Merge remote-tracking branch 'origin/fh/mymaster' into fh/oit

operands of the ?: operator"

.gitignore

@@ -48,6 +48,7 @@ reicast-ios.xccheckout
 shell/linux/.map
 shell/linux/nosym-reicast.elf
 shell/linux/reicast.elf
+shell/linux/reicast_naomi.elf
 
 # Visual Studio
 generated

core/hw/pvr/Renderer_if.cpp

@@ -103,7 +103,7 @@ void dump_frame(const char* file, TA_context* ctx, u8* vram, u8* vram_ref = NULL
 
 	u32 bytes = ctx->tad.End() - ctx->tad.thd_root;
 
-	fwrite("TAFRAME3", 1, 8, fw);
+	fwrite("TAFRAME4", 1, 8, fw);
 
 	fwrite(&ctx->rend.isRTT, 1, sizeof(ctx->rend.isRTT), fw);
 	u32 zero = 0;
@@ -168,10 +168,17 @@ TA_context* read_frame(const char* file, u8* vram_ref = NULL) {
 
 	fread(id0, 1, 8, fw);
 
-	if (memcmp(id0, "TAFRAME3", 8) != 0) {
+	if (memcmp(id0, "TAFRAME", 7) != 0 || (id0[7] != '3' && id0[7] != '4')) {
 		fclose(fw);
 		return 0;
 	}
+	int sizeofPolyParam = sizeof(PolyParam);
+	int sizeofVertex = sizeof(Vertex);
+	if (id0[7] == '3')
+	{
+		sizeofPolyParam -= 12;
+		sizeofVertex -= 16;
+	}
 
 	TA_context* ctx = tactx_Alloc();
 
@@ -184,8 +191,10 @@ TA_context* read_frame(const char* file, u8* vram_ref = NULL) {
 	fread(&ctx->rend.fb_X_CLIP.full, 1, sizeof(ctx->rend.fb_X_CLIP.full), fw);
 	fread(&ctx->rend.fb_Y_CLIP.full, 1, sizeof(ctx->rend.fb_Y_CLIP.full), fw);
 
-	fread(ctx->rend.global_param_op.Append(), 1, sizeof(PolyParam), fw);
-	fread(ctx->rend.verts.Append(4), 1, 4 * sizeof(Vertex), fw);
+	fread(ctx->rend.global_param_op.Append(), 1, sizeofPolyParam, fw);
+	Vertex *vtx = ctx->rend.verts.Append(4);
+	for (int i = 0; i < 4; i++)
+		fread(vtx + i, 1, sizeofVertex, fw);
 
 	fread(&t, 1, sizeof(t), fw);
 	verify(t == VRAM_SIZE);

core/hw/pvr/helper_classes.h

@@ -8,6 +8,7 @@ struct List
 
 	int size;
 	bool* overrun;
+	const char *list_name;
 
 	__forceinline int used() const { return size-avail; }
 	__forceinline int bytes() const { return used()* sizeof(T); }
@@ -17,6 +18,8 @@ struct List
 	{ 
 		*overrun |= true;
 		Clear();
+		if (list_name != NULL)
+			printf("List overrun for list %s\n", list_name);
 
 		return daty;
 	}
@@ -45,7 +48,7 @@ struct List
 
 	T* head() const { return daty-used(); }
 
-	void InitBytes(int maxbytes,bool* ovrn)
+	void InitBytes(int maxbytes,bool* ovrn, const char *name)
 	{
 		maxbytes-=maxbytes%sizeof(T);
 
@@ -58,11 +61,12 @@ struct List
 		overrun=ovrn;
 
 		Clear();
+		list_name = name;
 	}
 
-	void Init(int maxsize,bool* ovrn)
+	void Init(int maxsize,bool* ovrn, const char *name)
 	{
-		InitBytes(maxsize*sizeof(T),ovrn);
+		InitBytes(maxsize*sizeof(T),ovrn, name);
 	}
 
 	void Clear()
@@ -76,4 +80,4 @@ struct List
 		Clear();
 		free(daty);
 	}
-};
+};

core/hw/pvr/pvr_mem.cpp

@@ -37,9 +37,9 @@ void YUV_init()
 
 	YUV_dest=TA_YUV_TEX_BASE&VRAM_MASK;//TODO : add the masking needed
 	TA_YUV_TEX_CNT=0;
-	YUV_blockcount=(((TA_YUV_TEX_CTRL>>0)&0x3F)+1)*(((TA_YUV_TEX_CTRL>>8)&0x3F)+1);
+	YUV_blockcount = (TA_YUV_TEX_CTRL.yuv_u_size + 1) * (TA_YUV_TEX_CTRL.yuv_v_size + 1);
 
-	if ((TA_YUV_TEX_CTRL>>16 )&1)
+	if (TA_YUV_TEX_CTRL.yuv_tex != 0)
 	{
 		die ("YUV: Not supported configuration\n");
 		YUV_x_size=16;
@@ -47,8 +47,8 @@ void YUV_init()
 	}
 	else // yesh!!!
 	{
-		YUV_x_size=(((TA_YUV_TEX_CTRL>>0)&0x3F)+1)*16;
-		YUV_y_size=(((TA_YUV_TEX_CTRL>>8)&0x3F)+1)*16;
+		YUV_x_size = (TA_YUV_TEX_CTRL.yuv_u_size + 1) * 16;
+		YUV_y_size = (TA_YUV_TEX_CTRL.yuv_v_size + 1) * 16;
 	}
 }
 
@@ -164,7 +164,7 @@ void YUV_data(u32* data , u32 count)
 		YUV_init();
 	}
 
-	u32 block_size=(TA_YUV_TEX_CTRL & (1<<24))==0?384:512;
+	u32 block_size = TA_YUV_TEX_CTRL.yuv_form == 0 ? 384 : 512;
 
 	verify(block_size==384); //no support for 512
 

core/hw/pvr/pvr_regs.h

@@ -370,7 +370,22 @@ union TA_GLOB_TILE_CLIP_type
 	};
 	u32 full;
 };
- 
+
+union TA_YUV_TEX_CTRL_type
+{
+	struct
+	{
+		u32 yuv_u_size	: 6;
+		u32 reserved1	: 2;
+		u32 yuv_v_size	: 6;
+		u32 reserved2	: 2;
+		u32 yuv_tex		: 1;
+		u32 reserved3	: 7;
+		u32 yuv_form	: 1;
+		u32 reserved4	: 7;
+	};
+	u32 full;
+};
 
 // TA REGS
 #define TA_OL_BASE_addr         0x00000124 // RW  Object list write start address
@@ -483,7 +498,7 @@ union TA_GLOB_TILE_CLIP_type
 #define TA_ALLOC_CTRL     PvrReg(TA_ALLOC_CTRL_addr,u32)     // RW Object list control
 #define TA_LIST_INIT      PvrReg(TA_LIST_INIT_addr,u32)      // RW TA initialization
 #define TA_YUV_TEX_BASE   PvrReg(TA_YUV_TEX_BASE_addr,u32)   // RW YUV422 texture write start address
-#define TA_YUV_TEX_CTRL   PvrReg(TA_YUV_TEX_CTRL_addr,u32)   // RW YUV converter control
+#define TA_YUV_TEX_CTRL   PvrReg(TA_YUV_TEX_CTRL_addr, TA_YUV_TEX_CTRL_type)   // RW YUV converter control
 #define TA_YUV_TEX_CNT    PvrReg(TA_YUV_TEX_CNT_addr,u32)    // R  YUV converter macro block counter value
 
 #define TA_LIST_CONT      PvrReg(TA_LIST_CONT_addr,u32)      // RW TA continuation processing

core/hw/pvr/ta.h

@@ -23,10 +23,10 @@ bool ta_parse_vdrc(TA_context* ctx);
 
 
 #define STRIPS_AS_PPARAMS 1
-#define TRIG_SORT 1
+#define TRIG_SORT 0
 
 
 #if TRIG_SORT 
 #undef STRIPS_AS_PPARAMS
 #define STRIPS_AS_PPARAMS 1
-#endif
+#endif

core/hw/pvr/ta_ctx.h

@@ -17,6 +17,12 @@ struct Vertex
 	u8 spc[4];
 
 	float u,v;
+
+	// Two volumes format
+	u8 col1[4];
+	u8 spc1[4];
+
+	float u1,v1;
 };
 
 struct PolyParam
@@ -35,6 +41,9 @@ struct PolyParam
 	float zvZ;
 	u32 tileclip;
 	//float zMin,zMax;
+	TSP tsp1;
+	TCW tcw1;
+	u32 texid1;
 };
 
 struct ModifierVolumeParam
@@ -98,6 +107,7 @@ struct RenderPass {
 	u32 mvo_count;
 	u32 pt_count;
 	u32 tr_count;
+	u32 mvo_tr_count;
 };
 
 struct rend_context
@@ -124,6 +134,7 @@ struct rend_context
 	List<u16>         idx;
 	List<ModTriangle> modtrig;
 	List<ModifierVolumeParam>  global_param_mvo;
+	List<ModifierVolumeParam>  global_param_mvo_tr;
 
 	List<PolyParam>   global_param_op;
 	List<PolyParam>   global_param_pt;
@@ -139,6 +150,7 @@ struct rend_context
 		global_param_tr.Clear();
 		modtrig.Clear();
 		global_param_mvo.Clear();
+		global_param_mvo_tr.Clear();
 		render_passes.Clear();
 
