frontend: export fps stats to plugins
[pcsx_rearmed.git] / plugins / gpu_unai / gpu.cpp
index 5152327..5a0ad2c 100644 (file)
 #include "debug.h"
 
 int skipCount = 2; /* frame skip (0,1,2,3...) */
-int skCount=0; /* internal frame skip */
+int skCount = 0; /* internal frame skip */
 int linesInterlace = 0;  /* internal lines interlace */
 int linesInterlace_user = 0; /* Lines interlace */
 
-bool isSkip=false; /* skip frame (info coming from GPU) */
-bool skipFrame=false; /* skip frame (according to frame skip) */
+bool isSkip = false; /* skip frame (info coming from GPU) */
+bool wasSkip = false;
+bool skipFrame = false; /* skip frame (according to frame skip) */
 bool alt_fps = false; /* Alternative FPS algorithm */
 bool show_fps = false; /* Show FPS statistics */
 
-bool isPAL=false; /* PAL video timing */
+bool isPAL = false; /* PAL video timing */
 bool progressInterlace_flag = false; /* Progressive interlace flag */
 bool progressInterlace = false; /* Progressive interlace option*/
 bool frameLimit = false; /* frames to wait */
+
 bool light = true; /* lighting */
 bool blend = true; /* blending */
-
+bool FrameToRead = false; /* load image in progress */
+bool FrameToWrite = false; /* store image in progress */
 bool fb_dirty = false;
 
 bool enableAbbeyHack = false; /* Abe's Odyssey hack */
+
 u8 BLEND_MODE;
 u8 TEXT_MODE;
 u8 Masking;
@@ -60,9 +64,6 @@ s32           px,py;
 s32            x_end,y_end;
 u16*  pvram;
 
-u32 FrameToRead;
-u32 FrameToWrite;
-
 u32 GP0;
 s32 PacketCount;
 s32 PacketIndex;
@@ -95,7 +96,7 @@ u32   tInc, tMsk;
 
 GPUPacket PacketBuffer;
 // FRAME_BUFFER_SIZE is defined in bytes; 512K is guard memory for out of range reads
-u16   GPU_FrameBuffer[(FRAME_BUFFER_SIZE+512*1024)/2] __attribute__((aligned(16)));
+u16   GPU_FrameBuffer[(FRAME_BUFFER_SIZE+512*1024)/2] __attribute__((aligned(2048)));
 u32   GPU_GP1;
 
 ///////////////////////////////////////////////////////////////////////////////
@@ -273,8 +274,9 @@ void  GPU_writeDataMem(u32* dmaAddress, s32 dmaCount)
        {
                if (FrameToWrite) 
                {
-                       while (dmaCount--) 
+                       while (dmaCount)
                        {
+                               dmaCount--;
                                data = *dmaAddress++;
                                if ((&pvram[px])>(VIDEO_END)) pvram-=512*1024;
                                pvram[px] = data;
@@ -284,7 +286,7 @@ void  GPU_writeDataMem(u32* dmaAddress, s32 dmaCount)
                                        pvram += 1024;
                                        if (++py>=y_end) 
                                        {
-                                               FrameToWrite = 0;
+                                               FrameToWrite = false;
                                                GPU_GP1 &= ~0x08000000;
                                                break;
                                        }
@@ -297,7 +299,7 @@ void  GPU_writeDataMem(u32* dmaAddress, s32 dmaCount)
                                        pvram += 1024;
                                        if (++py>=y_end) 
                                        {
-                                               FrameToWrite = 0;
+                                               FrameToWrite = false;
                                                GPU_GP1 &= ~0x08000000;
                                                break;
                                        }
@@ -331,7 +333,7 @@ INLINE int CheckForEndlessLoop(u32 *laddr)
 }
 
 ///////////////////////////////////////////////////////////////////////////////
-void  GPU_dmaChain(u32* baseAddr, u32 dmaVAddr)
+long GPU_dmaChain(u32* baseAddr, u32 dmaVAddr)
 {
 #ifdef DEBUG_ANALYSIS
        dbg_anacnt_GPU_dmaChain++;
@@ -339,6 +341,7 @@ void  GPU_dmaChain(u32* baseAddr, u32 dmaVAddr)
        pcsx4all_prof_start_with_pause(PCSX4ALL_PROF_GPU,PCSX4ALL_PROF_HW_WRITE);
        u32 data, *address, count, offset;
        unsigned int DMACommandCounter = 0;
+       long dma_words = 0;
 
