texture_mask & psx_gpu->dirty_textures_4bpp_mask) \
3115 update_texture_4bpp_cache(psx_gpu) \
3117 #define setup_sprite_tiled_initialize_8bpp() \
3118 if(psx_gpu->current_texture_mask & psx_gpu->dirty_textures_8bpp_mask) \
3119 update_texture_8bpp_cache(psx_gpu) \
3122 #define setup_sprite_tile_fetch_texel_block_8bpp(offset) \
3123 texture_block_ptr = psx_gpu->texture_page_ptr + \
3124 ((texture_offset + offset) & texture_mask); \
3126 load_64b(texels, texture_block_ptr) \
3129 #define setup_sprite_tile_add_blocks(tile_num_blocks) \
3130 num_blocks += tile_num_blocks; \
3131 sprite_blocks += tile_num_blocks; \
3133 if(num_blocks > MAX_BLOCKS) \
3135 flush_render_block_buffer(psx_gpu); \
3136 num_blocks = tile_num_blocks; \
3137 block = psx_gpu->blocks; \
3140 #define setup_sprite_tile_full_4bpp(edge) \
3142 vec_8x8u texels_low, texels_high; \
3144 setup_sprite_tile_add_blocks(sub_tile_height * 2); \
3146 while(sub_tile_height) \
3148 setup_sprite_tile_fetch_texel_block_8bpp(0); \
3149 tbl_16(texels_low, texels, clut_low); \
3150 tbl_16(texels_high, texels, clut_high); \
3151 zip_8x16b(pixels, texels_low, texels_high); \
3153 block->texels = pixels; \
3154 block->draw_mask_bits = left_mask_bits; \
3155 block->fb_ptr = fb_ptr; \
3158 setup_sprite_tile_fetch_texel_block_8bpp(8); \
3159 tbl_16(texels_low, texels, clut_low); \
3160 tbl_16(texels_high, texels, clut_high); \
3161 zip_8x16b(pixels, texels_low, texels_high); \
3163 block->texels = pixels; \
3164 block->draw_mask_bits = right_mask_bits; \
3165 block->fb_ptr = fb_ptr + 8; \
3169 texture_offset += 0x10; \
3170 sub_tile_height--; \
3172 texture_offset += 0xF00; \
3173 psx_gpu->num_blocks = num_blocks; \
3176 #define setup_sprite_tile_half_4bpp(edge) \
3178 vec_8x8u texels_low, texels_high; \
3180 setup_sprite_tile_add_blocks(sub_tile_height); \
3182 while(sub_tile_height) \
3184 setup_sprite_tile_fetch_texel_block_8bpp(0); \
3185 tbl_16(texels_low, texels, clut_low); \
3186 tbl_16(texels_high, texels, clut_high); \
3187 zip_8x16b(pixels, texels_low, texels_high); \
3189 block->texels = pixels; \
3190 block->draw_mask_bits = edge##_mask_bits; \
3191 block->fb_ptr = fb_ptr; \
3195 texture_offset += 0x10; \
3196 sub_tile_height--; \
3198 texture_offset += 0xF00; \
3199 psx_gpu->num_blocks = num_blocks; \
3203 #define setup_sprite_tile_full_8bpp(edge) \
3205 setup_sprite_tile_add_blocks(sub_tile_height * 2); \
3207 while(sub_tile_height) \
3209 setup_sprite_tile_fetch_texel_block_8bpp(0); \
3210 block->r = texels; \
3211 block->draw_mask_bits = left_mask_bits; \
3212 block->fb_ptr = fb_ptr; \
3215 setup_sprite_tile_fetch_texel_block_8bpp(8); \
3216 block->r = texels; \
3217 block->draw_mask_bits = right_mask_bits; \
3218 block->fb_ptr = fb_ptr + 8; \
3222 texture_offset += 0x10; \
3223 sub_tile_height--; \
3225 texture_offset += 0xF00; \
3226 psx_gpu->num_blocks = num_blocks; \
3229 #define setup_sprite_tile_half_8bpp(edge) \
3231 setup_sprite_tile_add_blocks(sub_tile_height * 2); \
3233 while(sub_tile_height) \
3235 setup_sprite_tile_fetch_texel_block_8bpp(0); \
3236 block->r = texels; \
3237 block->draw_mask_bits = edge##_mask_bits; \
3238 block->fb_ptr = fb_ptr; \
3242 texture_offset += 0x10; \
3243 sub_tile_height--; \
3245 texture_offset += 0xF00; \
3246 psx_gpu->num_blocks = num_blocks; \
3250 #define setup_sprite_tile_column_edge_pre_adjust_half_right() \
3251 texture_offset = texture_offset_base + 8; \
3254 #define setup_sprite_tile_column_edge_pre_adjust_half_left() \
3255 texture_offset = texture_offset_base \
3257 #define setup_sprite_tile_column_edge_pre_adjust_half(edge) \
3258 setup_sprite_tile_column_edge_pre_adjust_half_##edge() \
3260 #define setup_sprite_tile_column_edge_pre_adjust_full(edge) \
3261 texture_offset = texture_offset_base \
3263 #define setup_sprite_tile_column_edge_post_adjust_half_right() \
3266 #define setup_sprite_tile_column_edge_post_adjust_half_left() \
3268 #define setup_sprite_tile_column_edge_post_adjust_half(edge) \
3269 setup_sprite_tile_column_edge_post_adjust_half_##edge() \
3271 #define setup_sprite_tile_column_edge_post_adjust_full(edge) \
3274 #define setup_sprite_tile_column_height_single(edge_mode, edge, texture_mode, \
3278 sub_tile_height = column_data; \
3279 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
3280 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
3281 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge); \
3284 #define setup_sprite_tile_column_height_multi(edge_mode, edge, texture_mode, \
3288 u32 tiles_remaining = column_data >> 16; \
3289 sub_tile_height = column_data & 0xFF; \
3290 setup_sprite_tile_column_edge_pre_adjust_##edge_mode##x4mode(edge); \
3291 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
3292 tiles_remaining -= 1; \
3294 while(tiles_remaining) \
3296 sub_tile_height = 16; \
3297 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
3298 tiles_remaining--; \
3301 sub_tile_height = (column_data >> 8) & 0xFF; \
3302 setup_sprite_tile_##edge_mode##_##texture_mode##x4mode(edge); \
3303 setup_sprite_tile_column_edge_post_adjust_##edge_mode##x4mode(edge); \
3307 #define setup_sprite_column_data_single() \
3308 column_data = height \
3310 #define setup_sprite_column_data_multi() \
3311 column_data = 16 - offset_v; \
3312 column_data |= ((height_rounded & 0xF) + 1) << 8; \
3313 column_data |= (tile_height - 1) << 16 \
3316 #define RIGHT_MASK_BIT_SHIFT 8
