Merge branch 'new_blit'

build/windows/example/example.vcxproj

@@ -22,6 +22,12 @@
     <ClCompile Include="..\..\..\examples\sdl-renderer.c" />
   </ItemGroup>
   <ItemGroup>
+    <ProjectReference Include="..\..\..\..\3rd-party\SDL2-2.0.5\VisualC\SDLmain\SDLmain.vcxproj">
+      <Project>{da956fd3-e142-46f2-9dd5-c78bebb56b7a}</Project>
+    </ProjectReference>
+    <ProjectReference Include="..\..\..\..\3rd-party\SDL2-2.0.5\VisualC\SDL\SDL.vcxproj">
+      <Project>{81ce8daf-ebb2-4761-8e45-b71abcca8c68}</Project>
+    </ProjectReference>
     <ProjectReference Include="..\tic80\tic80.vcxproj">
       <Project>{c4d8bc10-ebf6-42bb-9b5d-6712fb428a50}</Project>
     </ProjectReference>
@@ -100,7 +106,7 @@
     <Link>
       <SubSystem>Windows</SubSystem>
       <GenerateDebugInformation>true</GenerateDebugInformation>
-      <AdditionalDependencies>SDL2main.lib;SDL2.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalDependencies>;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
       <AdditionalLibraryDirectories>..\..\..\lib\windows</AdditionalLibraryDirectories>
     </Link>
   </ItemDefinitionGroup>
@@ -115,7 +121,7 @@
     <Link>
       <SubSystem>Windows</SubSystem>
       <GenerateDebugInformation>true</GenerateDebugInformation>
-      <AdditionalDependencies>SDL2main.lib;SDL2.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalDependencies>;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
       <AdditionalLibraryDirectories>..\..\..\lib\windows</AdditionalLibraryDirectories>
     </Link>
   </ItemDefinitionGroup>
@@ -134,7 +140,7 @@
       <EnableCOMDATFolding>true</EnableCOMDATFolding>
       <OptimizeReferences>true</OptimizeReferences>
       <GenerateDebugInformation>true</GenerateDebugInformation>
-      <AdditionalDependencies>SDL2main.lib;SDL2.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalDependencies>;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
       <AdditionalLibraryDirectories>..\..\..\lib\windows</AdditionalLibraryDirectories>
     </Link>
   </ItemDefinitionGroup>
@@ -153,7 +159,7 @@
       <EnableCOMDATFolding>true</EnableCOMDATFolding>
       <OptimizeReferences>true</OptimizeReferences>
       <GenerateDebugInformation>true</GenerateDebugInformation>
-      <AdditionalDependencies>SDL2main.lib;SDL2.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalDependencies>;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
       <AdditionalLibraryDirectories>..\..\..\lib\windows</AdditionalLibraryDirectories>
     </Link>
   </ItemDefinitionGroup>

examples/sdl-renderer.c

@@ -56,9 +56,9 @@ int main(int argc, char **argv)
 			SDL_Init(SDL_INIT_VIDEO | SDL_INIT_AUDIO);
 
 			{
-				SDL_Window* window = SDL_CreateWindow("TIC-80 SDL demo", SDL_WINDOWPOS_CENTERED, SDL_WINDOWPOS_CENTERED, TIC80_WIDTH, TIC80_HEIGHT, SDL_WINDOW_SHOWN | SDL_WINDOW_RESIZABLE);
+				SDL_Window* window = SDL_CreateWindow("TIC-80 SDL demo", SDL_WINDOWPOS_CENTERED, SDL_WINDOWPOS_CENTERED, TIC80_FULLWIDTH, TIC80_FULLHEIGHT, SDL_WINDOW_SHOWN | SDL_WINDOW_RESIZABLE);
 				SDL_Renderer* renderer = SDL_CreateRenderer(window, -1, SDL_RENDERER_ACCELERATED | SDL_RENDERER_PRESENTVSYNC);
-				SDL_Texture* texture = SDL_CreateTexture(renderer, SDL_PIXELFORMAT_ARGB8888, SDL_TEXTUREACCESS_STREAMING, TIC80_WIDTH, TIC80_HEIGHT);
+				SDL_Texture* texture = SDL_CreateTexture(renderer, SDL_PIXELFORMAT_ARGB8888, SDL_TEXTUREACCESS_STREAMING, TIC80_FULLWIDTH, TIC80_FULLHEIGHT);
 				