 		Overrun=false;
@@ -189,16 +201,23 @@ struct TA_context
 	{
 		tad.Reset((u8*)OS_aligned_malloc(32, 8*1024*1024));
 
-		rend.verts.InitBytes(2*1024*1024,&rend.Overrun); //up to 2 MB of vtx data/frame = ~ 75k vtx/frame
-		rend.idx.Init(120*1024,&rend.Overrun);			//up to 120K indexes ( idx have stripification overhead )
-		rend.global_param_op.Init(4096,&rend.Overrun);
-		rend.global_param_pt.Init(4096,&rend.Overrun);
-		rend.global_param_mvo.Init(4096,&rend.Overrun);
-		rend.global_param_tr.Init(8192,&rend.Overrun);
+		rend.verts.InitBytes(4 * 1024 * 1024, &rend.Overrun, "verts");	//up to 4 mb of vtx data/frame = ~ 96k vtx/frame
+		rend.idx.Init(120 * 1024, &rend.Overrun, "idx");				//up to 120K indexes ( idx have stripification overhead )
+		rend.global_param_op.Init(4096, &rend.Overrun, "global_param_op");
+		rend.global_param_pt.Init(4096, &rend.Overrun, "global_param_pt");
+		rend.global_param_mvo.Init(4096, &rend.Overrun, "global_param_mvo");
+#if STRIPS_AS_PPARAMS
+		// That makes a lot of polyparams but this is required for proper sorting...
+		// Rez uses more than 8192 translucent polygons sometimes
+		rend.global_param_tr.Init(10240, &rend.Overrun, "global_param_tr");
+#else
+		rend.global_param_tr.Init(4096, &rend.Overrun, "global_param_tr");
+#endif
+		rend.global_param_mvo_tr.Init(4096, &rend.Overrun, "global_param_mvo_tr");
 
-		rend.modtrig.Init(8192,&rend.Overrun);
+		rend.modtrig.Init(16384, &rend.Overrun, "modtrig");
 		
-		rend.render_passes.Init(sizeof(RenderPass) * 10, &rend.Overrun);	// 10 render passes
+		rend.render_passes.Init(sizeof(RenderPass) * 10, &rend.Overrun, "render_passes");	// 10 render passes
 
 		Reset();
 	}
@@ -222,6 +241,7 @@ struct TA_context
 		rend.global_param_tr.Free();
 		rend.modtrig.Free();
 		rend.global_param_mvo.Free();
+		rend.global_param_mvo_tr.Free();
 		rend.render_passes.Free();
 	}
 };

core/hw/pvr/ta_vtx.cpp

@@ -85,6 +85,8 @@ List<PolyParam>* CurrentPPlist;
 //TA state vars	
 DECL_ALIGN(4) u8 FaceBaseColor[4];
 DECL_ALIGN(4) u8 FaceOffsColor[4];
+DECL_ALIGN(4) u8 FaceBaseColor1[4];
+DECL_ALIGN(4) u8 FaceOffsColor1[4];
 DECL_ALIGN(4) u32 SFaceBaseColor;
 DECL_ALIGN(4) u32 SFaceOffsColor;
 
@@ -769,7 +771,7 @@ public:
 		CurrentPP=&nullPP;
 		CurrentPPlist=0;
 
-		if (ListType == ListType_Opaque_Modifier_Volume)
+		if (ListType == ListType_Opaque_Modifier_Volume || ListType == ListType_Translucent_Modifier_Volume)
 			EndModVol();
 	}
 
@@ -808,6 +810,9 @@ public:
 			if (d_pp->pcw.Texture) {
 				d_pp->texid = renderer->GetTexture(d_pp->tsp,d_pp->tcw);
 			}
+			d_pp->tsp1.full = -1;
+			d_pp->tcw1.full = -1;
+			d_pp->texid1 = -1;
 		}
 	}
 
@@ -860,6 +865,11 @@ public:
 		TA_PolyParam3* pp=(TA_PolyParam3*)vpp;
 
 		glob_param_bdc(pp);
+
+		CurrentPP->tsp1.full = pp->tsp1.full;
+		CurrentPP->tcw1.full = pp->tcw1.full;
+		if (pp->pcw.Texture)
+			CurrentPP->texid1 = renderer->GetTexture(pp->tsp1, pp->tcw1);
 	}
 	__forceinline
 		static void TACALL AppendPolyParam4A(void* vpp)
@@ -867,13 +877,19 @@ public:
 		TA_PolyParam4A* pp=(TA_PolyParam4A*)vpp;
 
 		glob_param_bdc(pp);
+
+		CurrentPP->tsp1.full = pp->tsp1.full;
+		CurrentPP->tcw1.full = pp->tcw1.full;
+		if (pp->pcw.Texture)
+			CurrentPP->texid1 = renderer->GetTexture(pp->tsp1, pp->tcw1);
 	}
 	__forceinline
 		static void TACALL AppendPolyParam4B(void* vpp)
 	{
 		TA_PolyParam4B* pp=(TA_PolyParam4B*)vpp;
 
-		poly_float_color(FaceBaseColor,FaceColor0);
+		poly_float_color(FaceBaseColor, FaceColor0);
+		poly_float_color(FaceBaseColor1, FaceColor1);
 	}
 
 	//Poly Strip handling
@@ -884,13 +900,6 @@ public:
 	{
 		CurrentPP->count=vdrc.idx.used() - CurrentPP->first;
 
-		int vbase=vdrc.verts.used();
-
-		*vdrc.idx.Append()=vbase-1;
-		*vdrc.idx.Append()=vbase;
-
-		if (CurrentPP->count&1)
-			*vdrc.idx.Append()=vbase;
 #if STRIPS_AS_PPARAMS
 		if (CurrentPPlist==&vdrc.global_param_tr)
 		{
@@ -900,7 +909,20 @@ public:
 			d_pp->first=vdrc.idx.used(); 
 			d_pp->count=0; 
 		}
+		else
+		{
 #endif
+			int vbase=vdrc.verts.used();
+
+			*vdrc.idx.Append()=vbase-1;
+			*vdrc.idx.Append()=vbase;
+
+			if (CurrentPP->count&1)
+				*vdrc.idx.Append()=vbase;
+#if STRIPS_AS_PPARAMS
+		}
+#endif
+
 	}
 
 
@@ -941,6 +963,14 @@ public:
 		cv->u = f16(vtx->u_name);\
 		cv->v = f16(vtx->v_name);
 
+	#define vert_uv1_32(u_name,v_name) \
+		cv->u1 = (vtx->u_name);\
+		cv->v1 = (vtx->v_name);
+
+	#define vert_uv1_16(u_name,v_name) \
+		cv->u1 = f16(vtx->u_name);\
+		cv->v1 = f16(vtx->v_name);
+
 		//Color conversions
 	#define vert_packed_color_(to,src) \
 		{ \
@@ -984,6 +1014,20 @@ public:
 		cv->spc[2] = FaceOffsColor[2]*satint/256;  \
 		cv->spc[3] = FaceOffsColor[3]; }
 
+	#define vert_face_base_color1(baseint) \
+		{ u32 satint=float_to_satu8(vtx->baseint); \
+		cv->col1[0] = FaceBaseColor1[0]*satint/256;  \
+		cv->col1[1] = FaceBaseColor1[1]*satint/256;  \
+		cv->col1[2] = FaceBaseColor1[2]*satint/256;  \
+		cv->col1[3] = FaceBaseColor1[3]; }
+
+	#define vert_face_offs_color1(offsint) \
+		{ u32 satint=float_to_satu8(vtx->offsint); \
+		cv->spc1[0] = FaceOffsColor1[0]*satint/256;  \
+		cv->spc1[1] = FaceOffsColor1[1]*satint/256;  \
+		cv->spc1[2] = FaceOffsColor1[2]*satint/256;  \
+		cv->spc1[3] = FaceOffsColor1[3]; }
+
 	//vert_float_color_(cv->spc,FaceOffsColor[3],FaceOffsColor[0]*satint/256,FaceOffsColor[1]*satint/256,FaceOffsColor[2]*satint/256); }
 
 
@@ -1109,6 +1153,7 @@ public:
 		vert_cvt_base;
 
 		vert_packed_color(col,BaseCol0);
+		vert_packed_color(col1, BaseCol1);
 	}
 
 	//(Non-Textured, Intensity,	with Two Volumes)
@@ -1118,6 +1163,7 @@ public:
 		vert_cvt_base;
 
 		vert_face_base_color(BaseInt0);
+		vert_face_base_color1(BaseInt1);
 	}
 
 	//(Textured, Packed Color,	with Two Volumes)	
@@ -1136,6 +1182,10 @@ public:
 	{
 		vert_res_base;
 
+		vert_packed_color(col1, BaseCol1);
+		vert_packed_color(spc1, OffsCol1);
+
+		vert_uv1_32(u1, v1);
 	}
 
 	//(Textured, Packed Color, 16bit UV, with Two Volumes)
@@ -1154,6 +1204,10 @@ public:
 	{
 		vert_res_base;
 
+		vert_packed_color(col1, BaseCol1);
+		vert_packed_color(spc1, OffsCol1);
+
+		vert_uv1_16(u1, v1);
 	}
 
 	//(Textured, Intensity,	with Two Volumes)
@@ -1172,6 +1226,10 @@ public:
 	{
 		vert_res_base;
 
+		vert_face_base_color1(BaseInt1);
+		vert_face_offs_color1(OffsInt1);
+
+		vert_uv1_32(u1,v1);
 	}
 
