| 1 | /*************************************************************************** |
| 2 | * Copyright (C) 2010 PCSX4ALL Team * |
| 3 | * Copyright (C) 2010 Unai * |
| 4 | * Copyright (C) 2016 Senquack (dansilsby <AT> gmail <DOT> com) * |
| 5 | * * |
| 6 | * This program is free software; you can redistribute it and/or modify * |
| 7 | * it under the terms of the GNU General Public License as published by * |
| 8 | * the Free Software Foundation; either version 2 of the License, or * |
| 9 | * (at your option) any later version. * |
| 10 | * * |
| 11 | * This program is distributed in the hope that it will be useful, * |
| 12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of * |
| 13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * |
| 14 | * GNU General Public License for more details. * |
| 15 | * * |
| 16 | * You should have received a copy of the GNU General Public License * |
| 17 | * along with this program; if not, write to the * |
| 18 | * Free Software Foundation, Inc., * |
| 19 | * 51 Franklin Street, Fifth Floor, Boston, MA 02111-1307 USA. * |
| 20 | ***************************************************************************/ |
| 21 | |
| 22 | #ifndef GPU_UNAI_H |
| 23 | #define GPU_UNAI_H |
| 24 | |
| 25 | #include "gpu.h" |
| 26 | |
| 27 | // Header shared between both standalone gpu_unai (gpu.cpp) and new |
| 28 | // gpulib-compatible gpu_unai (gpulib_if.cpp) |
| 29 | // -> Anything here should be for gpu_unai's private use. <- |
| 30 | |
| 31 | /////////////////////////////////////////////////////////////////////////////// |
| 32 | // Compile Options |
| 33 | |
| 34 | //#define ENABLE_GPU_NULL_SUPPORT // Enables NullGPU support |
| 35 | //#define ENABLE_GPU_LOG_SUPPORT // Enables gpu logger, very slow only for windows debugging |
| 36 | //#define ENABLE_GPU_ARMV7 // Enables ARMv7 optimized assembly |
| 37 | |
| 38 | //Poly routine options (default is integer math and accurate division) |
| 39 | //#define GPU_UNAI_USE_FLOATMATH // Use float math in poly routines |
| 40 | //#define GPU_UNAI_USE_FLOAT_DIV_MULTINV // If GPU_UNAI_USE_FLOATMATH is defined, |
| 41 | // use multiply-by-inverse for division |
| 42 | //#define GPU_UNAI_USE_INT_DIV_MULTINV // If GPU_UNAI_USE_FLOATMATH is *not* |
| 43 | // defined, use old inaccurate division |
| 44 | |
| 45 | |
| 46 | #define GPU_INLINE static inline __attribute__((always_inline)) |
| 47 | #define INLINE static inline __attribute__((always_inline)) |
| 48 | |
| 49 | #define u8 uint8_t |
| 50 | #define s8 int8_t |
| 51 | #define u16 uint16_t |
| 52 | #define s16 int16_t |
| 53 | #define u32 uint32_t |
| 54 | #define s32 int32_t |
| 55 | #define s64 int64_t |
| 56 | |
| 57 | typedef struct { |
| 58 | u32 v; |
| 59 | } le32_t; |
| 60 | |
| 61 | typedef struct { |
| 62 | u16 v; |
| 63 | } le16_t; |
| 64 | |
| 65 | static inline u32 le32_to_u32(le32_t le) |
| 66 | { |
| 67 | return LE32TOH(le.v); |
| 68 | } |
| 69 | |
| 70 | static inline s32 le32_to_s32(le32_t le) |
| 71 | { |
| 72 | return (int32_t) LE32TOH(le.v); |
| 73 | } |
| 74 | |
| 75 | static inline u32 le32_raw(le32_t le) |
| 76 | { |
| 77 | return le.v; |
| 78 | } |
| 79 | |
| 80 | static inline le32_t u32_to_le32(u32 u) |
| 81 | { |
| 82 | return (le32_t){ .v = HTOLE32(u) }; |
| 83 | } |
| 84 | |
| 85 | static inline u16 le16_to_u16(le16_t le) |
| 86 | { |
| 87 | return LE16TOH(le.v); |