3317 #define RIGHT_MASK_BIT_SHIFT_4x 16
3319 #define setup_sprite_tile_column_width_single(texture_mode, multi_height, \
3320 edge_mode, edge, x4mode) \
3322 setup_sprite_column_data_##multi_height(); \
3323 left_mask_bits = left_block_mask | right_block_mask; \
3324 right_mask_bits = left_mask_bits >> RIGHT_MASK_BIT_SHIFT##x4mode; \
3326 setup_sprite_tile_column_height_##multi_height(edge_mode, edge, \
3327 texture_mode, x4mode); \
3330 #define setup_sprite_tiled_advance_column() \
3331 texture_offset_base += 0x100; \
3332 if((texture_offset_base & 0xF00) == 0) \
3333 texture_offset_base -= (0x100 + 0xF00) \
3335 #define FB_PTR_MULTIPLIER 1
3336 #define FB_PTR_MULTIPLIER_4x 2
3338 #define setup_sprite_tile_column_width_multi(texture_mode, multi_height, \
3339 left_mode, right_mode, x4mode) \
3341 setup_sprite_column_data_##multi_height(); \
3342 s32 fb_ptr_advance_column = (16 - (1024 * height)) \
3343 * FB_PTR_MULTIPLIER##x4mode; \
3346 left_mask_bits = left_block_mask; \
3347 right_mask_bits = left_mask_bits >> RIGHT_MASK_BIT_SHIFT##x4mode; \
3349 setup_sprite_tile_column_height_##multi_height(left_mode, right, \
3350 texture_mode, x4mode); \
3351 fb_ptr += fb_ptr_advance_column; \
3353 left_mask_bits = 0x00; \
3354 right_mask_bits = 0x00; \
3358 setup_sprite_tiled_advance_column(); \
3359 setup_sprite_tile_column_height_##multi_height(full, none, \
3360 texture_mode, x4mode); \
3361 fb_ptr += fb_ptr_advance_column; \
3365 left_mask_bits = right_block_mask; \
3366 right_mask_bits = left_mask_bits >> RIGHT_MASK_BIT_SHIFT##x4mode; \
3368 setup_sprite_tiled_advance_column(); \
3369 setup_sprite_tile_column_height_##multi_height(right_mode, left, \
3370 texture_mode, x4mode); \
3375 #define setup_sprite_tiled_initialize_4bpp_4x() \
3376 setup_sprite_tiled_initialize_4bpp_clut() \
3378 #define setup_sprite_tiled_initialize_8bpp_4x() \
3381 #define setup_sprite_tile_full_4bpp_4x(edge) \
3383 vec_8x8u texels_low, texels_high; \
3384 vec_8x16u pixels, pixels_wide; \
3385 setup_sprite_tile_add_blocks(sub_tile_height * 2 * 4); \
3386 u32 left_mask_bits_a = left_mask_bits & 0xFF; \
3387 u32 left_mask_bits_b = left_mask_bits >> 8; \
3388 u32 right_mask_bits_a = right_mask_bits & 0xFF; \
3389 u32 right_mask_bits_b = right_mask_bits >> 8; \
3391 while(sub_tile_height) \
3393 setup_sprite_tile_fetch_texel_block_8bpp(0); \
3394 tbl_16(texels_low, texels, clut_low); \
3395 tbl_16(texels_high, texels, clut_high); \
3396 zip_8x16b(pixels, texels_low, texels_high); \
3398 zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.low, pixels.low); \
3399 block->texels = pixels_wide; \
3400 block->draw_mask_bits = left_mask_bits_a; \
3401 block->fb_ptr = fb_ptr; \
3404 block->texels = pixels_wide; \
3405 block->draw_mask_bits = left_mask_bits_a; \
3406 block->fb_ptr = fb_ptr + 1024; \
3409 zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.high, pixels.high); \
3410 block->texels = pixels_wide; \
3411 block->draw_mask_bits = left_mask_bits_b; \
3412 block->fb_ptr = fb_ptr + 8; \
3415 block->texels = pixels_wide; \
3416 block->draw_mask_bits = left_mask_bits_b; \
3417 block->fb_ptr = fb_ptr + 1024 + 8; \
3420 setup_sprite_tile_fetch_texel_block_8bpp(8); \
3421 tbl_16(texels_low, texels, clut_low); \
3422 tbl_16(texels_high, texels, clut_high); \
3423 zip_8x16b(pixels, texels_low, texels_high); \
3425 zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.low, pixels.low); \
3426 block->texels = pixels_wide; \
3427 block->draw_mask_bits = right_mask_bits_a; \
3428 block->fb_ptr = fb_ptr + 16; \
3431 block->texels = pixels_wide; \
3432 block->draw_mask_bits = right_mask_bits_a; \
3433 block->fb_ptr = fb_ptr + 1024 + 16; \
3436 zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.high, pixels.high); \
3437 block->texels = pixels_wide; \
3438 block->draw_mask_bits = right_mask_bits_b; \
3439 block->fb_ptr = fb_ptr + 24; \
3442 block->texels = pixels_wide; \
3443 block->draw_mask_bits = right_mask_bits_b; \
3444 block->fb_ptr = fb_ptr + 1024 + 24; \
3448 texture_offset += 0x10; \
3449 sub_tile_height--; \
3451 texture_offset += 0xF00; \
3452 psx_gpu->num_blocks = num_blocks; \
3455 #define setup_sprite_tile_half_4bpp_4x(edge) \
3457 vec_8x8u texels_low, texels_high; \
3458 vec_8x16u pixels, pixels_wide; \
3459 setup_sprite_tile_add_blocks(sub_tile_height * 4); \
3460 u32 edge##_mask_bits_a = edge##_mask_bits & 0xFF; \
3461 u32 edge##_mask_bits_b = edge##_mask_bits >> 8; \
3463 while(sub_tile_height) \
3465 setup_sprite_tile_fetch_texel_block_8bpp(0); \
3466 tbl_16(texels_low, texels, clut_low); \
3467 tbl_16(texels_high, texels, clut_high); \
3468 zip_8x16b(pixels, texels_low, texels_high); \
3470 zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.low, pixels.low); \
3471 block->texels = pixels_wide; \
3472 block->draw_mask_bits = edge##_mask_bits_a; \
3473 block->fb_ptr = fb_ptr; \
3476 block->texels = pixels_wide; \
3477 block->draw_mask_bits = edge##_mask_bits_a; \
3478 block->fb_ptr = fb_ptr + 1024; \
3481 zip_4x32b(vector_cast(vec_4x32u, pixels_wide), pixels.high, pixels.high); \
3482 block->texels = pixels_wide; \
3483 block->draw_mask_bits = edge##_mask_bits_b; \
3484 block->fb_ptr = fb_ptr + 8; \
3487 block->texels = pixels_wide; \
3488 block->draw_mask_bits = edge##_mask_bits_b; \
3489 block->fb_ptr = fb_ptr + 1024 + 8; \
3493 texture_offset += 0x10; \
3494 sub_tile_height--; \
3496 texture_offset += 0xF00; \
