1 /***************************************************************************
2 * Copyright (C) 2010 PCSX4ALL Team *
3 * Copyright (C) 2010 Unai *
4 * Copyright (C) 2016 Senquack (dansilsby <AT> gmail <DOT> com) *
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. *
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. *
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 ***************************************************************************/
27 // Header shared between both standalone gpu_senquack (gpu.cpp) and new
28 // gpulib-compatible gpu_senquack (gpulib_if.cpp)
29 // -> Anything here should be for gpu_senquack's private use. <-
31 ///////////////////////////////////////////////////////////////////////////////
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
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
46 #define GPU_INLINE static inline __attribute__((always_inline))
47 #define INLINE static inline __attribute__((always_inline))
78 template<class T> static inline void SwapValues(T &x, T &y)
80 T tmp(x); x = y; y = tmp;
84 static inline T Min2 (const T a, const T b)
90 static inline T Min3 (const T a, const T b, const T c)
92 return Min2(Min2(a,b),c);
96 static inline T Max2 (const T a, const T b)
102 static inline T Max3 (const T a, const T b, const T c)
104 return Max2(Max2(a,b),c);
108 ///////////////////////////////////////////////////////////////////////////////
111 // Convert 24bpp color parameter of GPU command to 16bpp (15bpp + mask bit)
112 #define GPU_RGB16(rgb) ((((rgb)&0xF80000)>>9)|(((rgb)&0xF800)>>6)|(((rgb)&0xF8)>>3))
114 // Sign-extend 11-bit coordinate command param
115 #define GPU_EXPANDSIGN(x) (((s32)(x)<<(32-11))>>(32-11))
117 // Max difference between any two X or Y primitive coordinates
118 #define CHKMAX_X 1024
121 #define FRAME_BUFFER_SIZE (1024*512*2)
122 #define FRAME_WIDTH 1024
123 #define FRAME_HEIGHT 512
124 #define FRAME_OFFSET(x,y) (((y)<<10)+(x))
125 #define FRAME_BYTE_STRIDE 2048
126 #define FRAME_BYTES_PER_PIXEL 2
128 static inline s32 GPU_DIV(s32 rs, s32 rt)
130 return rt ? (rs / rt) : (0);
133 // 'Unsafe' version of above that doesn't check for div-by-zero
134 #define GPU_FAST_DIV(rs, rt) ((signed)(rs) / (signed)(rt))
136 struct gpu_senquack_t {
138 GPUPacket PacketBuffer;
143 u32 tex_window; // Current texture window vals (set by GP0(E2h) cmd)
146 bool fb_dirty; // Framebuffer is dirty (according to GPU)
149 // NOTE: Standalone older gpu_senquack didn't care about horiz display range
150 u16 DisplayArea[6]; // [0] : Start of display area (in VRAM) X
151 // [1] : Start of display area (in VRAM) Y
152 // [2] : Display mode resolution HORIZONTAL
153 // [3] : Display mode resolution VERTICAL
154 // [4] : Vertical display range (on TV) START
155 // [5] : Vertical display range (on TV) END
157 ////////////////////////////////////////////////////////////////////////////
158 // Dma Transfers info
163 u32 *last_dma; // Last dma pointer
164 bool FrameToRead; // Load image in progress
165 bool FrameToWrite; // Store image in progress
168 ////////////////////////////////////////////////////////////////////////////
171 int skipCount; // Frame skip (0,1,2,3...)