 	//(Textured, Intensity, 16bit UV, with Two Volumes)
@@ -1190,6 +1248,10 @@ public:
 	{
 		vert_res_base;
 
+		vert_face_base_color1(BaseInt1);
+		vert_face_offs_color1(OffsInt1);
+
+		vert_uv1_16(u1, v1);
 	}
 
 	//Sprites
@@ -1217,6 +1279,9 @@ public:
 		if (d_pp->pcw.Texture) {
 			d_pp->texid = renderer->GetTexture(d_pp->tsp,d_pp->tcw);
 		}
+		d_pp->tcw1.full = -1;
+		d_pp->tsp1.full = -1;
+		d_pp->texid1 = -1;
 
 		SFaceBaseColor=spr->BaseCol;
 		SFaceOffsColor=spr->OffsCol;
@@ -1374,6 +1439,8 @@ public:
 		List<ModifierVolumeParam> *list = NULL;
 		if (CurrentList == ListType_Opaque_Modifier_Volume)
 			list = &vdrc.global_param_mvo;
+		else if (CurrentList == ListType_Translucent_Modifier_Volume)
+			list = &vdrc.global_param_mvo_tr;
 		else
 			return;
 		if (list->used() > 0)
@@ -1390,6 +1457,8 @@ public:
 		ModifierVolumeParam *p = NULL;
 		if (CurrentList == ListType_Opaque_Modifier_Volume)
 			p = vdrc.global_param_mvo.Append();
+		else if (CurrentList == ListType_Translucent_Modifier_Volume)
+			p = vdrc.global_param_mvo_tr.Append();
 		else
 			return;
 		p->isp.full = param->isp.full;
@@ -1399,7 +1468,7 @@ public:
 	__forceinline
 		static void AppendModVolVertexA(TA_ModVolA* mvv)
 	{
-		if (CurrentList!=ListType_Opaque_Modifier_Volume)
+		if (CurrentList != ListType_Opaque_Modifier_Volume && CurrentList != ListType_Translucent_Modifier_Volume)
 			return;
 		lmr=vdrc.modtrig.Append();
 
@@ -1419,7 +1488,7 @@ public:
 	__forceinline
 		static void AppendModVolVertexB(TA_ModVolB* mvv)
 	{
-		if (CurrentList!=ListType_Opaque_Modifier_Volume)
+		if (CurrentList != ListType_Opaque_Modifier_Volume && CurrentList != ListType_Translucent_Modifier_Volume)
 			return;
 		lmr->y2=mvv->y2;
 		lmr->z2=mvv->z2;
@@ -1486,6 +1555,7 @@ bool ta_parse_vdrc(TA_context* ctx)
 			render_pass->mvo_count = vd_rc.global_param_mvo.used();
 			render_pass->pt_count = vd_rc.global_param_pt.used();
 			render_pass->tr_count = vd_rc.global_param_tr.used();
+			render_pass->mvo_tr_count = vd_rc.global_param_mvo_tr.used();
 			render_pass->autosort = UsingAutoSort(pass);
 			render_pass->z_clear = ClearZBeforePass(pass);
 		}
@@ -1637,6 +1707,9 @@ void FillBGP(TA_context* ctx)
 	bgpp->isp.full=vri(strip_base);
 	bgpp->tsp.full=vri(strip_base+4);
 	bgpp->tcw.full=vri(strip_base+8);
+	bgpp->tcw1.full = -1;
+	bgpp->tsp1.full = -1;
+	bgpp->texid1 = -1;
 	bgpp->count=4;
 	bgpp->first=0;
 	bgpp->tileclip=0;//disabled ! HA ~

core/linux-dist/x11.cpp

@@ -704,9 +704,11 @@ void x11_window_create()
 			verify(glXCreateContextAttribsARB != 0);
 			int context_attribs[] =
 			{
-				GLX_CONTEXT_MAJOR_VERSION_ARB, 3,
-				GLX_CONTEXT_MINOR_VERSION_ARB, 1,
+				GLX_CONTEXT_MAJOR_VERSION_ARB, 4,
+				GLX_CONTEXT_MINOR_VERSION_ARB, 3,
+#ifndef RELEASE
 				GLX_CONTEXT_FLAGS_ARB, GLX_CONTEXT_DEBUG_BIT_ARB,
+#endif
 				GLX_CONTEXT_PROFILE_MASK_ARB, GLX_CONTEXT_CORE_PROFILE_BIT_ARB,
 				None
 			};
@@ -716,7 +718,7 @@ void x11_window_create()
 
 			if (!x11_glc)
 			{
-				die("Failed to create GL3.1 context\n");
+				die("Failed to create OpenGL 4.3 context\n");
 			}
 		#endif
 

core/oslib/audiobackend_alsa.cpp

@@ -7,6 +7,8 @@ static bool pcm_blocking = true;
 static snd_pcm_uframes_t buffer_size;
 static snd_pcm_uframes_t period_size;
 
+#define MAX_LATENCY 100
+
 // We're making these functions static - there's no need to pollute the global namespace
 static void alsa_init()
 {
@@ -89,7 +91,7 @@ static void alsa_init()
 	}
 	else
 		printf("ALSA: period size set to %ld\n", period_size);
-	buffer_size = (44100 * 100 /* settings.omx.Audio_Latency */ / 1000 / period_size + 1) * period_size;
+	buffer_size = (44100 * MAX_LATENCY / 1000 / period_size + 1) * period_size;
 	rc=snd_pcm_hw_params_set_buffer_size_near(handle, params, &buffer_size);
 	if (rc < 0)
 	{

core/rend/TexCache.h

@@ -161,11 +161,7 @@ struct pp_8888
 {
 	__forceinline static u32 packRGB(u8 R,u8 G,u8 B)
 	{
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__ && defined(GLES)
-		return (R << 0) | (G << 8) | (B << 16) | 0xFF000000;
-#else
 		return (R << 24) | (G << 16) | (B << 8) | 0xFF;
-#endif
 	}
 };
 