3497 psx_gpu->num_blocks = num_blocks; \
3501 #define setup_sprite_tile_full_8bpp_4x(edge) \
3503 setup_sprite_tile_add_blocks(sub_tile_height * 2 * 4); \
3504 vec_16x8u texels_wide; \
3505 u32 left_mask_bits_a = left_mask_bits & 0xFF; \
3506 u32 left_mask_bits_b = left_mask_bits >> 8; \
3507 u32 right_mask_bits_a = right_mask_bits & 0xFF; \
3508 u32 right_mask_bits_b = right_mask_bits >> 8; \
3510 while(sub_tile_height) \
3512 setup_sprite_tile_fetch_texel_block_8bpp(0); \
3513 zip_8x16b(vector_cast(vec_8x16u, texels_wide), texels, texels); \
3514 block->r = texels_wide.low; \
3515 block->draw_mask_bits = left_mask_bits_a; \
3516 block->fb_ptr = fb_ptr; \
3519 block->r = texels_wide.low; \
3520 block->draw_mask_bits = left_mask_bits_a; \
3521 block->fb_ptr = fb_ptr + 1024; \
3524 block->r = texels_wide.high; \
3525 block->draw_mask_bits = left_mask_bits_b; \
3526 block->fb_ptr = fb_ptr + 8; \
3529 block->r = texels_wide.high; \
3530 block->draw_mask_bits = left_mask_bits_b; \
3531 block->fb_ptr = fb_ptr + 1024 + 8; \
3534 setup_sprite_tile_fetch_texel_block_8bpp(8); \
3535 zip_8x16b(vector_cast(vec_8x16u, texels_wide), texels, texels); \
3536 block->r = texels_wide.low; \
3537 block->draw_mask_bits = right_mask_bits_a; \
3538 block->fb_ptr = fb_ptr + 16; \
3541 block->r = texels_wide.low; \
3542 block->draw_mask_bits = right_mask_bits_a; \
3543 block->fb_ptr = fb_ptr + 1024 + 16; \
3546 block->r = texels_wide.high; \
3547 block->draw_mask_bits = right_mask_bits_b; \
3548 block->fb_ptr = fb_ptr + 24; \
3551 block->r = texels_wide.high; \
3552 block->draw_mask_bits = right_mask_bits_b; \
3553 block->fb_ptr = fb_ptr + 24 + 1024; \
3557 texture_offset += 0x10; \
3558 sub_tile_height--; \
3560 texture_offset += 0xF00; \
3561 psx_gpu->num_blocks = num_blocks; \
3564 #define setup_sprite_tile_half_8bpp_4x(edge) \
3566 setup_sprite_tile_add_blocks(sub_tile_height * 4); \
3567 vec_16x8u texels_wide; \
3568 u32 edge##_mask_bits_a = edge##_mask_bits & 0xFF; \
3569 u32 edge##_mask_bits_b = edge##_mask_bits >> 8; \
3571 while(sub_tile_height) \
3573 setup_sprite_tile_fetch_texel_block_8bpp(0); \
3574 zip_8x16b(vector_cast(vec_8x16u, texels_wide), texels, texels); \
3575 block->r = texels_wide.low; \
3576 block->draw_mask_bits = edge##_mask_bits_a; \
3577 block->fb_ptr = fb_ptr; \
3580 block->r = texels_wide.low; \
3581 block->draw_mask_bits = edge##_mask_bits_a; \
3582 block->fb_ptr = fb_ptr + 1024; \
3585 block->r = texels_wide.high; \
3586 block->draw_mask_bits = edge##_mask_bits_b; \
3587 block->fb_ptr = fb_ptr + 8; \
3590 block->r = texels_wide.high; \
3591 block->draw_mask_bits = edge##_mask_bits_b; \
3592 block->fb_ptr = fb_ptr + 8 + 1024; \
3596 texture_offset += 0x10; \
3597 sub_tile_height--; \
3599 texture_offset += 0xF00; \
3600 psx_gpu->num_blocks = num_blocks; \
3604 #define setup_sprite_tile_column_edge_pre_adjust_half_right_4x() \
3605 texture_offset = texture_offset_base + 8; \
3608 #define setup_sprite_tile_column_edge_pre_adjust_half_left_4x() \
3609 texture_offset = texture_offset_base \
3611 #define setup_sprite_tile_column_edge_pre_adjust_half_4x(edge) \
3612 setup_sprite_tile_column_edge_pre_adjust_half_##edge##_4x() \
3614 #define setup_sprite_tile_column_edge_pre_adjust_full_4x(edge) \
3615 texture_offset = texture_offset_base \
3617 #define setup_sprite_tile_column_edge_post_adjust_half_right_4x() \
3620 #define setup_sprite_tile_column_edge_post_adjust_half_left_4x() \
3622 #define setup_sprite_tile_column_edge_post_adjust_half_4x(edge) \
3623 setup_sprite_tile_column_edge_post_adjust_half_##edge##_4x() \
3625 #define setup_sprite_tile_column_edge_post_adjust_full_4x(edge) \
3628 #define setup_sprite_offset_u_adjust() \
3630 #define setup_sprite_comapre_left_block_mask() \
3631 ((left_block_mask & 0xFF) == 0xFF) \
3633 #define setup_sprite_comapre_right_block_mask() \
3634 (((right_block_mask >> 8) & 0xFF) == 0xFF) \
3637 #define setup_sprite_offset_u_adjust_4x() \
3639 offset_u_right = offset_u_right * 2 + 1 \
3641 #define setup_sprite_comapre_left_block_mask_4x() \
3642 ((left_block_mask & 0xFFFF) == 0xFFFF) \
3644 #define setup_sprite_comapre_right_block_mask_4x() \
3645 (((right_block_mask >> 16) & 0xFFFF) == 0xFFFF) \
3648 #define setup_sprite_tiled_builder(texture_mode, x4mode) \
3649 void setup_sprite_##texture_mode##x4mode(psx_gpu_struct *psx_gpu, s32 x, s32 y,\
3650 s32 u, s32 v, s32 width, s32 height, u32 color) \
3652 s32 offset_u = u & 0xF; \
3653 s32 offset_v = v & 0xF; \
3655 s32 width_rounded = offset_u + width + 15; \
3656 s32 height_rounded = offset_v + height + 15; \
3657 s32 tile_height = height_rounded / 16; \
3658 s32 tile_width = width_rounded / 16; \
3659 u32 offset_u_right = width_rounded & 0xF; \
3661 setup_sprite_offset_u_adjust##x4mode(); \
3663 u32 left_block_mask = ~(0xFFFFFFFF << offset_u); \
3664 u32 right_block_mask = 0xFFFFFFFE << offset_u_right; \
3666 u32 left_mask_bits; \
3667 u32 right_mask_bits; \
3669 u32 sub_tile_height; \
3672 u32 texture_mask = (psx_gpu->texture_mask_width & 0xF) | \
3673 ((psx_gpu->texture_mask_height & 0xF) << 4) | \
3674 ((psx_gpu->texture_mask_width >> 4) << 8) | \
3675 ((psx_gpu->texture_mask_height >> 4) << 12); \
3676 u32 texture_offset = ((v & 0xF) << 4) | ((u & 0xF0) << 4) | \
3677 ((v & 0xF0) << 8); \
3678 u32 texture_offset_base = texture_offset; \
3681 u16 *fb_ptr = psx_gpu->vram_out_ptr + (y * 1024) + (x - offset_u); \