        GPU_GP1 &= ~0x14000000;
        lUsedAddr[0]=lUsedAddr[1]=lUsedAddr[2]=(u32*)0x1fffff;
@@ -355,9 +358,12 @@ void  GPU_dmaChain(u32* baseAddr, u32 dmaVAddr)
                else dmaVAddr = 0x1FFFFF;
 
                if(count>0) GPU_writeDataMem(address,count);
+               dma_words += 1 + count;
        }
        GPU_GP1 = (GPU_GP1 | 0x14000000) & ~0x60000000;
        pcsx4all_prof_end_with_resume(PCSX4ALL_PROF_GPU,PCSX4ALL_PROF_HW_WRITE);
+
+       return dma_words;
 }
 
 ///////////////////////////////////////////////////////////////////////////////
@@ -381,7 +387,7 @@ void  GPU_writeData(u32 data)
                        pvram += 1024;
                        if (++py>=y_end) 
                        {
-                               FrameToWrite = 0;
+                               FrameToWrite = false;
                                GPU_GP1 &= ~0x08000000;
                        }
                }
@@ -395,7 +401,7 @@ void  GPU_writeData(u32 data)
                                pvram += 1024;
                                if (++py>=y_end) 
                                {
-                                       FrameToWrite = 0;
+                                       FrameToWrite = false;
                                        GPU_GP1 &= ~0x08000000;
                                }
                        }
@@ -448,7 +454,7 @@ void  GPU_readDataMem(u32* dmaAddress, s32 dmaCount)
                        pvram += 1024;
                        if (++py>=y_end) 
                        {
-                               FrameToRead = 0;
+                               FrameToRead = false;
                                GPU_GP1 &= ~0x08000000;
                                break;
                        }
@@ -481,7 +487,7 @@ u32  GPU_readData(void)
                        pvram += 1024;
                        if (++py>=y_end) 
                        {
-                               FrameToRead = 0;
+                               FrameToRead = false;
                                GPU_GP1 &= ~0x08000000;
                        }
                }
@@ -493,7 +499,7 @@ u32  GPU_readData(void)
                        pvram +=1024;
                        if (++py>=y_end) 
                        {
-                               FrameToRead = 0;
+                               FrameToRead = false;
                                GPU_GP1 &= ~0x08000000;
                        }
                }
@@ -529,11 +535,11 @@ void  GPU_writeStatus(u32 data)
                break;
        case 0x01:
                GPU_GP1 &= ~0x08000000;
-               PacketCount = FrameToRead = FrameToWrite = 0;
+               PacketCount = 0; FrameToRead = FrameToWrite = false;
                break;
        case 0x02:
                GPU_GP1 &= ~0x08000000;
-               PacketCount = FrameToRead = FrameToWrite = 0;
+               PacketCount = 0; FrameToRead = FrameToWrite = false;
                break;
        case 0x03:
                GPU_GP1 = (GPU_GP1 & ~0x00800000) | ((data & 1) << 23);
@@ -547,6 +553,11 @@ void  GPU_writeStatus(u32 data)
                DisplayArea[0] = (data & 0x000003FF); //(short)(data & 0x3ff);
                DisplayArea[1] = ((data & 0x0007FC00)>>10); //(data & 0x000FFC00) >> 10; //(short)((data>>10)&0x1ff);
                fb_dirty = true;
+               wasSkip = isSkip;
+               if (isSkip)
+                       isSkip = false;
+               else
+                       isSkip = skipFrame;
                break;
        case 0x07:
                DisplayArea[4] = data & 0x000003FF; //(short)(data & 0x3ff);
@@ -634,15 +645,16 @@ INLINE void gpuVideoOutput(void)
        int incY = (h0==480) ? 2 : 1;
        h0=(h0==480 ? 2048 : 1024);
 