core/rend/gles/abuffer.cpp

@@ -0,0 +1,545 @@
+/*
+ * abuffer.cpp
+ *
+ *  Created on: May 26, 2018
+ *      Author: raph
+ */
+#include "glcache.h"
+
+GLuint pixels_buffer;
+GLuint pixels_pointers;
+GLuint atomic_buffer;
+PipelineShader g_abuffer_final_shader;
+PipelineShader g_abuffer_final_nosort_shader;
+PipelineShader g_abuffer_clear_shader;
+PipelineShader g_abuffer_tr_modvol_shaders[ModeCount];
+static GLuint g_quadBuffer = 0;
+static GLuint g_quadVertexArray = 0;
+
+static int g_imageWidth = 0;
+static int g_imageHeight = 0;
+
+GLuint pixel_buffer_size = 512 * 1024 * 1024;	// Initial size 512 MB
+
+#define MAX_PIXELS_PER_FRAGMENT "32"
+
+static const char *final_shader_source = SHADER_HEADER "\
+#define DEPTH_SORTED %d \n\
+#define MAX_PIXELS_PER_FRAGMENT " MAX_PIXELS_PER_FRAGMENT " \n\
+ \n\
+layout(binding = 0) uniform sampler2D tex; \n\
+uniform highp float shade_scale_factor; \n\
+ \n\
+out vec4 FragColor; \n\
+ \n\
+uint pixel_list[MAX_PIXELS_PER_FRAGMENT]; \n\
+ \n\
+ \n\
+int fillAndSortFragmentArray(ivec2 coords) \n\
+{ \n\
+	// Load fragments into a local memory array for sorting \n\
+	uint idx = imageLoad(abufferPointerImg, coords).x; \n\
+	int count = 0; \n\
+	for (; idx != EOL && count < MAX_PIXELS_PER_FRAGMENT; count++) \n\
+	{ \n\
+		const Pixel p = pixels[idx]; \n\
+		int j = count - 1; \n\
+		Pixel jp = pixels[pixel_list[j]]; \n\
+#if DEPTH_SORTED == 1 \n\
+		while (j >= 0 \n\
+			   && (jp.depth < p.depth \n\
+				   || (jp.depth == p.depth && getPolyNumber(jp) > getPolyNumber(p)))) \n\
+#else \n\
+		while (j >= 0 && getPolyNumber(jp) > getPolyNumber(p)) \n\
+#endif \n\
+		{ \n\
+			pixel_list[j + 1] = pixel_list[j]; \n\
+			j--; \n\
+			jp = pixels[pixel_list[j]]; \n\
+		} \n\
+		pixel_list[j + 1] = idx; \n\
+		idx = p.next; \n\
+	} \n\
+	return count; \n\
+} \n\
+ \n\
+// Blend fragments back-to-front \n\
+vec4 resolveAlphaBlend(ivec2 coords) { \n\
+	 \n\
+	// Copy and sort fragments into a local array \n\
+	int num_frag = fillAndSortFragmentArray(coords); \n\
+	 \n\
+	vec4 finalColor = texture(tex, gl_FragCoord.xy / textureSize(tex, 0)); \n\
+	vec4 secondaryBuffer = vec4(0.0); // Secondary accumulation buffer \n\
+	float depth = 1.0; \n\
+	 \n\
+	for (int i = 0; i < num_frag; i++) \n\
+	{ \n\
+		const Pixel pixel = pixels[pixel_list[i]]; \n\
+		const PolyParam pp = tr_poly_params[getPolyNumber(pixel)]; \n\
+#if DEPTH_SORTED != 1 \n\
+		const float frag_depth = pixel.depth; \n\
+		switch (getDepthFunc(pp)) \n\
+		{ \n\
+		case 0:		// Never \n\
+			continue; \n\
+		case 1:		// Greater \n\
+			if (frag_depth <= depth) \n\
+				continue; \n\
+			break; \n\
+		case 2:		// Equal \n\
+			if (frag_depth != depth) \n\
+				continue; \n\
+			break; \n\
+		case 3:		// Greater or equal \n\
+			if (frag_depth < depth) \n\
+				continue; \n\
+			break; \n\
+		case 4:		// Less \n\
+			if (frag_depth >= depth) \n\
+				continue; \n\
+			break; \n\
+		case 5:		// Not equal \n\
+			if (frag_depth == depth) \n\
+				continue; \n\
+			break; \n\
+		case 6:		// Less or equal \n\
+			if (frag_depth > depth) \n\
+				continue; \n\
+			break; \n\
+		case 7:		// Always \n\
+			break; \n\
+		} \n\
+		 \n\
+		if (getDepthMask(pp)) \n\
+			depth = frag_depth; \n\
+#endif \n\
+		bool area1 = false; \n\
+		bool shadowed = false; \n\
+		if (isShadowed(pixel)) \n\
+		{ \n\
+			if (isTwoVolumes(pp)) \n\
+				area1 = true; \n\
+			else \n\
+				shadowed = true; \n\
+		} \n\
+		vec4 srcColor; \n\
+		if (getSrcSelect(pp, area1)) \n\
+			srcColor = secondaryBuffer; \n\
+		else \n\
+		{ \n\
+			srcColor = pixel.color; \n\
+			if (shadowed) \n\
+				srcColor.rgb *= shade_scale_factor; \n\
+		} \n\
+		vec4 dstColor = getDstSelect(pp, area1) ? secondaryBuffer : finalColor; \n\
+		vec4 srcCoef; \n\
+		vec4 dstCoef; \n\
+		 \n\
+		int srcBlend = getSrcBlendFunc(pp, area1); \n\
+		switch (srcBlend) \n\
+		{ \n\
+			case ZERO: \n\
+				srcCoef = vec4(0.0); \n\
+				break; \n\
+			case ONE: \n\
+				srcCoef = vec4(1.0); \n\
+				break; \n\
+			case OTHER_COLOR: \n\
+				srcCoef = finalColor; \n\
+				break; \n\
+			case INVERSE_OTHER_COLOR: \n\
+				srcCoef = vec4(1.0) - dstColor; \n\
+				break; \n\
+			case SRC_ALPHA: \n\
+				srcCoef = vec4(srcColor.a); \n\
+				break; \n\
+			case INVERSE_SRC_ALPHA: \n\
+				srcCoef = vec4(1.0 - srcColor.a); \n\
+				break; \n\
+			case DST_ALPHA: \n\
+				srcCoef = vec4(dstColor.a); \n\
+				break; \n\
+			case INVERSE_DST_ALPHA: \n\
+				srcCoef = vec4(1.0 - dstColor.a); \n\
+				break; \n\
+		} \n\
+		int dstBlend = getDstBlendFunc(pp, area1); \n\
+		switch (dstBlend) \n\
+		{ \n\
+			case ZERO: \n\
+				dstCoef = vec4(0.0); \n\
+				break; \n\
+			case ONE: \n\
+				dstCoef = vec4(1.0); \n\
+				break; \n\
+			case OTHER_COLOR: \n\
+				dstCoef = srcColor; \n\
+				break; \n\
+			case INVERSE_OTHER_COLOR: \n\
+				dstCoef = vec4(1.0) - srcColor; \n\
+				break; \n\
+			case SRC_ALPHA: \n\
+				dstCoef = vec4(srcColor.a); \n\
+				break; \n\
+			case INVERSE_SRC_ALPHA: \n\
+				dstCoef = vec4(1.0 - srcColor.a); \n\
+				break; \n\
+			case DST_ALPHA: \n\
+				dstCoef = vec4(dstColor.a); \n\
+				break; \n\
+			case INVERSE_DST_ALPHA: \n\
+				dstCoef = vec4(1.0 - dstColor.a); \n\
+				break; \n\
+		} \n\
+		const vec4 result = clamp(dstColor * dstCoef + srcColor * srcCoef, 0.0, 1.0); \n\
+		if (getDstSelect(pp, area1)) \n\
+			secondaryBuffer = result; \n\
+		else \n\
+			finalColor = result; \n\
+	} \n\
+	 \n\
+	return finalColor; \n\
+	 \n\
+} \n\
+ \n\
+void main(void) \n\
+{ \n\
+	ivec2 coords = ivec2(gl_FragCoord.xy); \n\
+	// Compute and output final color for the frame buffer \n\
+	// Visualize the number of layers in use \n\
+	//FragColor = vec4(float(fillFragmentArray(coords)) / MAX_PIXELS_PER_FRAGMENT, 0, 0, 1); \n\
+	FragColor = resolveAlphaBlend(coords); \n\
+} \n\
+";
+
+static const char *clear_shader_source = SHADER_HEADER "\
+ \n\
+void main(void) \n\
+{ \n\
+	ivec2 coords = ivec2(gl_FragCoord.xy); \n\
+ \n\
+	// Reset pointers \n\
+	imageStore(abufferPointerImg, coords, uvec4(EOL)); \n\
+ \n\
+	// Discard fragment so nothing is written to the framebuffer \n\
+	discard; \n\
+} \n\
+";
+
+static const char *tr_modvol_shader_source = SHADER_HEADER "\
+#define MV_MODE %d \n\
+#define MAX_PIXELS_PER_FRAGMENT " MAX_PIXELS_PER_FRAGMENT " \n\
+ \n\
+// Must match ModifierVolumeMode enum values \n\
+#define MV_XOR		 0 \n\
+#define MV_OR		 1 \n\
+#define MV_INCLUSION 2 \n\
+#define MV_EXCLUSION 3 \n\
+ \n\
+void main(void) \n\
+{ \n\
+#if MV_MODE == MV_XOR || MV_MODE == MV_OR \n\
+	setFragDepth(); \n\
+#endif \n\
+	ivec2 coords = ivec2(gl_FragCoord.xy); \n\
+	 \n\
+	uint idx = imageLoad(abufferPointerImg, coords).x; \n\
+	int list_len = 0; \n\
+	while (idx != EOL && list_len < MAX_PIXELS_PER_FRAGMENT) \n\
+	{ \n\
+		const Pixel pixel = pixels[idx]; \n\
+		const PolyParam pp = tr_poly_params[getPolyNumber(pixel)]; \n\
+		if (getShadowEnable(pp)) \n\
+		{ \n\
+#if MV_MODE == MV_XOR \n\
+			if (gl_FragDepth <= pixel.depth) \n\
+				atomicXor(pixels[idx].seq_num, SHADOW_STENCIL); \n\
+#elif MV_MODE == MV_OR \n\
+			if (gl_FragDepth <= pixel.depth) \n\
+				atomicOr(pixels[idx].seq_num, SHADOW_STENCIL); \n\
+#elif MV_MODE == MV_INCLUSION \n\
+			uint prev_val = atomicAnd(pixels[idx].seq_num, ~(SHADOW_STENCIL)); \n\
+			if ((prev_val & (SHADOW_STENCIL|SHADOW_ACC)) == SHADOW_STENCIL) \n\
+				pixels[idx].seq_num = bitfieldInsert(pixel.seq_num, 1u, 31, 1); \n\
+#elif MV_MODE == MV_EXCLUSION \n\
+			uint prev_val = atomicAnd(pixels[idx].seq_num, ~(SHADOW_STENCIL|SHADOW_ACC)); \n\
+			if ((prev_val & (SHADOW_STENCIL|SHADOW_ACC)) == SHADOW_ACC) \n\
+				pixels[idx].seq_num = bitfieldInsert(pixel.seq_num, 1u, 31, 1); \n\
+#endif \n\
+		} \n\
+		idx = pixel.next; \n\
+		list_len++; \n\
+	} \n\
+	 \n\
+	discard; \n\
+} \n\
+";
+
+void DrawQuad();
+
+void initABuffer()
+{
+	g_imageWidth = screen_width;
+	g_imageHeight = screen_height;
+
+	if (g_imageWidth > 0 && g_imageHeight > 0)
+	{
+		if (pixels_pointers == 0)
+			pixels_pointers = glcache.GenTexture();
+		glActiveTexture(GL_TEXTURE4);