3682 u32 num_blocks = psx_gpu->num_blocks; \
3683 block_struct *block = psx_gpu->blocks + num_blocks; \
3685 u16 *texture_block_ptr; \
3688 setup_sprite_tiled_initialize_##texture_mode##x4mode(); \
3690 control_mask = tile_width == 1; \
3691 control_mask |= (tile_height == 1) << 1; \
3692 control_mask |= setup_sprite_comapre_left_block_mask##x4mode() << 2; \
3693 control_mask |= setup_sprite_comapre_right_block_mask##x4mode() << 3; \
3695 sprites_##texture_mode++; \
3697 switch(control_mask) \
3701 setup_sprite_tile_column_width_multi(texture_mode, multi, full, full, \
3706 setup_sprite_tile_column_width_single(texture_mode, multi, full, none, \
3711 setup_sprite_tile_column_width_multi(texture_mode, single, full, full, \
3716 setup_sprite_tile_column_width_single(texture_mode, single, full, none, \
3721 setup_sprite_tile_column_width_multi(texture_mode, multi, half, full, \
3726 setup_sprite_tile_column_width_single(texture_mode, multi, half, right, \
3731 setup_sprite_tile_column_width_multi(texture_mode, single, half, full, \
3736 setup_sprite_tile_column_width_single(texture_mode, single, half, right, \
3741 setup_sprite_tile_column_width_multi(texture_mode, multi, full, half, \
3746 setup_sprite_tile_column_width_single(texture_mode, multi, half, left, \
3751 setup_sprite_tile_column_width_multi(texture_mode, single, full, half, \
3756 setup_sprite_tile_column_width_single(texture_mode, single, half, left, \
3761 setup_sprite_tile_column_width_multi(texture_mode, multi, half, half, \
3766 setup_sprite_tile_column_width_multi(texture_mode, single, half, half, \
3773 setup_sprite_tiled_builder(4bpp,);
3774 setup_sprite_tiled_builder(8bpp,);
3776 setup_sprite_tiled_builder(4bpp,_4x);
3777 setup_sprite_tiled_builder(8bpp,_4x);
3779 void setup_sprite_16bpp(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u,
3780 s32 v, s32 width, s32 height, u32 color)
3782 u32 left_offset = u & 0x7;
3783 u32 width_rounded = width + left_offset + 7;
3785 u16 *fb_ptr = psx_gpu->vram_out_ptr + (y * 1024) + (s32)(x - left_offset);
3786 u32 right_width = width_rounded & 0x7;
3787 u32 block_width = width_rounded / 8;
3788 u32 fb_ptr_pitch = (1024 + 8) - (block_width * 8);
3790 u32 left_mask_bits = ~(0xFF << left_offset);
3791 u32 right_mask_bits = 0xFE << right_width;
3793 u32 texture_offset_base = u + (v * 1024);
3795 psx_gpu->texture_mask_width | (psx_gpu->texture_mask_height * 1024);
3797 u32 blocks_remaining;
3798 u32 num_blocks = psx_gpu->num_blocks;
3799 block_struct *block = psx_gpu->blocks + num_blocks;
3801 u16 *texture_page_ptr = psx_gpu->texture_page_ptr;
3802 u16 *texture_block_ptr;
3804 texture_offset_base &= ~0x7;
3808 if(block_width == 1)
3810 u32 mask_bits = left_mask_bits | right_mask_bits;
3817 if(num_blocks > MAX_BLOCKS)
3819 flush_render_block_buffer(psx_gpu);
3821 block = psx_gpu->blocks;
3825 texture_page_ptr + (texture_offset_base & texture_mask);
3827 load_128b(block->texels, texture_block_ptr);
3828 block->draw_mask_bits = mask_bits;
3829 block->fb_ptr = fb_ptr;
3833 texture_offset_base += 1024;
3837 psx_gpu->num_blocks = num_blocks;
3846 blocks_remaining = block_width - 2;
3847 num_blocks += block_width;
3848 sprite_blocks += block_width;
3850 if(num_blocks > MAX_BLOCKS)
3852 flush_render_block_buffer(psx_gpu);
3853 num_blocks = block_width;
3854 block = psx_gpu->blocks;
3857 texture_offset = texture_offset_base;
3858 texture_offset_base += 1024;
3860 texture_block_ptr = texture_page_ptr + (texture_offset & texture_mask);
3861 load_128b(block->texels, texture_block_ptr);
3863 block->draw_mask_bits = left_mask_bits;
3864 block->fb_ptr = fb_ptr;
3866 texture_offset += 8;
3870 while(blocks_remaining)
3872 texture_block_ptr = texture_page_ptr + (texture_offset & texture_mask);
3873 load_128b(block->texels, texture_block_ptr);
3875 block->draw_mask_bits = 0;
3876 block->fb_ptr = fb_ptr;
3878 texture_offset += 8;
3885 texture_block_ptr = texture_page_ptr + (texture_offset & texture_mask);
3886 load_128b(block->texels, texture_block_ptr);
3888 block->draw_mask_bits = right_mask_bits;
3889 block->fb_ptr = fb_ptr;
3891 fb_ptr += fb_ptr_pitch;
3895 psx_gpu->num_blocks = num_blocks;
3900 void setup_sprite_untextured(psx_gpu_struct *psx_gpu, s32 x, s32 y, s32 u,
3901 s32 v, s32 width, s32 height, u32 color)
3903 if((psx_gpu->render_state & (RENDER_STATE_MASK_EVALUATE |
3904 RENDER_FLAGS_MODULATE_TEXELS | RENDER_FLAGS_BLEND)) == 0 &&
3905 (psx_gpu->render_mode & RENDER_INTERLACE_ENABLED) == 0)
3907 setup_sprite_untextured_simple(psx_gpu, x, y, u, v, width, height, color);
3911 u32 right_width = ((width - 1) & 0x7) + 1;
3912 u32 right_mask_bits = (0xFF << right_width);
3913 u16 *fb_ptr = psx_gpu->vram_out_ptr + (y * 1024) + x;
3914 u32 block_width = (width + 7) / 8;
3915 u32 fb_ptr_pitch = 1024 - ((block_width - 1) * 8);
3916 u32 blocks_remaining;
3917 u32 num_blocks = psx_gpu->num_blocks;
3918 block_struct *block = psx_gpu->blocks + num_blocks;
3920 u32 color_r = color & 0xFF;
3921 u32 color_g = (color >> 8) & 0xFF;
3922 u32 color_b = (color >> 16) & 0xFF;
3924 vec_8x16u right_mask;
3925 vec_8x16u test_mask = psx_gpu->test_mask;
3926 vec_8x16u zero_mask;
3928 sprites_untextured++;
3930 color = (color_r >> 3) | ((color_g >> 3) << 5) | ((color_b >> 3) << 10);
3932 dup_8x16b(colors, color);
3933 dup_8x16b(zero_mask, 0x00);
3934 dup_8x16b(right_mask, right_mask_bits);
3935 tst_8x16b(right_mask, right_mask, test_mask);