 				SDL_AudioDeviceID audioDevice = 0;
 				SDL_AudioSpec audioSpec;

include/tic80/tic80.h

@@ -31,6 +31,8 @@ extern "C" {
 
 #define TIC80_WIDTH 240
 #define TIC80_HEIGHT 136
+#define TIC80_FULLWIDTH 256
+#define TIC80_FULLHEIGHT (TIC80_FULLWIDTH*9/16)
 
 typedef struct 
 {
@@ -47,15 +49,8 @@ typedef struct
 		s32 count;
 	} sound;
 
-	u32 screen[TIC80_WIDTH * TIC80_HEIGHT];
+	u32 screen[TIC80_FULLWIDTH * TIC80_FULLHEIGHT];
-	u32 border[TIC80_HEIGHT];
 	
-	struct
-	{
-		s8 x;
-		s8 y;
-		s8 rows[TIC80_HEIGHT];
-	} offset;
 } tic80;
 
 typedef union

src/console.c

@@ -1782,7 +1782,7 @@ static void onConsoleRamCommand(Console* console, const char* param)
 		ADDR_RECORD(console->tic->ram.vram.screen, 					"SCREEN"),
 		ADDR_RECORD(console->tic->ram.vram.palette, 				"PALETTE"),
 		ADDR_RECORD(console->tic->ram.vram.mapping, 				"PALETTE MAP"),
-		ADDR_RECORD(console->tic->ram.vram.vars.border, 			"BORDER COLOR"),
+		ADDR_RECORD(console->tic->ram.vram.vars.colors, 			"BORDER/BG COLOR"),
 		ADDR_RECORD(console->tic->ram.vram.vars.offset, 			"SCREEN OFFSET"),
 		ADDR_RECORD(console->tic->ram.vram.vars.mask, 				"GAMEPAD MASK"),
 		ADDR_RECORD(console->tic->ram.vram.input.gamepad, 			"GAMEPAD"),

src/studio.c

@@ -47,14 +47,15 @@
 #include "ext/md5.h"
 
 #define STUDIO_UI_SCALE 3
-#define STUDIO_UI_BORDER 16
+#define TEXTURE_SIZE (TIC80_FULLWIDTH)
 
 #define MAX_CONTROLLERS 4
 #define STUDIO_PIXEL_FORMAT SDL_PIXELFORMAT_ARGB8888
 
-#define MAX_OFFSET 128
+#define FRAME_SIZE (TIC80_FULLWIDTH * TIC80_FULLHEIGHT * sizeof(u32))
-#define FULL_WIDTH (TIC80_WIDTH + MAX_OFFSET*2)
+
-#define FRAME_SIZE (TIC80_WIDTH * TIC80_HEIGHT * sizeof(u32))
+#define OFFSET_LEFT ((TIC80_FULLWIDTH-TIC80_WIDTH)/2)
+#define OFFSET_TOP ((TIC80_FULLHEIGHT-TIC80_HEIGHT)/2)
 
 typedef struct
 {
@@ -80,8 +81,8 @@ static struct
 
 	SDL_Window* window;
 	SDL_Renderer* renderer;
+	SDL_Renderer* softwareRenderer;
 	SDL_Texture* texture;
-	SDL_Texture* borderTexture;
 
 	SDL_AudioSpec audioSpec;
 	SDL_AudioDeviceID audioDevice;
@@ -195,8 +196,8 @@ static struct
 
 	.window = NULL,
 	.renderer = NULL,
+	.softwareRenderer = NULL,
 	.texture = NULL,
-	.borderTexture = NULL,
 	.audioDevice = 0,
 
 	.joysticks = {NULL, NULL, NULL, NULL},
@@ -962,7 +963,9 @@ static void calcTextureRect(SDL_Rect* rect)
 
 		rect->x = (rect->w - discreteWidth) / 2;
 
-		rect->y = rect->w > rect->h ? (rect->h - discreteHeight) / 2 : 0;
+		rect->y = rect->w > rect->h 
+			? (rect->h - discreteHeight) / 2 
+			: OFFSET_LEFT*discreteWidth/TIC80_WIDTH;
 
 		rect->w = discreteWidth;
 		rect->h = discreteHeight;
@@ -1406,78 +1409,38 @@ static u32* paletteBlit()
 	return srcPaletteBlit(studio.tic->ram.vram.palette.data);
 }
 