-       if (!progressInterlace)
        {
                const int li=linesInterlace;
+               bool pi=progressInterlace;
+               bool pif=progressInterlace_flag;
                switch ( w0 )
                {
                        case 256:
                                for(int y1=y0+h1; y0<y1; y0+=incY)
                                {
-                                       if( 0 == (y0&li) ) GPU_BlitWWDWW(       src_screen16,   dest_screen16, isRGB24);
+                                       if(( 0 == (y0&li) ) && ((!pi) || (pif=!pif))) GPU_BlitWWDWW(    src_screen16,   dest_screen16, isRGB24);
                                        dest_screen16 += VIDEO_WIDTH;
                                        src_screen16  += h0;
                                }
@@ -650,7 +662,7 @@ INLINE void gpuVideoOutput(void)
                        case 368:
                                for(int y1=y0+h1; y0<y1; y0+=incY)
                                {
-                                       if( 0 == (y0&li) ) GPU_BlitWWWWWWWWS(   src_screen16,   dest_screen16, isRGB24, 4);
+                                       if(( 0 == (y0&li) ) && ((!pi) || (pif=!pif))) GPU_BlitWWWWWWWWS(        src_screen16,   dest_screen16, isRGB24, 4);
                                        dest_screen16 += VIDEO_WIDTH;
                                        src_screen16  += h0;
                                }
@@ -658,7 +670,7 @@ INLINE void gpuVideoOutput(void)
                        case 320:
                                for(int y1=y0+h1; y0<y1; y0+=incY)
                                {
-                                       if( 0 == (y0&li) ) GPU_BlitWW(  src_screen16,   dest_screen16, isRGB24);
+                                       if(( 0 == (y0&li) ) && ((!pi) || (pif=!pif))) GPU_BlitWW(       src_screen16,   dest_screen16, isRGB24);
                                        dest_screen16 += VIDEO_WIDTH;
                                        src_screen16  += h0;
                                }
@@ -666,7 +678,7 @@ INLINE void gpuVideoOutput(void)
                        case 384:
                                for(int y1=y0+h1; y0<y1; y0+=incY)
                                {
-                                       if( 0 == (y0&li) ) GPU_BlitWWWWWS(      src_screen16,   dest_screen16, isRGB24);
+                                       if(( 0 == (y0&li) ) && ((!pi) || (pif=!pif))) GPU_BlitWWWWWS(   src_screen16,   dest_screen16, isRGB24);
                                        dest_screen16 += VIDEO_WIDTH;
                                        src_screen16  += h0;
                                }
@@ -674,7 +686,7 @@ INLINE void gpuVideoOutput(void)
                        case 512:
                                for(int y1=y0+h1; y0<y1; y0+=incY)
                                {
-                                       if( 0 == (y0&li) ) GPU_BlitWWSWWSWS(    src_screen16, dest_screen16, isRGB24);
+                                       if(( 0 == (y0&li) ) && ((!pi) || (pif=!pif))) GPU_BlitWWSWWSWS( src_screen16, dest_screen16, isRGB24);
                                        dest_screen16 += VIDEO_WIDTH;
                                        src_screen16  += h0;
                                }
@@ -682,69 +694,13 @@ INLINE void gpuVideoOutput(void)
                        case 640:
                                for(int y1=y0+h1; y0<y1; y0+=incY)
                                {
-                                       if( 0 == (y0&li) ) GPU_BlitWS(  src_screen16, dest_screen16, isRGB24);
+                                       if(( 0 == (y0&li) ) && ((!pi) || (pif=!pif))) GPU_BlitWS(       src_screen16, dest_screen16, isRGB24);
                                        dest_screen16 += VIDEO_WIDTH;
                                        src_screen16  += h0;
                                }
                                break;
                }
-       }
-       else
-       {
-               const int li=linesInterlace;
-               bool flag=progressInterlace_flag; /* progressive interlace */
-               switch ( w0 )
-               {
-                       case 256:
-                               for(int y1=y0+h1; y0<y1; y0+=incY)
-                               {
-                                       if(( 0 == (y0&li) ) && (flag=!flag)) GPU_BlitWWDWW(     src_screen16,   dest_screen16, isRGB24);
-                                       dest_screen16 += VIDEO_WIDTH;
-                                       src_screen16  += h0;
-                               }
-                               break;
-                       case 368:
-                               for(int y1=y0+h1; y0<y1; y0+=incY)
-                               {
-                                       if(( 0 == (y0&li) ) && (flag=!flag)) GPU_BlitWWWWWWWWS( src_screen16,   dest_screen16, isRGB24, 4);
-                                       dest_screen16 += VIDEO_WIDTH;
-                                       src_screen16  += h0;
-                               }
-                               break;
-                       case 320:
-                               for(int y1=y0+h1; y0<y1; y0+=incY)
-                               {
-                                       if(( 0 == (y0&li) ) && (flag=!flag)) GPU_BlitWW(        src_screen16,   dest_screen16, isRGB24);
-                                       dest_screen16 += VIDEO_WIDTH;
-                                       src_screen16  += h0;
-                               }
-                               break;
-                       case 384:
-                               for(int y1=y0+h1; y0<y1; y0+=incY)
-                               {
-                                       if(( 0 == (y0&li) ) && (flag=!flag)) GPU_BlitWWWWWS(    src_screen16,   dest_screen16, isRGB24);
-                                       dest_screen16 += VIDEO_WIDTH;
-                                       src_screen16  += h0;
-                               }
-                               break;
-                       case 512:
-                               for(int y1=y0+h1; y0<y1; y0+=incY)
-                               {
-                                       if(( 0 == (y0&li) ) && (flag=!flag)) GPU_BlitWWSWWSWS(  src_screen16, dest_screen16, isRGB24);
-                                       dest_screen16 += VIDEO_WIDTH;
-                                       src_screen16  += h0;
-                               }
-                               break;
-                       case 640:
-                               for(int y1=y0+h1; y0<y1; y0+=incY)
-                               {
-                                       if(( 0 == (y0&li) ) && (flag=!flag)) GPU_BlitWS(        src_screen16, dest_screen16, isRGB24);
-                                       dest_screen16 += VIDEO_WIDTH;
-                                       src_screen16  += h0;
-                               }
-                               break;
-               }
-               progressInterlace_flag=!flag;
+               progressInterlace_flag=!progressInterlace_flag;
        }
        video_flip();
 }
@@ -793,7 +749,7 @@ void  GPU_updateLace(void)
        GPU_GP1 ^= 0x80000000;
 