3939 blocks_remaining = block_width - 1;
3940 num_blocks += block_width;
3943 sprite_blocks += block_width;
3946 if(num_blocks > MAX_BLOCKS)
3948 flush_render_block_buffer(psx_gpu);
3949 num_blocks = block_width;
3950 block = psx_gpu->blocks;
3953 while(blocks_remaining)
3955 block->pixels = colors;
3956 block->draw_mask = zero_mask;
3957 block->fb_ptr = fb_ptr;
3964 block->pixels = colors;
3965 block->draw_mask = right_mask;
3966 block->fb_ptr = fb_ptr;
3969 fb_ptr += fb_ptr_pitch;
3972 psx_gpu->num_blocks = num_blocks;
3978 void setup_sprite_untextured_simple(psx_gpu_struct *psx_gpu, s32 x, s32 y,
3979 s32 u, s32 v, s32 width, s32 height, u32 color)
3981 u32 r = color & 0xFF;
3982 u32 g = (color >> 8) & 0xFF;
3983 u32 b = (color >> 16) & 0xFF;
3984 u32 color_16bpp = (r >> 3) | ((g >> 3) << 5) | ((b >> 3) << 10) |
3986 u32 color_32bpp = color_16bpp | (color_16bpp << 16);
3988 u16 *vram_ptr16 = psx_gpu->vram_out_ptr + x + (y * 1024);
3993 if(psx_gpu->num_blocks > MAX_BLOCKS)
3995 flush_render_block_buffer(psx_gpu);
4002 vram_ptr = (void *)vram_ptr16;
4003 if((long)vram_ptr16 & 2)
4005 *vram_ptr16 = color_32bpp;
4006 vram_ptr = (void *)(vram_ptr16 + 1);
4010 while(num_width >= 4 * 2)
4012 vram_ptr[0] = color_32bpp;
4013 vram_ptr[1] = color_32bpp;
4014 vram_ptr[2] = color_32bpp;
4015 vram_ptr[3] = color_32bpp;
4021 while(num_width >= 2)
4023 *vram_ptr++ = color_32bpp;
4029 *(u16 *)vram_ptr = color_32bpp;
4038 #define setup_sprite_blocks_switch_textured(texture_mode) \
4039 setup_sprite_##texture_mode \
4041 #define setup_sprite_blocks_switch_untextured(texture_mode) \
4042 setup_sprite_untextured \
4044 #define setup_sprite_blocks_switch(texturing, texture_mode) \
4045 setup_sprite_blocks_switch_##texturing(texture_mode) \
4048 #define texture_sprite_blocks_switch_4bpp() \
4049 texture_blocks_untextured \
4051 #define texture_sprite_blocks_switch_8bpp() \
4052 texture_sprite_blocks_8bpp \
4054 #define texture_sprite_blocks_switch_16bpp() \
4055 texture_blocks_untextured \
4057 #define texture_sprite_blocks_switch_untextured(texture_mode) \
4058 texture_blocks_untextured \
4060 #define texture_sprite_blocks_switch_textured(texture_mode) \
4061 texture_sprite_blocks_switch_##texture_mode() \
4063 #define render_sprite_blocks_switch_block_modulation(texture_mode, blend_mode, \
4064 mask_evaluate, shading, dithering, texturing, blending, modulation) \
4066 setup_sprite_blocks_switch(texturing, texture_mode), \
4067 texture_sprite_blocks_switch_##texturing(texture_mode), \
4068 shade_blocks_switch(unshaded, texturing, modulation, undithered, blending, \
4070 blend_blocks_switch(texturing, blending, blend_mode, mask_evaluate) \
4073 #define render_sprite_blocks_switch_block_blending(texture_mode, blend_mode, \
4074 mask_evaluate, shading, dithering, texturing, blending) \
4075 render_sprite_blocks_switch_block_modulation(texture_mode, blend_mode, \
4076 mask_evaluate, shading, dithering, texturing, blending, modulated), \
4077 render_sprite_blocks_switch_block_modulation(texture_mode, blend_mode, \
4078 mask_evaluate, shading, dithering, texturing, blending, unmodulated) \
4080 #define render_sprite_blocks_switch_block_texturing(texture_mode, blend_mode, \
4081 mask_evaluate, shading, dithering, texturing) \
4082 render_sprite_blocks_switch_block_blending(texture_mode, blend_mode, \
4083 mask_evaluate, shading, dithering, texturing, unblended), \
4084 render_sprite_blocks_switch_block_blending(texture_mode, blend_mode, \
4085 mask_evaluate, shading, dithering, texturing, blended) \
4087 #define render_sprite_blocks_switch_block_dithering(texture_mode, blend_mode, \
4088 mask_evaluate, shading, dithering) \
4089 render_sprite_blocks_switch_block_texturing(texture_mode, blend_mode, \
4090 mask_evaluate, shading, dithering, untextured), \
4091 render_sprite_blocks_switch_block_texturing(texture_mode, blend_mode, \
4092 mask_evaluate, shading, dithering, textured) \
4094 #define render_sprite_blocks_switch_block_shading(texture_mode, blend_mode, \
4095 mask_evaluate, shading) \
4096 render_sprite_blocks_switch_block_dithering(texture_mode, blend_mode, \
4097 mask_evaluate, shading, undithered), \
4098 render_sprite_blocks_switch_block_dithering(texture_mode, blend_mode, \
4099 mask_evaluate, shading, dithered) \
4101 #define render_sprite_blocks_switch_block_mask_evaluate(texture_mode, \
4102 blend_mode, mask_evaluate) \
4103 render_sprite_blocks_switch_block_shading(texture_mode, blend_mode, \
4104 mask_evaluate, unshaded), \
4105 render_sprite_blocks_switch_block_shading(texture_mode, blend_mode, \
4106 mask_evaluate, shaded) \
4108 #define render_sprite_blocks_switch_block_blend_mode(texture_mode, blend_mode) \
4109 render_sprite_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, \
4111 render_sprite_blocks_switch_block_mask_evaluate(texture_mode, blend_mode, \
4114 #define render_sprite_blocks_switch_block_texture_mode(texture_mode) \
4115 render_sprite_blocks_switch_block_blend_mode(texture_mode, average), \
4116 render_sprite_blocks_switch_block_blend_mode(texture_mode, add), \
4117 render_sprite_blocks_switch_block_blend_mode(texture_mode, subtract), \
4118 render_sprite_blocks_switch_block_blend_mode(texture_mode, add_fourth) \
4120 #define render_sprite_blocks_switch_block() \
4121 render_sprite_blocks_switch_block_texture_mode(4bpp), \
4122 render_sprite_blocks_switch_block_texture_mode(8bpp), \