-static void blit(u32* out, u32* bgOut, s32 pitch, s32 bgPitch)
+static void screen2buffer(u32* buffer, const u32* pixels, SDL_Rect rect)
 {
-	const s32 pitchWidth = pitch/sizeof *out;
+	pixels += rect.y * TIC80_FULLWIDTH;
-	const s32 bgPitchWidth = bgPitch/sizeof *bgOut;
-	u32* row = out;
-	const u32* pal = paletteBlit();
 
-	void(*scanline)(tic_mem* memory, s32 row) = NULL;
+	for(s32 i = 0; i < rect.h; i++)
-
-	switch(studio.mode)
 	{
-	case TIC_RUN_MODE:
+		SDL_memcpy(buffer, pixels + rect.x, rect.w * sizeof(pixels[0]));
-		scanline = studio.tic->api.scanline;
+		pixels += TIC80_FULLWIDTH;
-		break;
+		buffer += rect.w;
-	case TIC_SPRITE_MODE:
-		scanline = studio.sprite.scanline;
-		break;
-	case TIC_MAP_MODE:
-		scanline = studio.map.scanline;
-		break;
-	default:
-		break;
-	}
-
-	for(s32 r = 0, pos = 0; r < TIC80_HEIGHT; r++, row += pitchWidth)
-	{
-		if(scanline)
-		{
-			scanline(studio.tic, r);
-			pal = paletteBlit();
-		}
-
-		if(bgOut)
-		{
-			u8 border = tic_tool_peek4(studio.tic->ram.vram.mapping, studio.tic->ram.vram.vars.border & 0xf);
-			SDL_memset4(bgOut, pal[border], TIC80_WIDTH);
-			bgOut += bgPitchWidth;
-		}
-
-		SDL_memset4(row, 0, pitchWidth);
-
-		for(u32* ptr = row + MAX_OFFSET + studio.tic->ram.vram.vars.offset.x, c = 0; c < TIC80_WIDTH; c++, ptr++)
-			*ptr = pal[tic_tool_peek4(studio.tic->ram.vram.screen.data, pos++)];
-	}
-}
-
-static void screen2buffer(u32* buffer, const u8* pixels, s32 pitch)
-{
-	for(s32 i = 0; i < TIC80_HEIGHT; i++)
-	{
-		SDL_memcpy(buffer, pixels+MAX_OFFSET * sizeof(u32), TIC80_WIDTH * sizeof(u32));
-		pixels += pitch;
-		buffer += TIC80_WIDTH;
 	}
 }
 
 static void setCoverImage()
 {
+	tic_mem* tic = studio.tic;
+
 	if(studio.mode == TIC_RUN_MODE)
 	{
-		enum {Pitch = FULL_WIDTH*sizeof(u32)};
+		enum {Pitch = TIC80_FULLWIDTH*sizeof(u32)};
-		u32* pixels = SDL_malloc(Pitch * TIC80_HEIGHT);
+		u32* pixels = SDL_malloc(Pitch * TIC80_FULLHEIGHT);
 
 		if(pixels)
 		{
-			blit(pixels, NULL, Pitch, 0);
+			tic->api.blit(tic, pixels, tic->api.scanline);
 
 			u32* buffer = SDL_malloc(TIC80_WIDTH * TIC80_HEIGHT * sizeof(u32));
 
 			if(buffer)
 			{
-				screen2buffer(buffer, (const u8*)pixels, Pitch);
+				SDL_Rect rect = {OFFSET_LEFT, OFFSET_TOP, TIC80_WIDTH, TIC80_HEIGHT};
+
+				screen2buffer(buffer, pixels, rect);
 
 				gif_write_animation(studio.tic->cart.cover.data, &studio.tic->cart.cover.size,
 					TIC80_WIDTH, TIC80_HEIGHT, (const u8*)buffer, 1, TIC_FRAMERATE, 1);
@@ -1508,7 +1471,7 @@ static void stopVideoRecord()
 			s32 size = 0;
 			u8* data = SDL_malloc(FRAME_SIZE * studio.video.frame);
 