172 bool isSkip; // Skip frame (according to GPU)
173 bool skipFrame; // Skip this frame (according to frame skip)
174 bool wasSkip; // Skip frame old value (according to GPU)
175 bool skipGPU; // Skip GPU primitives
178 // END of standalone gpu_senquack variables
179 ////////////////////////////////////////////////////////////////////////////
181 u32 TextureWindowCur; // Current setting from last GP0(0xE2) cmd (raw form)
182 u8 TextureWindow[4]; // [0] : Texture window offset X
183 // [1] : Texture window offset Y
184 // [2] : Texture window mask X
185 // [3] : Texture window mask Y
187 u16 DrawingArea[4]; // [0] : Drawing area top left X
188 // [1] : Drawing area top left Y
189 // [2] : Drawing area bottom right X
190 // [3] : Drawing area bottom right Y
192 s16 DrawingOffset[2]; // [0] : Drawing offset X (signed)
193 // [1] : Drawing offset Y (signed)
195 u16* TBA; // Ptr to current texture in VRAM
196 u16* CBA; // Ptr to current CLUT in VRAM
198 ////////////////////////////////////////////////////////////////////////////
199 // Inner Loop parameters
201 // 22.10 Fixed-pt texture coords, mask, scanline advance
202 // NOTE: U,V are no longer packed together into one u32, this proved to be
203 // too imprecise, leading to pixel dropouts. Example: NFS3's skybox.
208 // Color for Gouraud-shaded prims
209 // Packed fixed-pt 8.3:8.3:8.2 rgb triplet
210 // layout: rrrrrrrrXXXggggggggXXXbbbbbbbbXX
213 u32 gInc; // Increment along scanline for gCol
215 // Color for flat-shaded, texture-blended prims
216 u8 r5, g5, b5; // 5-bit light for undithered prims
217 u8 r8, g8, b8; // 8-bit light for dithered prims
219 // Color for flat-shaded, untextured prims
220 u16 PixelData; // bgr555 color for untextured flat-shaded polys
222 // End of inner Loop parameters
223 ////////////////////////////////////////////////////////////////////////////
226 u8 blit_mask; // Determines what pixels to skip when rendering.
227 // Only useful on low-resolution devices using
228 // a simple pixel-dropping downscaler for PS1
229 // high-res modes. See 'pixel_skip' option.
231 u8 ilace_mask; // Determines what lines to skip when rendering.
232 // Normally 0 when PS1 240 vertical res is in
233 // use and ilace_force is 0. When running in
234 // PS1 480 vertical res on a low-resolution
235 // device (320x240), will usually be set to 1
236 // so odd lines are not rendered. (Unless future
237 // full-screen scaling option is in use ..TODO)
239 bool prog_ilace_flag; // Tracks successive frames for 'prog_ilace' option
247 gpu_senquack_config_t config;
249 u8 LightLUT[32*32]; // 5-bit lighting LUT (gpu_inner_light.h)
250 u32 DitherMatrix[64]; // Matrix of dither coefficients
253 static gpu_senquack_t gpu_senquack;
255 // Global config that frontend can alter.. Values are read in GPU_init().
256 // TODO: if frontend menu modifies a setting, add a function that can notify
257 // GPU plugin to use new setting.
258 gpu_senquack_config_t gpu_senquack_config_ext;
260 ///////////////////////////////////////////////////////////////////////////////
261 // Internal inline funcs to get option status: (Allows flexibility)
262 static inline bool LightingEnabled()
264 return gpu_senquack.config.lighting;
267 static inline bool FastLightingEnabled()
269 return gpu_senquack.config.fast_lighting;
272 static inline bool BlendingEnabled()
274 return gpu_senquack.config.blending;
277 static inline bool DitheringEnabled()
279 return gpu_senquack.config.dithering;
282 // For now, this is just for development/experimentation purposes..
283 // If modified to return true, it will allow ignoring the status register
284 // bit 9 setting (dither enable). It will still restrict dithering only
285 // to Gouraud-shaded or texture-blended polys.
286 static inline bool ForcedDitheringEnabled()
291 static inline bool ProgressiveInterlaceEnabled()
294 // Using this old option greatly decreases quality of image. Disabled
295 // for now when using new gpulib, since it also adds more work in loops.
298 return gpu_senquack.config.prog_ilace;
302 // For now, 320x240 output resolution is assumed, using simple line-skipping
303 // and pixel-skipping downscaler.
304 // TODO: Flesh these out so they return useful values based on whether
305 // running on higher-res device or a resampling downscaler is enabled.
306 static inline bool PixelSkipEnabled()
308 return gpu_senquack.config.pixel_skip || gpu_senquack.config.scale_hires;
311 static inline bool LineSkipEnabled()