| 88 | } |
| 89 | |
| 90 | static inline s16 le16_to_s16(le16_t le) |
| 91 | { |
| 92 | return (int16_t) LE16TOH(le.v); |
| 93 | } |
| 94 | |
| 95 | static inline u16 le16_raw(le16_t le) |
| 96 | { |
| 97 | return le.v; |
| 98 | } |
| 99 | |
| 100 | static inline le16_t u16_to_le16(u16 u) |
| 101 | { |
| 102 | return (le16_t){ .v = HTOLE16(u) }; |
| 103 | } |
| 104 | |
| 105 | union PtrUnion |
| 106 | { |
| 107 | le32_t *U4; |
| 108 | le16_t *U2; |
| 109 | u8 *U1; |
| 110 | void *ptr; |
| 111 | }; |
| 112 | |
| 113 | union GPUPacket |
| 114 | { |
| 115 | le32_t U4[16]; |
| 116 | le16_t U2[32]; |
| 117 | u8 U1[64]; |
| 118 | }; |
| 119 | |
| 120 | template<class T> static inline void SwapValues(T &x, T &y) |
| 121 | { |
| 122 | T tmp(x); x = y; y = tmp; |
| 123 | } |
| 124 | |
| 125 | template<typename T> |
| 126 | static inline T Min2 (const T a, const T b) |
| 127 | { |
| 128 | return (a<b)?a:b; |
| 129 | } |
| 130 | |
| 131 | template<typename T> |
| 132 | static inline T Min3 (const T a, const T b, const T c) |
| 133 | { |
| 134 | return Min2(Min2(a,b),c); |
| 135 | } |
| 136 | |
| 137 | template<typename T> |
| 138 | static inline T Max2 (const T a, const T b) |
| 139 | { |
| 140 | return (a>b)?a:b; |
| 141 | } |
| 142 | |
| 143 | template<typename T> |
| 144 | static inline T Max3 (const T a, const T b, const T c) |
| 145 | { |
| 146 | return Max2(Max2(a,b),c); |
| 147 | } |
| 148 | |
| 149 | |
| 150 | /////////////////////////////////////////////////////////////////////////////// |
| 151 | // GPU Raster Macros |
| 152 | |
| 153 | // Convert 24bpp color parameter of GPU command to 16bpp (15bpp + mask bit) |
| 154 | #define GPU_RGB16(rgb) ((((rgb)&0xF80000)>>9)|(((rgb)&0xF800)>>6)|(((rgb)&0xF8)>>3)) |
| 155 | |
| 156 | // Sign-extend 11-bit coordinate command param |
| 157 | #define GPU_EXPANDSIGN(x) (((s32)(x)<<(32-11))>>(32-11)) |
| 158 | |
| 159 | // Max difference between any two X or Y primitive coordinates |
| 160 | #define CHKMAX_X 1024 |
| 161 | #define CHKMAX_Y 512 |
| 162 | |
| 163 | #define FRAME_BUFFER_SIZE (1024*512*2) |
| 164 | #define FRAME_WIDTH 1024 |
| 165 | #define FRAME_HEIGHT 512 |
| 166 | #define FRAME_OFFSET(x,y) (((y)<<10)+(x)) |
| 167 | #define FRAME_BYTE_STRIDE 2048 |
| 168 | #define FRAME_BYTES_PER_PIXEL 2 |
| 169 | |
| 170 | static inline s32 GPU_DIV(s32 rs, s32 rt) |
| 171 | { |
| 172 | return rt ? (rs / rt) : (0); |
| 173 | } |
| 174 | |
| 175 | // 'Unsafe' version of above that doesn't check for div-by-zero |
| 176 | #define GPU_FAST_DIV(rs, rt) ((signed)(rs) / (signed)(rt)) |
| 177 | |
| 178 | struct gpu_unai_t { |
| 179 | u32 GPU_GP1; |
| 180 | GPUPacket PacketBuffer; |
| 181 | le16_t *vram; |
| 182 | |
| 183 | #ifdef USE_GPULIB |
| 184 | le16_t *downscale_vram; |
| 185 | #endif |
| 186 | //////////////////////////////////////////////////////////////////////////// |
| 187 | // Variables used only by older standalone version of gpu_unai (gpu.cpp) |
| 188 | #ifndef USE_GPULIB |
| 189 | u32 GPU_GP0; |
| 190 | u32 tex_window; // Current texture window vals (set by GP0(E2h) cmd) |
| 191 | s32 PacketCount; |
| 192 | s32 PacketIndex; |
| 193 | bool fb_dirty; // Framebuffer is dirty (according to GPU) |
| 194 | |
| 195 | // Display status |
| 196 | // NOTE: Standalone older gpu_unai didn't care about horiz display range |