-			gif_write_animation(data, &size, TIC80_WIDTH, TIC80_HEIGHT, (const u8*)studio.video.buffer, studio.video.frame, TIC_FRAMERATE, getConfig()->gifScale);
+			gif_write_animation(data, &size, TIC80_FULLWIDTH, TIC80_FULLHEIGHT, (const u8*)studio.video.buffer, studio.video.frame, TIC_FRAMERATE, getConfig()->gifScale);
 
 			fsGetFileData(onVideoExported, "screen.gif", data, size, DEFAULT_CHMOD, NULL);
 		}
@@ -1835,6 +1798,7 @@ static void blitSound()
 {
 	s32 samples = studio.audioSpec.freq / TIC_FRAMERATE;
 
+	// TODO: use SDL_ConvertAudio to covert format
 	if(studio.audioSpec.format == AUDIO_F32)
 	{
 		if(!studio.floatSamples)
@@ -1852,7 +1816,7 @@ static void blitSound()
 		SDL_QueueAudio(studio.audioDevice, studio.tic->samples.buffer, studio.tic->samples.size);
 }
 
-static void drawRecordLabel(u8* frame, s32 pitch, s32 sx, s32 sy, const u32* color)
+static void drawRecordLabel(u32* frame, s32 pitch, s32 sx, s32 sy, const u32* color)
 {
 	static const u16 RecLabel[] =
 	{
@@ -1868,23 +1832,24 @@ static void drawRecordLabel(u8* frame, s32 pitch, s32 sx, s32 sy, const u32* col
 		for(s32 x = 0; x < sizeof RecLabel[0]*BITS_IN_BYTE; x++)
 		{
 			if(RecLabel[y] & (1 << x))
-				memcpy(&frame[((MAX_OFFSET + sx) + 15 - x + (y+sy)*(pitch/4))*4], color, sizeof *color);
+				memcpy(&frame[sx + 15 - x + (y+sy)*TIC80_FULLWIDTH], color, sizeof *color);
 		}
 	}
 }
 
-static void recordFrame(u8* pixels, s32 pitch)
+static void recordFrame(u32* pixels)
 {
 	if(studio.video.record)
 	{
 		if(studio.video.frame < studio.video.frames)
 		{
-			screen2buffer(studio.video.buffer + (TIC80_WIDTH*TIC80_HEIGHT) * studio.video.frame, pixels, pitch);
+			SDL_Rect rect = {0, 0, TIC80_FULLWIDTH, TIC80_FULLHEIGHT};
+			screen2buffer(studio.video.buffer + (TIC80_FULLWIDTH*TIC80_FULLHEIGHT) * studio.video.frame, pixels, rect);
 
 			if(studio.video.frame % TIC_FRAMERATE < TIC_FRAMERATE / 2)
 			{
 				const u32* pal = srcPaletteBlit(studio.tic->config.palette.data);
-				drawRecordLabel(pixels, pitch, TIC80_WIDTH-24, 8, &pal[tic_color_red]);
+				drawRecordLabel(pixels, TIC80_FULLWIDTH, TIC80_WIDTH-24, 8, &pal[tic_color_red]);
 			}
 
 			studio.video.frame++;
@@ -1899,42 +1864,73 @@ static void recordFrame(u8* pixels, s32 pitch)
 
 static void blitTexture()
 {
+	tic_mem* tic = studio.tic;
 	SDL_Rect rect = {0, 0, 0, 0};
 	calcTextureRect(&rect);
 
-	const s32 Pixel = rect.w / TIC80_WIDTH;
-	rect.x -= MAX_OFFSET * Pixel;
-	rect.w = rect.w * FULL_WIDTH / TIC80_WIDTH;
-
 	void* pixels = NULL;
 	s32 pitch = 0;
 	SDL_LockTexture(studio.texture, NULL, &pixels, &pitch);
 
-	if(studio.mode == TIC_RUN_MODE)
+	tic_scanline scanline = NULL;
-	{
-		rect.y += studio.tic->ram.vram.vars.offset.y * Pixel;
 