+		glBindTexture(GL_TEXTURE_2D, pixels_pointers);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+		glTexImage2D(GL_TEXTURE_2D, 0, GL_R32UI, g_imageWidth, g_imageHeight, 0, GL_RED_INTEGER, GL_UNSIGNED_INT, 0);
+		glBindImageTexture(4, pixels_pointers, 0, false, 0,  GL_READ_WRITE, GL_R32UI);
+		glCheck();
+	}
+
+	if (pixels_buffer == 0 )
+	{
+		// Create the buffer
+		glGenBuffers(1, &pixels_buffer);
+		// Bind it
+		glBindBuffer(GL_SHADER_STORAGE_BUFFER, pixels_buffer);
+		// Declare storage
+		glBufferData(GL_SHADER_STORAGE_BUFFER, pixel_buffer_size, NULL, GL_DYNAMIC_COPY);
+		glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, pixels_buffer);
+		glCheck();
+	}
+
+	if (atomic_buffer == 0 )
+	{
+		// Create the buffer
+		glGenBuffers(1, &atomic_buffer);
+		// Bind it
+		glBindBuffer(GL_ATOMIC_COUNTER_BUFFER, atomic_buffer);
+		// Declare storage
+		glBufferData(GL_ATOMIC_COUNTER_BUFFER, 4, NULL, GL_DYNAMIC_COPY);
+		glBindBufferBase(GL_ATOMIC_COUNTER_BUFFER, 0, atomic_buffer);
+		GLint zero = 0;
+		glBufferSubData(GL_SHADER_STORAGE_BUFFER, 0, sizeof(GLint), &zero);
+		glCheck();
+	}
+
+	if (g_abuffer_final_shader.program == 0)
+	{
+		char source[16384];
+		sprintf(source, final_shader_source, 1);
+		CompilePipelineShader(&g_abuffer_final_shader, source);
+	}
+	if (g_abuffer_final_nosort_shader.program == 0)
+	{
+		char source[16384];
+		sprintf(source, final_shader_source, 0);
+		CompilePipelineShader(&g_abuffer_final_nosort_shader, source);
+	}
+	if (g_abuffer_clear_shader.program == 0)
+		CompilePipelineShader(&g_abuffer_clear_shader, clear_shader_source);
+	if (g_abuffer_tr_modvol_shaders[0].program == 0)
+	{
+		char source[16384];
+		for (int mode = 0; mode < ModeCount; mode++)
+		{
+			sprintf(source, tr_modvol_shader_source, mode);
+			CompilePipelineShader(&g_abuffer_tr_modvol_shaders[mode], source);
+		}
+	}
+
+	if (g_quadVertexArray == 0)
+		glGenVertexArrays(1, &g_quadVertexArray);
+	if (g_quadBuffer == 0)
+		glGenBuffers(1, &g_quadBuffer);
+
+	glCheck();
+
+	// Clear A-buffer pointers
+	glcache.UseProgram(g_abuffer_clear_shader.program);
+	ShaderUniforms.Set(&g_abuffer_clear_shader);
+
+	DrawQuad();
+	glMemoryBarrier(GL_SHADER_IMAGE_ACCESS_BARRIER_BIT);
+
+	glCheck();
+}
+
+void reshapeABuffer(int w, int h)
+{
+	if (w != g_imageWidth || h != g_imageHeight) {
+		if (pixels_pointers != 0)
+		{
+			glcache.DeleteTextures(1, &pixels_pointers);
+			pixels_pointers = 0;
+		}
+
+		initABuffer();
+	}
+}
+
+void DrawQuad()
+{
+	glBindVertexArray(g_quadVertexArray);
+
+	float xmin = (ShaderUniforms.scale_coefs[2] - 1) / ShaderUniforms.scale_coefs[0];
+	float xmax = (ShaderUniforms.scale_coefs[2] + 1) / ShaderUniforms.scale_coefs[0];
+	float ymin = (ShaderUniforms.scale_coefs[3] - 1) / ShaderUniforms.scale_coefs[1];
+	float ymax = (ShaderUniforms.scale_coefs[3] + 1) / ShaderUniforms.scale_coefs[1];
+	if (ymin > ymax)
+	{
+		float t = ymin;
+		ymin = ymax;
+		ymax = t;
+	}
+	struct Vertex vertices[] = {
+			{ xmin, ymax, 1, { 255, 255, 255, 255 }, { 0, 0, 0, 0 }, 0, 1 },
+			{ xmin, ymin, 1, { 255, 255, 255, 255 }, { 0, 0, 0, 0 }, 0, 0 },
+			{ xmax, ymax, 1, { 255, 255, 255, 255 }, { 0, 0, 0, 0 }, 1, 1 },
+			{ xmax, ymin, 1, { 255, 255, 255, 255 }, { 0, 0, 0, 0 }, 1, 0 },
+	};
+	GLushort indices[] = { 0, 1, 2, 1, 3 };
+
+	glBindBuffer(GL_ARRAY_BUFFER, g_quadBuffer); glCheck();
+	glBufferData(GL_ARRAY_BUFFER, sizeof(vertices), vertices, GL_STREAM_DRAW); glCheck();
+	glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0); glCheck();
+
+	glEnableVertexAttribArray(VERTEX_POS_ARRAY); glCheck();
+	glVertexAttribPointer(VERTEX_POS_ARRAY, 3, GL_FLOAT, GL_FALSE, sizeof(Vertex), (void*)offsetof(Vertex,x)); glCheck();
+
+	glEnableVertexAttribArray(VERTEX_COL_BASE_ARRAY); glCheck();
+	glVertexAttribPointer(VERTEX_COL_BASE_ARRAY, 4, GL_UNSIGNED_BYTE, GL_TRUE, sizeof(Vertex), (void*)offsetof(Vertex,col)); glCheck();
+
+	glEnableVertexAttribArray(VERTEX_COL_OFFS_ARRAY); glCheck();
+	glVertexAttribPointer(VERTEX_COL_OFFS_ARRAY, 4, GL_UNSIGNED_BYTE, GL_TRUE, sizeof(Vertex), (void*)offsetof(Vertex,spc)); glCheck();
+
+	glEnableVertexAttribArray(VERTEX_UV_ARRAY); glCheck();
+	glVertexAttribPointer(VERTEX_UV_ARRAY, 2, GL_FLOAT, GL_FALSE, sizeof(Vertex), (void*)offsetof(Vertex,u)); glCheck();
+
+	glDisableVertexAttribArray(VERTEX_UV1_ARRAY);
+	glDisableVertexAttribArray(VERTEX_COL_OFFS1_ARRAY);
+	glDisableVertexAttribArray(VERTEX_COL_BASE1_ARRAY);
+
+	glDrawElements(GL_TRIANGLE_STRIP, 5, GL_UNSIGNED_SHORT, indices); glCheck();
+}
+
+void DrawTranslucentModVols(int first, int count)
+{
+	if (count == 0 || pvrrc.modtrig.used() == 0)
+		return;
+	SetupModvolVBO();
+
+	glActiveTexture(GL_TEXTURE2);
+	glBindTexture(GL_TEXTURE_2D, 0);
+	glActiveTexture(GL_TEXTURE3);
+	glBindTexture(GL_TEXTURE_2D, 0);
+	glActiveTexture(GL_TEXTURE1);
+	glBindTexture(GL_TEXTURE_2D, 0);
+	glActiveTexture(GL_TEXTURE0);
+	glBindTexture(GL_TEXTURE_2D, 0);
+
+	glcache.Disable(GL_DEPTH_TEST);
+	glcache.Disable(GL_STENCIL_TEST);
+
+	glCheck();
+
+	ModifierVolumeParam* params = &pvrrc.global_param_mvo_tr.head()[first];
+
+	glMemoryBarrier(GL_SHADER_IMAGE_ACCESS_BARRIER_BIT | GL_BUFFER_UPDATE_BARRIER_BIT);
+
+	int mod_base = -1;
+
+	for (u32 cmv = 0; cmv < count; cmv++)
+	{
+		ModifierVolumeParam& param = params[cmv];
+
+		if (param.count == 0)
+			continue;
+
+		u32 mv_mode = param.isp.DepthMode;
+
+		verify(param.first >= 0 && param.first + param.count <= pvrrc.modtrig.used());
+
+		if (mod_base == -1)
+			mod_base = param.first;
+
+		PipelineShader *shader;
+		if (!param.isp.VolumeLast && mv_mode > 0)
+			shader = &g_abuffer_tr_modvol_shaders[Or];	// OR'ing (open volume or quad)
+		else
+			shader = &g_abuffer_tr_modvol_shaders[Xor];	// XOR'ing (closed volume)
+		glcache.UseProgram(shader->program);
+		ShaderUniforms.Set(shader);
+
+		SetCull(param.isp.CullMode); glCheck();
+
+		glMemoryBarrier(GL_BUFFER_UPDATE_BARRIER_BIT);
+
+		glDrawArrays(GL_TRIANGLES, param.first * 3, param.count * 3); glCheck();
+
+		if (mv_mode == 1 || mv_mode == 2)
+		{
+			//Sum the area
+			shader = &g_abuffer_tr_modvol_shaders[mv_mode == 1 ? Inclusion : Exclusion];
+			glcache.UseProgram(shader->program);
+			ShaderUniforms.Set(shader);
+
+			glMemoryBarrier(GL_BUFFER_UPDATE_BARRIER_BIT);
+			glDrawArrays(GL_TRIANGLES, mod_base * 3, (param.first + param.count - mod_base) * 3); glCheck();
+			mod_base = -1;
+		}
+	}
+}
+
+void checkOverflowAndReset()
+{
+	// Using atomic counter
+	GLuint max_pixel_index = 0;
+//	glGetBufferSubData(GL_ATOMIC_COUNTER_BUFFER, 0, sizeof(GLuint), &max_pixel_index);
+////	printf("ABUFFER %d pixels used\n", max_pixel_index);
+//	if ((max_pixel_index + 1) * 32 - 1 >= pixel_buffer_size)
+//	{
+//		GLint64 size;
+//		glGetInteger64v(GL_MAX_SHADER_STORAGE_BLOCK_SIZE, &size);
+//		if (pixel_buffer_size == size)
+//			printf("A-buffer overflow: %d pixels. Buffer size already maxed out\n", max_pixel_index);
+//		else
+//		{
+//			pixel_buffer_size = (GLuint)min(2 * (GLint64)pixel_buffer_size, size);
+//
+//			printf("A-buffer overflow: %d pixels. Resizing buffer to %d MB\n", max_pixel_index, pixel_buffer_size / 1024 / 1024);
+//
+//			glBindBuffer(GL_SHADER_STORAGE_BUFFER, pixels_buffer);
+//			glBufferData(GL_SHADER_STORAGE_BUFFER, pixel_buffer_size, NULL, GL_DYNAMIC_COPY);
+//			glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, pixels_buffer);
+//			glCheck();
+//		}
+//	}
+	// Reset counter
+	max_pixel_index = 0;
+ 	glBufferSubData(GL_ATOMIC_COUNTER_BUFFER, 0 , sizeof(GLuint), &max_pixel_index);
+}
+
+void renderABuffer(bool sortFragments)
+{
+	// Render to output FBO
+	glcache.UseProgram(sortFragments ? g_abuffer_final_shader.program : g_abuffer_final_nosort_shader.program);
+	ShaderUniforms.Set(&g_abuffer_final_shader);
+
+	glcache.Disable(GL_DEPTH_TEST);
+	glcache.Disable(GL_CULL_FACE);
+	glMemoryBarrier(GL_SHADER_IMAGE_ACCESS_BARRIER_BIT | GL_BUFFER_UPDATE_BARRIER_BIT);
+
+	DrawQuad();
+
+	glCheck();
+
+	// Clear A-buffer pointers
+	glcache.UseProgram(g_abuffer_clear_shader.program);
+	ShaderUniforms.Set(&g_abuffer_clear_shader);
+
+	glMemoryBarrier(GL_SHADER_IMAGE_ACCESS_BARRIER_BIT);
+	DrawQuad();
+
+	glActiveTexture(GL_TEXTURE0);
+
+	glCheck();
+}