        // Update display
-       if ((!skipFrame) && (!isSkip) && (!(((GPU_GP1&0x08000000))||((GPU_GP1&0x00800000)))))
+       if ((!skipFrame) && (!isSkip) && (fb_dirty) && (!(((GPU_GP1&0x08000000))||((GPU_GP1&0x00800000)))))
        {
                gpuVideoOutput(); // Display updated
 
@@ -807,18 +763,18 @@ void  GPU_updateLace(void)
                        linesInterlace = linesInterlace_user; // resolution changed from 480 to lower one
                        video_clear();
                }
+       }
 
-               // Limit FPS
-               if (frameLimit)
+       // Limit FPS
+       if (frameLimit)
+       {
+               static unsigned next=get_ticks();
+               if (!skipFrame)
                {
-                       static unsigned next=get_ticks();
                        unsigned now=get_ticks();
-                       if (!skipFrame)
-                       {
-                               if (now<next) wait_ticks(next-now);
-                       }
-                       next+=(isPAL?(1000000/50):((unsigned)(1000000.0/59.94)));
+                       if (now<next) wait_ticks(next-now);
                }
+               next+=(isPAL?(1000000/50):((unsigned)(1000000.0/59.94)));
        }
 
        // Show FPS statistics
@@ -826,7 +782,7 @@ void  GPU_updateLace(void)
        {
                static u32 real_fps=0;
                static u32 prev=get_ticks();
-               static char msg[24]="FPS=000/00 SPD=000%";
+               static char msg[32]="FPS=000/00 SPD=000%";
                u32 now=get_ticks();
                real_fps++;
                if ((now-prev)>=1000000)
@@ -857,6 +813,7 @@ void  GPU_updateLace(void)
                        isSkip=true;
                }
        }
+       fb_dirty=false;
 
        pcsx4all_prof_end_with_resume(PCSX4ALL_PROF_GPU,PCSX4ALL_PROF_COUNTERS);
 }
@@ -864,24 +821,25 @@ void  GPU_updateLace(void)
 #else
 
 #include "../../frontend/plugin_lib.h"
-#include "../../frontend/arm_utils.h"
+#include "../gpulib/cspace.h"
 
 extern "C" {
 
 static const struct rearmed_cbs *cbs;
 static void *screen_buf;
+static s16 old_res_horz, old_res_vert, old_rgb24;
 
 static void blit(void)
 {
-       static s16 old_res_horz, old_res_vert, old_rgb24;
+       u16 *base = (u16 *)GPU_FrameBuffer;
        s16 isRGB24 = (GPU_GP1 & 0x00200000) ? 1 : 0;
        s16 h0, x0, y0, w0, h1;
-       u16 *srcs;
+       u32 fb_offs;
        u8  *dest;
 