+	switch(studio.mode)
 	{
-			void* bgPixels = NULL;
+	case TIC_RUN_MODE:
-			s32 bgPitch = 0;
+		scanline = tic->api.scanline;
-			SDL_LockTexture(studio.borderTexture, NULL, &bgPixels, &bgPitch);
+		break;
-			blit(pixels, bgPixels, pitch, bgPitch);
+	case TIC_SPRITE_MODE:
-			SDL_UnlockTexture(studio.borderTexture);
+		scanline = studio.sprite.scanline;
+		break;
+	case TIC_MAP_MODE:
+		scanline = studio.map.scanline;
+		break;
+	default:
+		break;
+	}
 
-			{
+	tic->api.blit(tic, pixels, scanline);
-				SDL_Rect srcRect = {0, 0, TIC80_WIDTH, TIC80_HEIGHT};
-				SDL_RenderCopy(studio.renderer, studio.borderTexture, &srcRect, NULL);
-			}
-		}
-	}
-	else blit(pixels, NULL, pitch, 0);
 
-	recordFrame(pixels, pitch);
+	recordFrame(pixels);
 
 	SDL_UnlockTexture(studio.texture);
 
 	{
-		SDL_Rect srcRect = {0, 0, FULL_WIDTH, TIC80_HEIGHT};
+		enum {Header = OFFSET_TOP};
+		SDL_Rect srcRect = {0, 0, TIC80_FULLWIDTH, Header};
+		SDL_Rect dstRect = {0};
+		SDL_GetWindowSize(studio.window, &dstRect.w, &dstRect.h);
+		dstRect.h = rect.y;
+		SDL_RenderCopy(studio.renderer, studio.texture, &srcRect, &dstRect);
+	}
+
+	{
+		enum {Header = OFFSET_TOP};
+		SDL_Rect srcRect = {0, TIC80_FULLHEIGHT - Header, TIC80_FULLWIDTH, Header};
+		SDL_Rect dstRect = {0};
+		SDL_GetWindowSize(studio.window, &dstRect.w, &dstRect.h);
+		dstRect.y = rect.y + rect.h;
+		dstRect.h = rect.y;
+		SDL_RenderCopy(studio.renderer, studio.texture, &srcRect, &dstRect);
+	}
+
+	{
+		enum {Header = OFFSET_TOP};
+		enum {Left = OFFSET_LEFT};
+		SDL_Rect srcRect = {0, Header, Left, TIC80_HEIGHT};
+		SDL_Rect dstRect = {0};
+		SDL_GetWindowSize(studio.window, &dstRect.w, &dstRect.h);
+		dstRect.y = rect.y;
+		dstRect.h = rect.h;
+		SDL_RenderCopy(studio.renderer, studio.texture, &srcRect, &dstRect);
+	}
+
+	{
+		enum {Top = OFFSET_TOP};
+		enum {Left = OFFSET_LEFT};
+
+		SDL_Rect srcRect = {Left, Top, TIC80_WIDTH, TIC80_HEIGHT};
+
         SDL_RenderCopy(studio.renderer, studio.texture, &srcRect, &rect);
 	}
 }
@@ -2180,11 +2176,6 @@ static void tick()
 	SDL_SystemCursor cursor = studio.mouse.system;
 	studio.mouse.system = SDL_SYSTEM_CURSOR_ARROW;
 
-	{
-		const u8* pal = (u8*)(paletteBlit() + tic_tool_peek4(studio.tic->ram.vram.mapping, studio.tic->ram.vram.vars.border & 0xf));
-		SDL_SetRenderDrawColor(studio.renderer, pal[2], pal[1], pal[0], SDL_ALPHA_OPAQUE);
-	}
-
 	SDL_RenderClear(studio.renderer);
 
 	renderStudio();
@@ -2217,14 +2208,6 @@ static void initSound()
 		SDL_PauseAudioDevice(studio.audioDevice, 0);
 }
 