core/rend/gles/glcache.h

@@ -9,7 +9,7 @@ public:
 	GLCache() { Reset(); }
 
 	void BindTexture(GLenum target,  GLuint texture) {
-		if (target == GL_TEXTURE_2D && texture != _texture) {
+		if ((target == GL_TEXTURE_2D && texture != _texture && !_disable_cache) || _disable_cache) {
 			glBindTexture(target, texture);
 			_texture = texture;
 		}
@@ -18,7 +18,7 @@ public:
 	}
 
 	void BlendFunc(GLenum sfactor, GLenum dfactor) {
-		if (sfactor != _src_blend_factor || dfactor != _dst_blend_factor) {
+		if (sfactor != _src_blend_factor || dfactor != _dst_blend_factor || _disable_cache) {
 			_src_blend_factor = sfactor;
 			_dst_blend_factor = dfactor;
 			glBlendFunc(sfactor, dfactor);
@@ -26,7 +26,7 @@ public:
 	}
 
 	void ClearColor(GLclampf red, GLclampf green, GLclampf blue, GLclampf alpha) {
-		if (red != _clear_r || green != _clear_g || blue != _clear_b || alpha != _clear_a) {
+		if (red != _clear_r || green != _clear_g || blue != _clear_b || alpha != _clear_a || _disable_cache) {
 			_clear_r = red;
 			_clear_g = green;
 			_clear_b = blue;
@@ -36,7 +36,7 @@ public:
 	}
 
 	void CullFace(GLenum mode) {
-		if (mode != _cull_face) {
+		if (mode != _cull_face || _disable_cache) {
 			_cull_face = mode;
 			glCullFace(mode);
 		}
@@ -52,14 +52,14 @@ public:
 	}
 
 	void DepthFunc(GLenum func) {
-		if (func != _depth_func) {
+		if (func != _depth_func || _disable_cache) {
 			_depth_func = func;
 			glDepthFunc(func);
 		}
 	}
 
 	void DepthMask(GLboolean flag) {
-		if (flag != _depth_mask) {
+		if (flag != _depth_mask || _disable_cache) {
 			_depth_mask = flag;
 			glDepthMask(flag);
 		}
@@ -74,14 +74,14 @@ public:
 	}
 
 	void UseProgram(GLuint program) {
-		if (program != _program) {
+		if (program != _program || _disable_cache) {
 			_program = program;
 			glUseProgram(program);
 		}
 	}
 
 	void StencilFunc(GLenum func, GLint ref, GLuint mask) {
-		if (_stencil_func != func || _stencil_ref != ref || _stencil_fmask != mask) {
+		if (_stencil_func != func || _stencil_ref != ref || _stencil_fmask != mask || _disable_cache) {
 			_stencil_func = func;
 			_stencil_ref = ref;
 			_stencil_fmask = mask;
@@ -90,7 +90,7 @@ public:
 	}
 
 	void StencilOp(GLenum sfail, GLenum dpfail, GLenum dppass) {
-		if (_stencil_sfail != sfail ||_stencil_dpfail != dpfail || _stencil_dppass != dppass) {
+		if (_stencil_sfail != sfail ||_stencil_dpfail != dpfail || _stencil_dppass != dppass || _disable_cache) {
 			_stencil_sfail = sfail;
 			_stencil_dpfail = dpfail;
 			_stencil_dppass = dppass;
@@ -99,14 +99,14 @@ public:
 	}
 
 	void StencilMask(GLuint mask) {
-		if (_stencil_mask != mask) {
+		if (_stencil_mask != mask || _disable_cache) {
 			_stencil_mask = mask;
 			glStencilMask(mask);
 		}
 	}
 
 	void TexParameteri(GLenum target,  GLenum pname,  GLint param) {
-		if (target == GL_TEXTURE_2D)
+		if (target == GL_TEXTURE_2D && !_disable_cache)
 		{
 			TextureParameters &cur_params = _texture_params[_texture];
 			switch (pname) {
@@ -201,7 +201,7 @@ private:
 			break;
 		}
 		if (pCap != NULL) {
-			if (*pCap == value)
+			if (*pCap == value && !_disable_cache)
 				return;
 			*pCap = value;
 		}
@@ -237,6 +237,7 @@ private:
 	GLuint _texture_ids[TEXTURE_ID_CACHE_SIZE];
 	GLuint _texture_cache_size;
 	std::map<GLuint, TextureParameters> _texture_params;
+	bool _disable_cache = true;
 };
 
 extern GLCache glcache;

core/rend/gles/gles.h

@@ -1,39 +1,12 @@
 #pragma once
 #include "rend/rend.h"
+#include <map>
 
-
-#ifdef GLES
-#if defined(TARGET_IPHONE) //apple-specific ogles2 headers
-//#include <APPLE/egl.h>
-#include <OpenGLES/ES2/gl.h>
-#include <OpenGLES/ES2/glext.h>
-#else
-#if !defined(TARGET_NACL32)
-#include <EGL/egl.h>
-#endif
-#include <GLES2/gl2.h>
-#include <GLES2/gl2ext.h>
-#endif
-
-#ifndef GL_NV_draw_path
-//IMGTEC GLES emulation
-#pragma comment(lib,"libEGL.lib")
-#pragma comment(lib,"libGLESv2.lib")
-#else /* NV gles emulation*/
-#pragma comment(lib,"libGLES20.lib")
-#endif
-
-#else
 #if HOST_OS == OS_DARWIN
     #include <OpenGL/gl3.h>
 #else
 	#include <GL3/gl3w.h>
 #endif
-#endif
-
-#ifndef GL_UNSIGNED_INT_8_8_8_8
-#define GL_UNSIGNED_INT_8_8_8_8 0x8035
-#endif
 
 #define glCheck() do { if (unlikely(settings.validate.OpenGlChecks)) { verify(glGetError()==GL_NO_ERROR); } } while(0)
 #define eglCheck() false
@@ -42,6 +15,9 @@
 #define VERTEX_COL_BASE_ARRAY 1
 #define VERTEX_COL_OFFS_ARRAY 2
 #define VERTEX_UV_ARRAY 3
+#define VERTEX_COL_BASE1_ARRAY 4
+#define VERTEX_COL_OFFS1_ARRAY 5
+#define VERTEX_UV1_ARRAY 6
 
 #ifndef GL_UNSIGNED_INT_8_8_8_8
 #define GL_UNSIGNED_INT_8_8_8_8 0x8035
@@ -52,71 +28,71 @@ extern u32 gcflip;
 extern float scale_x, scale_y;
 
 
-void DrawStrips();
+void DrawStrips(GLuint output_fbo);
 
 struct PipelineShader
 {
 	GLuint program;
 
-	GLuint scale,depth_scale;
+	GLuint scale;
 	GLuint extra_depth_scale;
 	GLuint pp_ClipTest,cp_AlphaTestValue;
 	GLuint sp_FOG_COL_RAM,sp_FOG_COL_VERT,sp_FOG_DENSITY;
+	GLuint shade_scale_factor;
+	GLuint pp_Number;
+	GLuint blend_mode;
+	GLuint use_alpha;
+	GLuint ignore_tex_alpha;
+	GLuint shading_instr;
+	GLuint fog_control;
 	GLuint trilinear_alpha;
 	GLuint fog_clamp_min, fog_clamp_max;
 
 	//
 	u32 cp_AlphaTest; s32 pp_ClipTestMode;
 	u32 pp_Texture, pp_UseAlpha, pp_IgnoreTexA, pp_ShadInstr, pp_Offset, pp_FogCtrl;
-	bool pp_Gouraud, pp_BumpMap;
+	u32 pp_DepthFunc;
+	int pass;
+	bool pp_TwoVolumes;
+	bool pp_Gouraud;
+	bool pp_BumpMap;
 	bool fog_clamping;
 };
 
 
 struct gl_ctx
 {
-#if defined(GLES) && HOST_OS != OS_DARWIN && !defined(TARGET_NACL32)
-	struct
-	{
-		EGLNativeWindowType native_wind;
-		EGLNativeDisplayType native_disp;
-		EGLDisplay display;
-		EGLSurface surface;
-		EGLContext context;
-	} setup;
-#endif
-
 	struct
 	{
 		GLuint program;
 