4123 render_sprite_blocks_switch_block_texture_mode(16bpp), \
4124 render_sprite_blocks_switch_block_texture_mode(16bpp) \
4127 render_block_handler_struct render_sprite_block_handlers[] =
4129 render_sprite_blocks_switch_block()
4133 void render_sprite(psx_gpu_struct *psx_gpu, s32 x, s32 y, u32 u, u32 v,
4134 s32 width, s32 height, u32 flags, u32 color)
4136 s32 x_right = x + width - 1;
4137 s32 y_bottom = y + height - 1;
4143 if(invalidate_texture_cache_region_viewport(psx_gpu, x, y, x_right,
4149 if(x < psx_gpu->viewport_start_x)
4151 u32 clip = psx_gpu->viewport_start_x - x;
4157 if(y < psx_gpu->viewport_start_y)
4159 s32 clip = psx_gpu->viewport_start_y - y;
4165 if(x_right > psx_gpu->viewport_end_x)
4166 width -= x_right - psx_gpu->viewport_end_x;
4168 if(y_bottom > psx_gpu->viewport_end_y)
4169 height -= y_bottom - psx_gpu->viewport_end_y;
4171 if((width <= 0) || (height <= 0))
4175 span_pixels += width * height;
4179 u32 render_state = flags &
4180 (RENDER_FLAGS_MODULATE_TEXELS | RENDER_FLAGS_BLEND |
4181 RENDER_FLAGS_TEXTURE_MAP);
4183 (psx_gpu->render_state_base & ~RENDER_STATE_DITHER);
4185 if((psx_gpu->render_state != render_state) ||
4186 (psx_gpu->primitive_type != PRIMITIVE_TYPE_SPRITE))
4188 psx_gpu->render_state = render_state;
4189 flush_render_block_buffer(psx_gpu);
4195 psx_gpu->primitive_type = PRIMITIVE_TYPE_SPRITE;
4199 if(psx_gpu->triangle_color != color)
4201 flush_render_block_buffer(psx_gpu);
4202 psx_gpu->triangle_color = color;
4205 if(color == 0x808080)
4206 render_state |= RENDER_FLAGS_MODULATE_TEXELS;
4208 render_block_handler_struct *render_block_handler =
4209 &(render_sprite_block_handlers[render_state]);
4210 psx_gpu->render_block_handler = render_block_handler;
4212 ((setup_sprite_function_type *)render_block_handler->setup_blocks)
4213 (psx_gpu, x, y, u, v, width, height, color);
4216 #define draw_pixel_line_mask_evaluate_yes() \
4217 if((*vram_ptr & 0x8000) == 0) \
4219 #define draw_pixel_line_mask_evaluate_no() \
4222 #define draw_pixel_line_shaded() \
4224 color_r = fixed_to_int(current_r); \
4225 color_g = fixed_to_int(current_g); \
4226 color_b = fixed_to_int(current_b); \
4228 current_r += gradient_r; \
4229 current_g += gradient_g; \
4230 current_b += gradient_b; \
4233 #define draw_pixel_line_unshaded() \
4235 color_r = color & 0xFF; \
4236 color_g = (color >> 8) & 0xFF; \
4237 color_b = (color >> 16) & 0xFF; \
4241 #define draw_pixel_line_dithered(_x, _y) \
4243 u32 dither_xor = _x ^ _y; \
4244 s32 dither_offset = (dither_xor >> 1) & 0x1; \
4245 dither_offset |= (_y & 0x1) << 1; \
4246 dither_offset |= (dither_xor & 0x1) << 2; \
4247 dither_offset -= 4; \
4249 color_r += dither_offset; \
4250 color_g += dither_offset; \
4251 color_b += dither_offset; \
4272 #define draw_pixel_line_undithered(_x, _y) \
4275 #define draw_pixel_line_average() \
4276 color_r = (color_r + fb_r) / 2; \
4277 color_g = (color_g + fb_g) / 2; \
4278 color_b = (color_b + fb_b) / 2 \
4280 #define draw_pixel_line_add() \
4295 #define draw_pixel_line_subtract() \
4296 color_r = fb_r - color_r; \
4297 color_g = fb_g - color_g; \
4298 color_b = fb_b - color_b; \
4309 #define draw_pixel_line_add_fourth() \
4310 color_r = fb_r + (color_r / 4); \
4311 color_g = fb_g + (color_g / 4); \
4312 color_b = fb_b + (color_b / 4); \
4324 #define draw_pixel_line_blended(blend_mode) \
4325 s32 fb_pixel = *vram_ptr; \
4326 s32 fb_r = fb_pixel & 0x1F; \
4327 s32 fb_g = (fb_pixel >> 5) & 0x1F; \
4328 s32 fb_b = (fb_pixel >> 10) & 0x1F; \
4330 draw_pixel_line_##blend_mode() \
4332 #define draw_pixel_line_unblended(blend_mode) \
4335 #define draw_pixel_line(_x, _y, shading, blending, dithering, mask_evaluate, \
4337 if((_x >= psx_gpu->viewport_start_x) && (_y >= psx_gpu->viewport_start_y) && \
4338 (_x <= psx_gpu->viewport_end_x) && (_y <= psx_gpu->viewport_end_y)) \
4340 draw_pixel_line_mask_evaluate_##mask_evaluate() \
4342 draw_pixel_line_##shading(); \
4343 draw_pixel_line_##dithering(_x, _y); \
4349 draw_pixel_line_##blending(blend_mode); \
4351 *vram_ptr = color_r | (color_g << 5) | (color_b << 10) | \
4352 psx_gpu->mask_msb; \
4356 #define update_increment(value) \
4359 #define update_decrement(value) \
4362 #define update_vram_row_increment(value) \
4365 #define update_vram_row_decrement(value) \
4368 #define compare_increment(a, b) \
4371 #define compare_decrement(a, b) \
4374 #define set_line_gradients(minor) \
4376 s32 gradient_divisor = delta_##minor; \
4377 if(gradient_divisor != 0) \
4379 gradient_r = int_to_fixed(vertex_b->r - vertex_a->r) / gradient_divisor; \
4380 gradient_g = int_to_fixed(vertex_b->g - vertex_a->g) / gradient_divisor; \
4381 gradient_b = int_to_fixed(vertex_b->b - vertex_a->b) / gradient_divisor; \
4389 current_r = fixed_center(vertex_a->r); \
4390 current_g = fixed_center(vertex_a->g); \
4391 current_b = fixed_center(vertex_a->b); \
4394 #define draw_line_span_horizontal(direction, shading, blending, dithering, \
4395 mask_evaluate, blend_mode) \
4398 error_step = delta_y * 2; \
4399 error_wrap = delta_x * 2; \
4403 set_line_gradients(x); \
4405 for(current_x = x_a; current_x <= x_b; current_x++) \
4407 draw_pixel_line(current_x, current_y, shading, blending, dithering, \
4408 mask_evaluate, blend_mode); \
4409 error += error_step; \
4412 if(error >= error_wrap) \
4414 update_##direction(current_y); \
4415 update_vram_row_##direction(); \
4416 error -= error_wrap; \