-static s32 textureLog2(s32 val)
-{
-	u32 rom = 0;
-	while( val >>= 1 ) rom++;
-
-	return 1 << ++rom;
-}
-
 static void initTouchGamepad()
 {
 	if (!studio.renderer)
@@ -2234,8 +2217,7 @@ static void initTouchGamepad()
 
 	if(!studio.gamepad.texture)
 	{
-		studio.gamepad.texture = SDL_CreateTexture(studio.renderer, STUDIO_PIXEL_FORMAT, SDL_TEXTUREACCESS_STREAMING,
+		studio.gamepad.texture = SDL_CreateTexture(studio.renderer, STUDIO_PIXEL_FORMAT, SDL_TEXTUREACCESS_STREAMING, TEXTURE_SIZE, TEXTURE_SIZE);
-			textureLog2(TIC80_WIDTH), textureLog2(TIC80_HEIGHT));
 		SDL_SetTextureBlendMode(studio.gamepad.texture, SDL_BLENDMODE_BLEND);
 	}
 
@@ -2334,8 +2316,8 @@ static void onFSInitialized(FileSystem* fs)
 	SDL_Init(SDL_INIT_VIDEO | SDL_INIT_AUDIO | SDL_INIT_JOYSTICK);
 
 	studio.window = SDL_CreateWindow( TIC_TITLE, SDL_WINDOWPOS_CENTERED, SDL_WINDOWPOS_CENTERED,
-		(TIC80_WIDTH+STUDIO_UI_BORDER) * STUDIO_UI_SCALE,
+		(TIC80_FULLWIDTH) * STUDIO_UI_SCALE,
-		(TIC80_HEIGHT+STUDIO_UI_BORDER) * STUDIO_UI_SCALE,
+		(TIC80_FULLHEIGHT) * STUDIO_UI_SCALE,
 		SDL_WINDOW_SHOWN | SDL_WINDOW_RESIZABLE
 #if defined(__ARM_LINUX__)
 		| SDL_WINDOW_FULLSCREEN_DESKTOP
@@ -2368,22 +2350,12 @@ static void onFSInitialized(FileSystem* fs)
 	// set the window icon before renderer is created (issues on Linux)
 	setWindowIcon();
 
-#if defined(__ARM_LINUX__)
+	studio.renderer = SDL_CreateRenderer(studio.window, -1, SDL_RENDERER_ACCELERATED | SDL_RENDERER_PRESENTVSYNC);
-	s32 renderFlags = SDL_RENDERER_SOFTWARE;
-#else
-	s32 renderFlags = SDL_RENDERER_ACCELERATED | SDL_RENDERER_PRESENTVSYNC;
-#endif
 
-	studio.renderer = SDL_CreateRenderer(studio.window, -1, renderFlags);
+	if(!studio.renderer)
-	studio.texture = SDL_CreateTexture(studio.renderer, STUDIO_PIXEL_FORMAT, SDL_TEXTUREACCESS_STREAMING,
+		studio.softwareRenderer = studio.renderer = SDL_CreateRenderer(studio.window, -1, SDL_RENDERER_SOFTWARE);
-		textureLog2(FULL_WIDTH), textureLog2(TIC80_HEIGHT));
 
-#if !defined(__ARM_LINUX__)
+	studio.texture = SDL_CreateTexture(studio.renderer, STUDIO_PIXEL_FORMAT, SDL_TEXTUREACCESS_STREAMING, TEXTURE_SIZE, TEXTURE_SIZE);
-	SDL_SetTextureBlendMode(studio.texture, SDL_BLENDMODE_BLEND);
-#endif
-
-	studio.borderTexture = SDL_CreateTexture(studio.renderer, STUDIO_PIXEL_FORMAT, SDL_TEXTUREACCESS_STREAMING,
-		textureLog2(TIC80_WIDTH), textureLog2(TIC80_HEIGHT));
 
 	initTouchGamepad();
 }
@@ -2429,6 +2401,7 @@ s32 main(s32 argc, char **argv)
 
 	createFileSystem(onFSInitialized);
 
+	{
 		u64 nextTick = SDL_GetPerformanceCounter();
 		const u64 Delta = SDL_GetPerformanceFrequency() / TIC_FRAMERATE;
 		
@@ -2443,7 +2416,7 @@ s32 main(s32 argc, char **argv)
 				SDL_Delay((u32)(delay * 1000 / SDL_GetPerformanceFrequency()));
 			else nextTick -= delay;
 		}
-
+	}
 
 #endif
 
@@ -2455,7 +2428,6 @@ s32 main(s32 argc, char **argv)
 