-		GLuint scale,depth_scale;
+		GLuint scale;
 		GLuint extra_depth_scale;
-		GLuint sp_ShaderColor;
-
 	} modvol_shader;
 
-	PipelineShader pogram_table[12288];
+	std::map<int, PipelineShader *> shaders;
 	struct
 	{
-		GLuint program,scale,depth_scale;
+		GLuint program,scale;
 		GLuint extra_depth_scale;
 	} OSD_SHADER;
 
 	struct
 	{
 		GLuint geometry,modvols,idxs,idxs2;
-#ifndef GLES
 		GLuint vao;
-#endif
+		GLuint tr_poly_params;
 	} vbo;
 
-	const char *gl_version;
-	const char *glsl_version_header;
-	int gl_major;
-	bool is_gles;
-	GLuint fog_image_format;
-	//GLuint matrix;
+	PipelineShader *getShader(int programId) {
+		PipelineShader *shader = shaders[programId];
+		if (shader == NULL) {
+			shader = new PipelineShader();
+			shaders[programId] = shader;
+			shader->program = -1;
+		}
+		return shader;
+	}
 };
 
 extern gl_ctx gl;
@@ -135,34 +111,197 @@ void CollectCleanup();
 void DoCleanup();
 void SortPParams(int first, int count);
 
-void BindRTT(u32 addy, u32 fbw, u32 fbh, u32 channels, u32 fmt);
+extern int screen_width;
+extern int screen_height;
+
+GLuint BindRTT(u32 addy, u32 fbw, u32 fbh, u32 channels, u32 fmt);
 void ReadRTTBuffer();
 void RenderFramebuffer();
 void DrawFramebuffer(float w, float h);
 
 int GetProgramID(u32 cp_AlphaTest, u32 pp_ClipTestMode,
 							u32 pp_Texture, u32 pp_UseAlpha, u32 pp_IgnoreTexA, u32 pp_ShadInstr, u32 pp_Offset,
-							u32 pp_FogCtrl, bool pp_Gouraud, bool pp_BumpMap, bool fog_clamping);
+							u32 pp_FogCtrl, bool two_volumes, u32 pp_DepthFunc, bool pp_Gouraud, bool pp_BumpMap, bool fog_clamping, int pass);
+void SetCull(u32 CulliMode);
 
-bool CompilePipelineShader(PipelineShader* s);
+extern const char *PixelPipelineShader;
+bool CompilePipelineShader(PipelineShader* s, const char *source = PixelPipelineShader);
 #define TEXTURE_LOAD_ERROR 0
 GLuint loadPNG(const string& subpath, int &width, int &height);
 
+extern GLuint stencilTexId;
+extern GLuint depthTexId;
+extern GLuint opaqueTexId;
+extern GLuint depthSaveTexId;
+
+#define SHADER_HEADER "#version 430 \n\
+\n\
+layout(r32ui, binding = 4) uniform coherent restrict uimage2D abufferPointerImg; \n\
+struct Pixel { \n\
+	highp vec4 color; \n\
+	highp float depth; \n\
+	uint seq_num; \n\
+	uint next; \n\
+}; \n\
+#define EOL 0xFFFFFFFFu \n\
+layout (binding = 0, std430) coherent restrict buffer PixelBuffer { \n\
+	Pixel pixels[]; \n\
+}; \n\
+layout(binding = 0, offset = 0) uniform atomic_uint buffer_index; \n\
+\n\
+#define ZERO				0 \n\
+#define ONE					1 \n\
+#define OTHER_COLOR			2 \n\
+#define INVERSE_OTHER_COLOR	3 \n\
+#define SRC_ALPHA			4 \n\
+#define INVERSE_SRC_ALPHA	5 \n\
+#define DST_ALPHA			6 \n\
+#define INVERSE_DST_ALPHA	7 \n\
+ \n\
+uint getNextPixelIndex() \n\
+{ \n\
+	uint index = atomicCounterIncrement(buffer_index); \n\
+	if (index >= pixels.length()) \n\
+		// Buffer overflow \n\
+		discard; \n\
+	 \n\
+	return index; \n\
+} \n\
+\n\
+void setFragDepth(void) \n\
+{ \n\
+	highp float w = 100000.0 * gl_FragCoord.w; \n\
+	gl_FragDepth = 1.0 - log2(1.0 + w) / 34.0; \n\
+} \n\
+struct PolyParam { \n\
+	int first; \n\
+	int count; \n\
+	int texid; \n\
+	int tsp; \n\
+	int tcw; \n\
+	int pcw; \n\
+	int isp; \n\
+	float zvZ; \n\
+	int tileclip; \n\
+	int tsp1; \n\
+	int tcw1; \n\
+	int texid1; \n\
+}; \n\
+layout (binding = 1, std430) readonly buffer TrPolyParamBuffer { \n\
+	PolyParam tr_poly_params[]; \n\
+}; \n\
+ \n\
+#define GET_TSP_FOR_AREA int tsp; if (area1) tsp = pp.tsp1; else tsp = pp.tsp; \n\
+ \n\
+int getSrcBlendFunc(const PolyParam pp, bool area1) \n\
+{ \n\
+	GET_TSP_FOR_AREA \n\
+	return (tsp >> 29) & 7; \n\
+} \n\
+\n\
+int getDstBlendFunc(const PolyParam pp, bool area1) \n\
+{ \n\
+	GET_TSP_FOR_AREA \n\
+	return (tsp >> 26) & 7; \n\
+} \n\
+\n\
+bool getSrcSelect(const PolyParam pp, bool area1) \n\
+{ \n\
+	GET_TSP_FOR_AREA \n\
+	return ((tsp >> 25) & 1) != 0; \n\
+} \n\
+\n\
+bool getDstSelect(const PolyParam pp, bool area1) \n\
+{ \n\
+	GET_TSP_FOR_AREA \n\
+	return ((tsp >> 24) & 1) != 0; \n\
+} \n\
+\n\
+int getFogControl(const PolyParam pp, bool area1) \n\
+{ \n\
+	GET_TSP_FOR_AREA \n\
+	return (tsp >> 22) & 3; \n\
+} \n\
+\n\
+bool getUseAlpha(const PolyParam pp, bool area1) \n\
+{ \n\
+	GET_TSP_FOR_AREA \n\
+	return ((tsp >> 20) & 1) != 0; \n\
+} \n\
+\n\
+bool getIgnoreTexAlpha(const PolyParam pp, bool area1) \n\
+{ \n\
+	GET_TSP_FOR_AREA \n\
+	return ((tsp >> 19) & 1) != 0; \n\
+} \n\
+\n\
+int getShadingInstruction(const PolyParam pp, bool area1) \n\
+{ \n\
+	GET_TSP_FOR_AREA \n\
+	return (tsp >> 6) & 3; \n\
+} \n\
+\n\
+int getDepthFunc(const PolyParam pp) \n\
+{ \n\
+	return (pp.isp >> 29) & 7; \n\
+} \n\
+\n\
+bool getDepthMask(const PolyParam pp) \n\
+{ \n\
+	return ((pp.isp >> 26) & 1) != 1; \n\
+} \n\
+\n\
+bool getShadowEnable(const PolyParam pp) \n\
+{ \n\
+	return ((pp.pcw >> 7) & 1) != 0; \n\
+} \n\
+\n\
+uint getPolyNumber(const Pixel pixel) \n\
+{ \n\
+	return pixel.seq_num & 0x3FFFFFFFu; \n\
+} \n\
+\n\
+#define SHADOW_STENCIL 0x40000000u \n\
+#define SHADOW_ACC	   0x80000000u \n\
+\n\
+bool isShadowed(const Pixel pixel) \n\
+{ \n\
+	return (pixel.seq_num & SHADOW_ACC) == SHADOW_ACC; \n\
+} \n\
+\n\
+bool isTwoVolumes(const PolyParam pp) \n\
+{ \n\
+	return pp.tsp1 != -1 || pp.tcw1 != -1; \n\
+} \n\
+ \n\
+"
+
+void SetupModvolVBO();
 enum ModifierVolumeMode { Xor, Or, Inclusion, Exclusion, ModeCount };
 
 extern struct ShaderUniforms_t
 {
 	float PT_ALPHA;
 	float scale_coefs[4];
-	float depth_coefs[4];
 	float extra_depth_scale;
 	float fog_den_float;
 	float ps_FOG_COL_RAM[3];
 	float ps_FOG_COL_VERT[3];
+	int poly_number;
 	float trilinear_alpha;
+	TSP tsp0;
+	TSP tsp1;
+	TCW tcw0;
+	TCW tcw1;
 	float fog_clamp_min[4];
 	float fog_clamp_max[4];
 
+	void setUniformArray(GLuint location, int v0, int v1)
+	{
+		int array[] = { v0, v1 };
+		glUniform1iv(location, 2, array);
+	}
+
 	void Set(PipelineShader* s)
 	{
 		if (s->cp_AlphaTestValue!=-1)
@@ -171,9 +310,6 @@ extern struct ShaderUniforms_t
 		if (s->scale!=-1)
 			glUniform4fv( s->scale, 1, scale_coefs);
 