4421 #define draw_line_span_vertical(direction, shading, blending, dithering, \
4422 mask_evaluate, blend_mode) \
4425 error_step = delta_x * 2; \
4426 error_wrap = delta_y * 2; \
4430 set_line_gradients(y); \
4432 for(current_y = y_a; compare_##direction(current_y, y_b); \
4433 update_##direction(current_y)) \
4435 draw_pixel_line(current_x, current_y, shading, blending, dithering, \
4436 mask_evaluate, blend_mode); \
4437 error += error_step; \
4438 update_vram_row_##direction(); \
4440 if(error > error_wrap) \
4444 error -= error_wrap; \
4450 #define render_line_body(shading, blending, dithering, mask_evaluate, \
4456 if(delta_x > delta_y) \
4458 draw_line_span_horizontal(decrement, shading, blending, dithering, \
4459 mask_evaluate, blend_mode); \
4463 draw_line_span_vertical(decrement, shading, blending, dithering, \
4464 mask_evaluate, blend_mode); \
4469 if(delta_x > delta_y) \
4471 draw_line_span_horizontal(increment, shading, blending, dithering, \
4472 mask_evaluate, blend_mode); \
4476 draw_line_span_vertical(increment, shading, blending, dithering, \
4477 mask_evaluate, blend_mode); \
4482 void render_line(psx_gpu_struct *psx_gpu, vertex_struct *vertexes, u32 flags,
4483 u32 color, int double_resolution)
4485 s32 color_r, color_g, color_b;
4486 u32 triangle_winding = 0;
4488 fixed_type gradient_r = 0;
4489 fixed_type gradient_g = 0;
4490 fixed_type gradient_b = 0;
4491 fixed_type current_r = 0;
4492 fixed_type current_g = 0;
4493 fixed_type current_b = 0;
4498 s32 delta_x, delta_y;
4509 flush_render_block_buffer(psx_gpu);
4510 psx_gpu->primitive_type = PRIMITIVE_TYPE_LINE;
4512 vertex_struct *vertex_a = &(vertexes[0]);
4513 vertex_struct *vertex_b = &(vertexes[1]);
4521 if(vertex_a->x >= vertex_b->x)
4523 vertex_swap(vertex_a, vertex_b);
4532 delta_x = x_b - x_a;
4533 delta_y = y_b - y_a;
4535 if(delta_x >= 1024 || delta_y >= 512 || delta_y <= -512)
4538 if(double_resolution)
4548 flags &= ~RENDER_FLAGS_TEXTURE_MAP;
4550 vram_ptr = psx_gpu->vram_out_ptr + (y_a * 1024) + x_a;
4554 if(flags & RENDER_FLAGS_SHADE)
4555 control_mask |= 0x1;
4557 if(flags & RENDER_FLAGS_BLEND)
4559 control_mask |= 0x2;
4560 control_mask |= ((psx_gpu->render_state_base >> 6) & 0x3) << 4;
4563 if(psx_gpu->render_state_base & RENDER_STATE_DITHER)
4564 control_mask |= 0x4;
4566 if(psx_gpu->render_state_base & RENDER_STATE_MASK_EVALUATE)
4567 control_mask |= 0x8;
4569 switch(control_mask)
4572 render_line_body(unshaded, unblended, undithered, no, none);
4576 render_line_body(shaded, unblended, undithered, no, none);
4580 render_line_body(unshaded, blended, undithered, no, average);
4584 render_line_body(shaded, blended, undithered, no, average);
4588 render_line_body(unshaded, unblended, dithered, no, none);
4592 render_line_body(shaded, unblended, dithered, no, none);
4596 render_line_body(unshaded, blended, dithered, no, average);
4600 render_line_body(shaded, blended, dithered, no, average);
4604 render_line_body(unshaded, unblended, undithered, yes, none);
4608 render_line_body(shaded, unblended, undithered, yes, none);
4612 render_line_body(unshaded, blended, undithered, yes, average);
4616 render_line_body(shaded, blended, undithered, yes, average);
4620 render_line_body(unshaded, unblended, dithered, yes, none);
4624 render_line_body(shaded, unblended, dithered, yes, none);
4628 render_line_body(unshaded, blended, dithered, yes, average);
4632 render_line_body(shaded, blended, dithered, yes, average);
4636 render_line_body(unshaded, blended, undithered, no, add);
4640 render_line_body(shaded, blended, undithered, no, add);
4644 render_line_body(unshaded, blended, dithered, no, add);
4648 render_line_body(shaded, blended, dithered, no, add);
4652 render_line_body(unshaded, blended, undithered, yes, add);
4656 render_line_body(shaded, blended, undithered, yes, add);
4660 render_line_body(unshaded, blended, dithered, yes, add);
4664 render_line_body(shaded, blended, dithered, yes, add);
4668 render_line_body(unshaded, blended, undithered, no, subtract);
4672 render_line_body(shaded, blended, undithered, no, subtract);
4676 render_line_body(unshaded, blended, dithered, no, subtract);
4680 render_line_body(shaded, blended, dithered, no, subtract);
4684 render_line_body(unshaded, blended, undithered, yes, subtract);
4688 render_line_body(shaded, blended, undithered, yes, subtract);
4692 render_line_body(unshaded, blended, dithered, yes, subtract);
4696 render_line_body(shaded, blended, dithered, yes, subtract);
4700 render_line_body(unshaded, blended, undithered, no, add_fourth);
4704 render_line_body(shaded, blended, undithered, no, add_fourth);
4708 render_line_body(unshaded, blended, dithered, no, add_fourth);
4712 render_line_body(shaded, blended, dithered, no, add_fourth);
4716 render_line_body(unshaded, blended, undithered, yes, add_fourth);
4720 render_line_body(shaded, blended, undithered, yes, add_fourth);
4724 render_line_body(unshaded, blended, dithered, yes, add_fourth);
4728 render_line_body(shaded, blended, dithered, yes, add_fourth);
4734 void render_block_fill(psx_gpu_struct *psx_gpu, u32 color, u32 x, u32 y,
4735 u32 width, u32 height)
4737 if((width == 0) || (height == 0))
4740 invalidate_texture_cache_region(psx_gpu, x, y, x + width - 1, y + height - 1);
4742 u32 r = color & 0xFF;
4743 u32 g = (color >> 8) & 0xFF;
4744 u32 b = (color >> 16) & 0xFF;
4745 u32 color_16bpp = (r >> 3) | ((g >> 3) << 5) | ((b >> 3) << 10) |