 	SDL_DestroyTexture(studio.gamepad.texture);
 	SDL_DestroyTexture(studio.texture);
-	SDL_DestroyTexture(studio.borderTexture);
 
 	if(studio.mouse.texture)
 		SDL_DestroyTexture(studio.mouse.texture);

src/tic.c

@@ -493,6 +493,8 @@ static void api_clear(tic_mem* memory, u8 color)
 	{
 		api_rect(memory, machine->state.clip.l, machine->state.clip.t, machine->state.clip.r - machine->state.clip.l, machine->state.clip.b - machine->state.clip.t, color);
 	}
+
+	memory->ram.vram.vars.bg = color & 0xf;
 }
 
 static s32 drawChar(tic_mem* memory, u8 symbol, s32 x, s32 y, s32 width, s32 height, u8 color, s32 scale)
@@ -1638,6 +1640,102 @@ static s32 api_save(const tic_cartridge* cart, u8* buffer)
 	return (s32)(buffer - start);
 }
 
+// copied from SDL2
+inline void memset4(void *dst, u32 val, u32 dwords)
+{
+#if defined(__GNUC__) && defined(i386)
+    s32 u0, u1, u2;
+    __asm__ __volatile__ (
+        "cld \n\t"
+        "rep ; stosl \n\t"
+        : "=&D" (u0), "=&a" (u1), "=&c" (u2)
+        : "0" (dst), "1" (val), "2" (dwords)
+        : "memory"
+    );
+#else
+    u32 _n = (dwords + 3) / 4;
+    u32 *_p = (u32*)dst;
+    u32 _val = (val);
+    if (dwords == 0)
+        return;
+    switch (dwords % 4)
+    {
+        case 0: do {    *_p++ = _val;
+        case 3:         *_p++ = _val;
+        case 2:         *_p++ = _val;
+        case 1:         *_p++ = _val;
+        } while ( --_n );
+    }
+#endif
+}
+
+static u32* paletteBlit(tic_mem* tic)
+{
+	static u32 pal[TIC_PALETTE_SIZE] = {0};
+
+	const u8* src = tic->ram.vram.palette.data;
+
+	memset(pal, 0xff, sizeof pal);
+
+	u8* dst = (u8*)pal;
+	const u8* end = src + sizeof(tic_palette);
+
+	enum{RGB = sizeof(tic_rgb)};
+
+	for(; src != end; dst++, src+=RGB)
+		for(s32 j = 0; j < RGB; j++)
+			*dst++ = *(src+(RGB-1)-j);
+
+	return pal;
+}
+
+static void api_blit(tic_mem* tic, u32* out, tic_scanline scanline)
+{
+	const u32* pal = paletteBlit(tic);
+
+	if(scanline)
+	{
+		scanline(tic, 0);
+		pal = paletteBlit(tic);
+	}
+
+	enum {Top = (TIC80_FULLHEIGHT-TIC80_HEIGHT)/2, Bottom = Top};
+	enum {Left = (TIC80_FULLWIDTH-TIC80_WIDTH)/2, Right = Left};
+
+	memset4(&out[0 * TIC80_FULLWIDTH], pal[tic->ram.vram.vars.border], TIC80_FULLWIDTH*Top);
+
+	for(s32 r = 0; r < TIC80_HEIGHT; r++)
+	{
+		memset4(&out[(r+Top) * TIC80_FULLWIDTH], pal[tic->ram.vram.vars.border], Left);
+		memset4(&out[(r+Top) * TIC80_FULLWIDTH + Left], pal[tic->ram.vram.vars.bg], TIC80_WIDTH);
+
+		{
+			s32 y = r + tic->ram.vram.vars.offset.y;
+
+			if(y < 0 || y >= TIC80_HEIGHT) continue;
+			
+			for(s32 c = 0; c < TIC80_WIDTH; c++)
+			{
+				s32 x = c + tic->ram.vram.vars.offset.x;
+
+				if(x < 0 || x >= TIC80_WIDTH) continue;
+
+				out[(c + Left) + (r+Top) * TIC80_FULLWIDTH] = pal[tic_tool_peek4(tic->ram.vram.screen.data, x + y * TIC80_WIDTH)];
+			}			
+		}
+
+		memset4(&out[(r+Top) * TIC80_FULLWIDTH + (TIC80_FULLWIDTH-Right)], pal[tic->ram.vram.vars.border], Right);
+
+		if(scanline && (r < TIC80_HEIGHT-1))
+		{
+			scanline(tic, r+1);
+			pal = paletteBlit(tic);
+		}
+	}
+
+	memset4(&out[(TIC80_FULLHEIGHT-Bottom) * TIC80_FULLWIDTH], pal[tic->ram.vram.vars.border], TIC80_FULLWIDTH*Bottom);
+}
+
 static void initApi(tic_api* api)
 {
 #define INIT_API(func) api->func = api_##func
@@ -1682,6 +1780,7 @@ static void initApi(tic_api* api)
 	INIT_API(save);
 	INIT_API(tick_start);
 	INIT_API(tick_end);
+	INIT_API(blit);
 