-		if (s->depth_scale!=-1)
-			glUniform4fv( s->depth_scale, 1, depth_coefs);
-
 		if (s->extra_depth_scale != -1)
 			glUniform1f(s->extra_depth_scale, extra_depth_scale);
 
@@ -186,6 +322,29 @@ extern struct ShaderUniforms_t
 		if (s->sp_FOG_COL_VERT!=-1)
 			glUniform3fv( s->sp_FOG_COL_VERT, 1, ps_FOG_COL_VERT);
 
+		if (s->shade_scale_factor != -1)
+			glUniform1f(s->shade_scale_factor, FPU_SHAD_SCALE.scale_factor / 256.f);
+
+		if (s->blend_mode != -1) {
+			u32 blend_mode[] = { tsp0.SrcInstr, tsp0.DstInstr, tsp1.SrcInstr, tsp1.DstInstr };
+			glUniform2iv(s->blend_mode, 2, (GLint *)blend_mode);
+		}
+
+		if (s->use_alpha != -1)
+			setUniformArray(s->use_alpha, tsp0.UseAlpha, tsp1.UseAlpha);
+
+		if (s->ignore_tex_alpha != -1)
+			setUniformArray(s->ignore_tex_alpha, tsp0.IgnoreTexA, tsp1.IgnoreTexA);
+
+		if (s->shading_instr != -1)
+			setUniformArray(s->shading_instr, tsp0.ShadInstr, tsp1.ShadInstr);
+
+		if (s->fog_control != -1)
+			setUniformArray(s->fog_control, tsp0.FogCtrl, tsp1.FogCtrl);
+
+		if (s->pp_Number != -1)
+			glUniform1i(s->pp_Number, poly_number);
+
 		if (s->trilinear_alpha != -1)
 			glUniform1f(s->trilinear_alpha, trilinear_alpha);
 		

core/rend/gles/gltex.cpp

@@ -206,7 +206,7 @@ static void dumpTexture(int texID, int w, int h, GLuint textype, void *temp_tex_
 	fclose(fp);
 
 	for (int y = 0; y < h; y++)
-	free(rows[y]);
+		free(rows[y]);
 	free(rows);
 }
 
@@ -576,29 +576,26 @@ TextureCacheData *getTextureCacheData(TSP tsp, TCW tcw);
 struct FBT
 {
 	u32 TexAddr;
-	GLuint depthb,stencilb;
 	GLuint tex;
 	GLuint fbo;
 };
 
 FBT fb_rtt;
 
-void BindRTT(u32 addy, u32 fbw, u32 fbh, u32 channels, u32 fmt)
+GLuint BindRTT(u32 addy, u32 fbw, u32 fbh, u32 channels, u32 fmt)
 {
 	FBT& rv=fb_rtt;
 
 	if (rv.fbo) glDeleteFramebuffers(1,&rv.fbo);
 	if (rv.tex) glcache.DeleteTextures(1,&rv.tex);
-	if (rv.depthb) glDeleteRenderbuffers(1,&rv.depthb);
-	if (rv.stencilb) glDeleteRenderbuffers(1,&rv.stencilb);
 
 	rv.TexAddr=addy>>3;
 
-	// Find the largest square power of two texture that fits into the viewport
-	int fbh2 = 2;
+	// Find the smallest power of two texture that fits the viewport
+	int fbh2 = 8;
 	while (fbh2 < fbh)
 		fbh2 *= 2;
-	int fbw2 = 2;
+	int fbw2 = 8;
 	while (fbw2 < fbw)
 		fbw2 *= 2;
 
@@ -612,26 +609,6 @@ void BindRTT(u32 addy, u32 fbw, u32 fbh, u32 channels, u32 fmt)
 	// Get the currently bound frame buffer object. On most platforms this just gives 0.
 	//glGetIntegerv(GL_FRAMEBUFFER_BINDING, &m_i32OriginalFbo);
 
-	// Generate and bind a render buffer which will become a depth buffer shared between our two FBOs
-	glGenRenderbuffers(1, &rv.depthb);
-	glBindRenderbuffer(GL_RENDERBUFFER, rv.depthb);
-
-	/*
-		Currently it is unknown to GL that we want our new render buffer to be a depth buffer.
-		glRenderbufferStorage will fix this and in this case will allocate a depth buffer
-		m_i32TexSize by m_i32TexSize.
-	*/
-
-#ifdef GLES
-	glRenderbufferStorage(GL_RENDERBUFFER, GL_DEPTH_COMPONENT24_OES, fbw2, fbh2);
-#else
-	glRenderbufferStorage(GL_RENDERBUFFER, GL_DEPTH_COMPONENT24, fbw2, fbh2);
-#endif
-
-	glGenRenderbuffers(1, &rv.stencilb);
-	glBindRenderbuffer(GL_RENDERBUFFER, rv.stencilb);
-	glRenderbufferStorage(GL_RENDERBUFFER, GL_STENCIL_INDEX8, fbw2, fbh2);
-
 	// Create a texture for rendering to
 	rv.tex = glcache.GenTexture();
 	glcache.BindTexture(GL_TEXTURE_2D, rv.tex);
@@ -645,15 +622,14 @@ void BindRTT(u32 addy, u32 fbw, u32 fbh, u32 channels, u32 fmt)
 	// Attach the texture to the FBO
 	glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, rv.tex, 0);
 
-	// Attach the depth buffer we created earlier to our FBO.
-	glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_RENDERBUFFER, rv.depthb);
-
 	// Check that our FBO creation was successful
 	GLuint uStatus = glCheckFramebufferStatus(GL_FRAMEBUFFER);
 
 	verify(uStatus == GL_FRAMEBUFFER_COMPLETE);
 
 	glViewport(0, 0, fbw, fbh);		// TODO CLIP_X/Y min?
+
+	return rv.fbo;
 }
 
 void ReadRTTBuffer() {
@@ -798,10 +774,11 @@ void ReadRTTBuffer() {
     }
     fb_rtt.tex = 0;
 
-	if (fb_rtt.fbo) { glDeleteFramebuffers(1,&fb_rtt.fbo); fb_rtt.fbo = 0; }
-	if (fb_rtt.depthb) { glDeleteRenderbuffers(1,&fb_rtt.depthb); fb_rtt.depthb = 0; }
-	if (fb_rtt.stencilb) { glDeleteRenderbuffers(1,&fb_rtt.stencilb); fb_rtt.stencilb = 0; }
-
+	if (fb_rtt.fbo)
+	{
+		glDeleteFramebuffers(1,&fb_rtt.fbo);
+		fb_rtt.fbo = 0;
+	}
 }
 
 static int TexCacheLookups;

core/rend/soft/softrend.cpp

@@ -20,6 +20,7 @@
 #include <smmintrin.h>
 
 #include <cmath>
+#include <algorithm>
 
 #include "rend/gles/gles.h"
 
@@ -53,6 +54,54 @@ union m128i {
 	uint32_t m128i_u32[4];
 };
 
+
+bool operator<(const PolyParam &left, const PolyParam &right)
+{
+/* put any condition you want to sort on here */
+	return left.zvZ<right.zvZ;
+	//return left.zMin<right.zMax;
+}
+
+//Sort based on min-z of each strip
+void SortPParams(int first, int count)
+{
+	if (pvrrc.verts.used() == 0 || count <= 1)
+		return;
+
+	Vertex* vtx_base=pvrrc.verts.head();
+	u16* idx_base=pvrrc.idx.head();
+
+	PolyParam* pp = &pvrrc.global_param_tr.head()[first];
+	PolyParam* pp_end = pp + count;
+
+	while(pp!=pp_end)
+	{
+		if (pp->count<2)
+		{
+			pp->zvZ=0;
+		}
+		else
+		{
+			u16* idx=idx_base+pp->first;
+
+			Vertex* vtx=vtx_base+idx[0];
+			Vertex* vtx_end=vtx_base + idx[pp->count-1]+1;
+
+			u32 zv=0xFFFFFFFF;
+			while(vtx!=vtx_end)
+			{
+				zv=min(zv,(u32&)vtx->z);
+				vtx++;
+			}
+
+			pp->zvZ=(f32&)zv;
+		}
+		pp++;
+	}
+
+	std::stable_sort(pvrrc.global_param_tr.head() + first, pvrrc.global_param_tr.head() + first + count);
+}
+
 static __m128 _mm_load_scaled_float(float v, float s)
 {
 	return _mm_setr_ps(v, v + s, v + s + s, v + s + s + s);

shell/reicast.vcxproj

@@ -1,4 +1,4 @@
-<?xml version="1.0" encoding="utf-8"?>
+<?xml version="1.0" encoding="utf-8"?>
 <Project DefaultTargets="Build" ToolsVersion="14.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
   <ItemGroup Label="ProjectConfigurations">
     <ProjectConfiguration Include="Fast|Win32">
@@ -191,6 +191,7 @@
     <ClCompile Include="..\core\reios\reios.cpp" />
     <ClCompile Include="..\core\reios\reios_elf.cpp" />
     <ClCompile Include="..\core\rend\d3d11\d3d11.cpp" />
+    <ClCompile Include="..\core\rend\gles\abuffer.cpp" />
     <ClCompile Include="..\core\rend\gles\gldraw.cpp" />
     <ClCompile Include="..\core\rend\gles\gles.cpp" />
     <ClCompile Include="..\core\rend\gles\gltex.cpp" />