4747 u32 color_32bpp = color_16bpp | (color_16bpp << 16);
4749 u32 *vram_ptr = (u32 *)(psx_gpu->vram_out_ptr + x + (y * 1024));
4751 u32 pitch = 512 - (width / 2);
4754 if(psx_gpu->render_mode & RENDER_INTERLACE_ENABLED)
4759 if(psx_gpu->render_mode & RENDER_INTERLACE_ODD)
4768 vram_ptr[0] = color_32bpp;
4769 vram_ptr[1] = color_32bpp;
4770 vram_ptr[2] = color_32bpp;
4771 vram_ptr[3] = color_32bpp;
4772 vram_ptr[4] = color_32bpp;
4773 vram_ptr[5] = color_32bpp;
4774 vram_ptr[6] = color_32bpp;
4775 vram_ptr[7] = color_32bpp;
4786 void render_block_fill_enh(psx_gpu_struct *psx_gpu, u32 color, u32 x, u32 y,
4787 u32 width, u32 height)
4789 if((width == 0) || (height == 0))
4795 u32 r = color & 0xFF;
4796 u32 g = (color >> 8) & 0xFF;
4797 u32 b = (color >> 16) & 0xFF;
4798 u32 color_16bpp = (r >> 3) | ((g >> 3) << 5) | ((b >> 3) << 10) |
4800 u32 color_32bpp = color_16bpp | (color_16bpp << 16);
4802 u32 *vram_ptr = (u32 *)(psx_gpu->vram_out_ptr + x + (y * 1024));
4804 u32 pitch = 1024 / 2 - (width / 2);
4812 vram_ptr[0] = color_32bpp;
4813 vram_ptr[1] = color_32bpp;
4814 vram_ptr[2] = color_32bpp;
4815 vram_ptr[3] = color_32bpp;
4816 vram_ptr[4] = color_32bpp;
4817 vram_ptr[5] = color_32bpp;
4818 vram_ptr[6] = color_32bpp;
4819 vram_ptr[7] = color_32bpp;
4830 void render_block_copy(psx_gpu_struct *psx_gpu, u16 *source, u32 x, u32 y,
4831 u32 width, u32 height, u32 pitch)
4833 u16 *vram_ptr = psx_gpu->vram_ptr + x + (y * 1024);
4835 u32 mask_msb = psx_gpu->mask_msb;
4837 if((width == 0) || (height == 0))
4840 flush_render_block_buffer(psx_gpu);
4841 invalidate_texture_cache_region(psx_gpu, x, y, x + width - 1, y + height - 1);
4843 for(draw_y = 0; draw_y < height; draw_y++)
4845 for(draw_x = 0; draw_x < width; draw_x++)
4847 vram_ptr[draw_x] = source[draw_x] | mask_msb;
4855 void render_block_move(psx_gpu_struct *psx_gpu, u32 source_x, u32 source_y,
4856 u32 dest_x, u32 dest_y, u32 width, u32 height)
4858 render_block_copy(psx_gpu, psx_gpu->vram_ptr + source_x + (source_y * 1024),
4859 dest_x, dest_y, width, height, 1024);
4863 void initialize_reciprocal_table(void)
4866 u32 height_normalized;
4867 u32 height_reciprocal;
4870 for(height = 1; height < sizeof(reciprocal_table)
4871 / sizeof(reciprocal_table[0]); height++)
4873 shift = __builtin_clz(height);
4874 height_normalized = height << shift;
4875 height_reciprocal = ((1ULL << 51) + (height_normalized - 1)) /
4878 shift = 32 - (51 - shift);
4880 reciprocal_table[height] = (height_reciprocal << 10) | shift;
4885 #define dither_table_row(a, b, c, d) \
4886 ((a & 0xFF) | ((b & 0xFF) << 8) | ((c & 0xFF) << 16) | ((d & 0xFF) << 24)) \
4888 void initialize_psx_gpu(psx_gpu_struct *psx_gpu, u16 *vram)
4890 vec_8x16u test_mask =
4891 { { { 0x01, 0x02, 0x04, 0x08, 0x10, 0x20, 0x40, 0x80 } } };
4893 psx_gpu->test_mask = test_mask;
4895 psx_gpu->dirty_textures_4bpp_mask = 0xFFFFFFFF;
4896 psx_gpu->dirty_textures_8bpp_mask = 0xFFFFFFFF;
4897 psx_gpu->dirty_textures_8bpp_alternate_mask = 0xFFFFFFFF;
4898 psx_gpu->viewport_mask = 0;
4899 psx_gpu->current_texture_page = 0;
4900 psx_gpu->current_texture_mask = 0;
4901 psx_gpu->last_8bpp_texture_page = 0;
4903 psx_gpu->clut_settings = 0;
4904 psx_gpu->texture_settings = 0;
4905 psx_gpu->render_state = 0;
4906 psx_gpu->render_state_base = 0;
4907 psx_gpu->num_blocks = 0;
4908 psx_gpu->uvrgb_phase = 0x8000;
4910 psx_gpu->vram_ptr = vram;
4911 psx_gpu->vram_out_ptr = vram;
4913 psx_gpu->texture_page_base = psx_gpu->vram_ptr;
4914 psx_gpu->texture_page_ptr = psx_gpu->vram_ptr;
4915 psx_gpu->clut_ptr = psx_gpu->vram_ptr;
4917 psx_gpu->mask_msb = 0;
4919 psx_gpu->texture_window_x = 0;
4920 psx_gpu->texture_window_y = 0;
4921 psx_gpu->texture_mask_width = 0xFF;
4922 psx_gpu->texture_mask_height = 0xFF;
4924 psx_gpu->render_mode = 0;
4926 memset(psx_gpu->vram_ptr, 0, sizeof(u16) * 1024 * 512);
4928 initialize_reciprocal_table();
4929 psx_gpu->reciprocal_table_ptr = reciprocal_table;
4936 // (minus ones(4) * 4)
4938 // d0: (1 3 5 7): x1 ^ y1
4939 // d1: (2 3 6 7): y0
4940 // d2: (4 5 6 7): x0 ^ y0
4942 psx_gpu->dither_table[0] = dither_table_row(-4, 0, -3, 1);
4943 psx_gpu->dither_table[1] = dither_table_row(2, -2, 3, -1);
4944 psx_gpu->dither_table[2] = dither_table_row(-3, 1, -4, 0);
4945 psx_gpu->dither_table[3] = dither_table_row(3, -1, 2, -2);
4947 psx_gpu->primitive_type = PRIMITIVE_TYPE_UNKNOWN;
4949 psx_gpu->enhancement_x_threshold = 256;
4955 gettimeofday(&tv, NULL);
4957 return (tv.tv_sec * 1000000ULL) + tv.tv_usec;
4960 #if 0 //def NEON_BUILD
4965 __asm__ volatile("mrc p15, 0, %0, c9, c13, 0" : "=r"(counter));
4970 void init_counter(void)
4973 asm volatile("mrc p15, 0, %0, c9, c12, 0" : "=r"(value));
4974 value |= 5; // master enable, ccnt reset
4975 value &= ~8; // ccnt divider 0
4976 asm volatile("mcr p15, 0, %0, c9, c12, 0" :: "r"(value));
4977 // enable cycle counter
4978 asm volatile("mcr p15, 0, %0, c9, c12, 1" :: "r"(1 << 31));
4981 void triangle_benchmark(psx_gpu_struct *psx_gpu)
4988 const u32 iterations = 500000;
4990 psx_gpu->num_blocks = 64;
4991 psx_gpu->clut_ptr = psx_gpu->vram_ptr;
4993 for(i = 0; i < 64; i++)
4995 memset(&(psx_gpu->blocks[i].r), 0, 16);
5000 ticks = get_counter();
5002 for(i = 0; i < iterations; i++)
5004 texture_sprite_blocks_8bpp(psx_gpu);
5007 ticks_elapsed = get_counter() - ticks;
5009 printf("benchmark: %lf cycles\n", (double)ticks_elapsed / (iterations * 64));
5014 #include "psx_gpu_4x.c"