-       x0 = DisplayArea[0] & ~3; // alignment needed by blitter
+       x0 = DisplayArea[0] & ~1; // alignment needed by blitter
        y0 = DisplayArea[1];
-       srcs = &((u16*)GPU_FrameBuffer)[FRAME_OFFSET(x0,y0)];
+       fb_offs = FRAME_OFFSET(x0, y0);
 
        w0 = DisplayArea[2];
        h0 = DisplayArea[3];  // video mode
@@ -897,33 +855,59 @@ static void blit(void)
                old_res_horz = w0;
                old_res_vert = h1;
                old_rgb24 = (s16)isRGB24;
-               screen_buf = cbs->pl_fbdev_set_mode(w0, h1, isRGB24 ? 24 : 16);
+               screen_buf = cbs->pl_vout_set_mode(w0, h1, isRGB24 ? 24 : 16);
        }
        dest = (u8 *)screen_buf;
 
        if (isRGB24)
        {
 #ifndef MAEMO
-               for (; h1-- > 0; dest += w0 * 3, srcs += 1024)
+               for (; h1-- > 0; dest += w0 * 3, fb_offs += 1024)
                {
-                       bgr888_to_rgb888(dest, srcs, w0 * 3);
+                       fb_offs &= 1024*512-1;
+                       bgr888_to_rgb888(dest, base + fb_offs, w0 * 3);
                }
 #else
-               for (; h1-- > 0; dest += w0 * 2, srcs += 1024)
+               for (; h1-- > 0; dest += w0 * 2, fb_offs += 1024)
                {
-                       bgr888_to_rgb565(dest, srcs, w0 * 3);
+                       fb_offs &= 1024*512-1;
+                       bgr888_to_rgb565(dest, base + fb_offs, w0 * 3);
                }
 #endif
        }
        else
        {
-               for (; h1-- > 0; dest += w0 * 2, srcs += 1024)
+               for (; h1-- > 0; dest += w0 * 2, fb_offs += 1024)
                {
-                       bgr555_to_rgb565(dest, srcs, w0 * 2);
+                       fb_offs &= 1024*512-1;
+                       bgr555_to_rgb565(dest, base + fb_offs, w0 * 2);
                }
        }
 
-       screen_buf = cbs->pl_fbdev_flip();
+       screen_buf = cbs->pl_vout_flip();
+}
+
+static void blit_raw(void)
+{
+       s16 isRGB24 = (GPU_GP1 & 0x00200000) ? 1 : 0;
+       s16 h0, w0, h1;
+
+       w0 = DisplayArea[2];
+       h0 = DisplayArea[3];  // video mode
+       h1 = DisplayArea[5] - DisplayArea[4]; // display needed
+       if (h0 == 480) h1 = Min2(h1*2,480);
+
+       if (h1 <= 0)
+               return;
+
+       if (w0 != old_res_horz || h1 != old_res_vert || isRGB24 != old_rgb24)
+       {
+               old_res_horz = w0;
+               old_res_vert = h1;
+               old_rgb24 = (s16)isRGB24;
+               screen_buf = cbs->pl_vout_set_mode(w0, h1, isRGB24 ? 24 : 16);
+       }
+       cbs->pl_vout_raw_flip(DisplayArea[0], DisplayArea[1]);
 }
 
 void GPU_updateLace(void)
@@ -934,27 +918,33 @@ void GPU_updateLace(void)
        if (!fb_dirty || (GPU_GP1&0x08800000))
                return;
 
-       if (!isSkip) {
-               blit();
-
+       if (!wasSkip) {
+               if (cbs->pl_vout_raw_flip != NULL)
+                       blit_raw();
+               else
+                       blit();
                fb_dirty = false;
-               if (*cbs->fskip_option)
-                       isSkip = true;
+               skCount = 0;
        }
-       else
-               isSkip = false;
+       else {
+               skCount++;
+               if (skCount >= 8)
+                       wasSkip = isSkip = 0;
+       }
+
+       skipFrame = cbs->fskip_advice || cbs->frameskip == 1;
 }
 
 long GPUopen(unsigned long *, char *, char *)
 {
-       cbs->pl_fbdev_open();
-       screen_buf = cbs->pl_fbdev_flip();
+       cbs->pl_vout_open();
+       screen_buf = cbs->pl_vout_flip();
        return 0;
 }
 
 long GPUclose(void)
 {
-       cbs->pl_fbdev_close();
+       cbs->pl_vout_close();
        return 0;
 }
 
@@ -968,6 +958,12 @@ long GPUfreeze(unsigned int ulGetFreezeData, GPUFreeze_t* p2)
 
 void GPUrearmedCallbacks(const struct rearmed_cbs *cbs_)
 {
+       enableAbbeyHack = cbs_->gpu_unai.abe_hack;
+       light = !cbs_->gpu_unai.no_light;
+       blend = !cbs_->gpu_unai.no_blend;
+       if (cbs_->pl_vout_set_raw_vram)
+               cbs_->pl_vout_set_raw_vram((void *)GPU_FrameBuffer);
+
        cbs = cbs_;
 }