 #undef INIT_API
 }

src/tic.h

@@ -355,7 +355,16 @@ typedef union
 
 		struct
 		{
-			u8 border;
+			union
+			{
+				u8 colors;
+							
+				struct
+				{
+					u8 border:TIC_PALETTE_BPP;
+					u8 bg:TIC_PALETTE_BPP;
+				};
+			};
 
 			struct
 			{

src/tic80.c

@@ -111,52 +111,6 @@ TIC80_API void tic80_load(tic80* tic, void* cart, s32 size)
 	}
 }
 
-static u32* srcPaletteBlit(const u8* src)
-{
-	static u32 pal[TIC_PALETTE_SIZE] = { 0 };
-
-	memset(pal, 0xff, sizeof pal);
-
-	u8* dst = (u8*)pal;
-	const u8* end = src + sizeof(tic_palette);
-
-	enum { RGB = sizeof(tic_rgb) };
-
-	for (; src != end; dst++, src += RGB)
-		for (s32 j = 0; j < RGB; j++)
-			*dst++ = *(src + (RGB - 1) - j);
-
-	return pal;
-}
-
-static u32* paletteBlit(tic_mem* memory)
-{
-	return srcPaletteBlit(memory->ram.vram.palette.data);
-}
-
-static void blit(tic80* tic)
-{
-	tic80_local* tic80 = (tic80_local*)tic;
-
-	u32* screen = tic->screen;
-	u32* border = tic->border;
-
-	tic->offset.x = tic80->memory->ram.vram.vars.offset.x;
-	tic->offset.y = tic80->memory->ram.vram.vars.offset.y;
-
-	for (s32 r = 0, pos = 0; r < TIC80_HEIGHT; r++, screen += TIC80_WIDTH)
-	{
-		tic80->memory->api.scanline(tic80->memory, r);
-		const u32* pal = paletteBlit(tic80->memory);
-
-		tic->offset.rows[r] = tic80->memory->ram.vram.vars.offset.x;
-
-		*border++ = pal[tic_tool_peek4(tic80->memory->ram.vram.mapping, tic80->memory->ram.vram.vars.border & 0xf)];
-		for (u32* ptr = screen, c = 0; c < TIC80_WIDTH; c++, ptr++)
-			*ptr = pal[tic_tool_peek4(tic80->memory->ram.vram.screen.data, pos++)];
-	}
-}
-
 TIC80_API void tic80_tick(tic80* tic, tic80_input input)
 {
 	tic80_local* tic80 = (tic80_local*)tic;
@@ -167,7 +121,7 @@ TIC80_API void tic80_tick(tic80* tic, tic80_input input)
 	tic80->memory->api.tick(tic80->memory, &tic80->tickData);
 	tic80->memory->api.tick_end(tic80->memory);
 
-	blit(tic);
+	tic80->memory->api.blit(tic80->memory, tic->screen, tic80->memory->api.scanline);
 
 	TickCounter++;
 }

src/ticapi.h

@@ -60,6 +60,7 @@ typedef struct
 } tic_tick_data;
 
 typedef struct tic_mem tic_mem;
+typedef void(*tic_scanline)(tic_mem* memory, s32 row);
 
 typedef struct
 {
@@ -104,6 +105,7 @@ typedef struct
 
 	void (*tick_start)			(tic_mem* memory, const tic_sound* src);
 	void (*tick_end)			(tic_mem* memory);
+	void (*blit)				(tic_mem* tic, u32* out, tic_scanline scanline);
 
 	tic_script_lang (*get_script)(tic_mem* memory);
 } tic_api;