| 197 | u16 DisplayArea[6]; // [0] : Start of display area (in VRAM) X |
| 198 | // [1] : Start of display area (in VRAM) Y |
| 199 | // [2] : Display mode resolution HORIZONTAL |
| 200 | // [3] : Display mode resolution VERTICAL |
| 201 | // [4] : Vertical display range (on TV) START |
| 202 | // [5] : Vertical display range (on TV) END |
| 203 | |
| 204 | //////////////////////////////////////////////////////////////////////////// |
| 205 | // Dma Transfers info |
| 206 | struct { |
| 207 | s32 px,py; |
| 208 | s32 x_end,y_end; |
| 209 | le16_t* pvram; |
| 210 | u32 *last_dma; // Last dma pointer |
| 211 | bool FrameToRead; // Load image in progress |
| 212 | bool FrameToWrite; // Store image in progress |
| 213 | } dma; |
| 214 | |
| 215 | //////////////////////////////////////////////////////////////////////////// |
| 216 | // Frameskip |
| 217 | struct { |
| 218 | int skipCount; // Frame skip (0,1,2,3...) |
| 219 | bool isSkip; // Skip frame (according to GPU) |
| 220 | bool skipFrame; // Skip this frame (according to frame skip) |
| 221 | bool wasSkip; // Skip frame old value (according to GPU) |
| 222 | bool skipGPU; // Skip GPU primitives |
| 223 | } frameskip; |
| 224 | #endif |
| 225 | // END of standalone gpu_unai variables |
| 226 | //////////////////////////////////////////////////////////////////////////// |
| 227 | |
| 228 | u32 TextureWindowCur; // Current setting from last GP0(0xE2) cmd (raw form) |
| 229 | u8 TextureWindow[4]; // [0] : Texture window offset X |
| 230 | // [1] : Texture window offset Y |
| 231 | // [2] : Texture window mask X |
| 232 | // [3] : Texture window mask Y |
| 233 | |
| 234 | u16 DrawingArea[4]; // [0] : Drawing area top left X |
| 235 | // [1] : Drawing area top left Y |
| 236 | // [2] : Drawing area bottom right X |
| 237 | // [3] : Drawing area bottom right Y |
| 238 | |
| 239 | s16 DrawingOffset[2]; // [0] : Drawing offset X (signed) |
| 240 | // [1] : Drawing offset Y (signed) |
| 241 | |
| 242 | le16_t* TBA; // Ptr to current texture in VRAM |
| 243 | le16_t* CBA; // Ptr to current CLUT in VRAM |
| 244 | |
| 245 | //////////////////////////////////////////////////////////////////////////// |
| 246 | // Inner Loop parameters |
| 247 | |
| 248 | // 22.10 Fixed-pt texture coords, mask, scanline advance |
| 249 | // NOTE: U,V are no longer packed together into one u32, this proved to be |
| 250 | // too imprecise, leading to pixel dropouts. Example: NFS3's skybox. |
| 251 | u32 u, v; |
| 252 | u32 u_msk, v_msk; |
| 253 | s32 u_inc, v_inc; |
| 254 | |
| 255 | // Color for Gouraud-shaded prims |
| 256 | // Packed fixed-pt 8.3:8.3:8.2 rgb triplet |
| 257 | // layout: rrrrrrrrXXXggggggggXXXbbbbbbbbXX |
| 258 | // ^ bit 31 ^ bit 0 |
| 259 | u32 gCol; |
| 260 | u32 gInc; // Increment along scanline for gCol |
| 261 | |
| 262 | // Color for flat-shaded, texture-blended prims |
| 263 | u8 r5, g5, b5; // 5-bit light for undithered prims |
| 264 | u8 r8, g8, b8; // 8-bit light for dithered prims |
| 265 | |
| 266 | // Color for flat-shaded, untextured prims |
| 267 | u16 PixelData; // bgr555 color for untextured flat-shaded polys |
| 268 | |
| 269 | // End of inner Loop parameters |
| 270 | //////////////////////////////////////////////////////////////////////////// |
| 271 | |
| 272 | |
| 273 | u8 blit_mask; // Determines what pixels to skip when rendering. |
| 274 | // Only useful on low-resolution devices using |
| 275 | // a simple pixel-dropping downscaler for PS1 |
| 276 | // high-res modes. See 'pixel_skip' option. |
| 277 | |
| 278 | u8 ilace_mask; // Determines what lines to skip when rendering. |
| 279 | // Normally 0 when PS1 240 vertical res is in |
| 280 | // use and ilace_force is 0. When running in |
| 281 | // PS1 480 vertical res on a low-resolution |
| 282 | // device (320x240), will usually be set to 1 |
| 283 | // so odd lines are not rendered. (Unless future |
| 284 | // full-screen scaling option is in use ..TODO) |
| 285 | |
| 286 | bool prog_ilace_flag; // Tracks successive frames for 'prog_ilace' option |
| 287 | |
| 288 | u8 BLEND_MODE; |
| 289 | u8 TEXT_MODE; |
| 290 | u8 Masking; |
| 291 | |
| 292 | u16 PixelMSB; |
| 293 | |
| 294 | gpu_unai_config_t config; |
| 295 | |
| 296 | u8 LightLUT[32*32]; // 5-bit lighting LUT (gpu_inner_light.h) |
| 297 | u32 DitherMatrix[64]; // Matrix of dither coefficients |
| 298 | }; |
| 299 | |
| 300 | static gpu_unai_t gpu_unai; |
| 301 | |
| 302 | // Global config that frontend can alter.. Values are read in GPU_init(). |
| 303 | // TODO: if frontend menu modifies a setting, add a function that can notify |
| 304 | // GPU plugin to use new setting. |
| 305 | gpu_unai_config_t gpu_unai_config_ext; |
| 306 | |
| 307 | /////////////////////////////////////////////////////////////////////////////// |
| 308 | // Internal inline funcs to get option status: (Allows flexibility) |
| 309 | static inline bool LightingEnabled() |
| 310 | { |
| 311 | return gpu_unai.config.lighting; |
| 312 | } |
| 313 | |
| 314 | static inline bool FastLightingEnabled() |
| 315 | { |
| 316 | return gpu_unai.config.fast_lighting; |
| 317 | } |
| 318 | |
| 319 | static inline bool BlendingEnabled() |
| 320 | { |
| 321 | return gpu_unai.config.blending; |
| 322 | } |
| 323 | |
| 324 | static inline bool DitheringEnabled() |
| 325 | { |
| 326 | return gpu_unai.config.dithering; |
| 327 | } |
| 328 | |
| 329 | // For now, this is just for development/experimentation purposes.. |
| 330 | // If modified to return true, it will allow ignoring the status register |
| 331 | // bit 9 setting (dither enable). It will still restrict dithering only |
| 332 | // to Gouraud-shaded or texture-blended polys. |
| 333 | static inline bool ForcedDitheringEnabled() |
| 334 | { |
| 335 | return false; |
| 336 | } |
| 337 | |
| 338 | static inline bool ProgressiveInterlaceEnabled() |
| 339 | { |
| 340 | #ifdef USE_GPULIB |
| 341 | // Using this old option greatly decreases quality of image. Disabled |
| 342 | // for now when using new gpulib, since it also adds more work in loops. |
| 343 | return false; |
| 344 | #else |
| 345 | return gpu_unai.config.prog_ilace; |
| 346 | #endif |
| 347 | } |
| 348 | |
| 349 | // For now, 320x240 output resolution is assumed, using simple line-skipping |
| 350 | // and pixel-skipping downscaler. |
| 351 | // TODO: Flesh these out so they return useful values based on whether |
| 352 | // running on higher-res device or a resampling downscaler is enabled. |
| 353 | static inline bool PixelSkipEnabled() |
| 354 | { |
| 355 | return gpu_unai.config.pixel_skip || gpu_unai.config.scale_hires; |
| 356 | } |
| 357 | |
| 358 | static inline bool LineSkipEnabled() |
| 359 | { |
| 360 | return true; |
| 361 | } |
| 362 | |
| 363 | #endif // GPU_UNAI_H |