LowLevel/Luma3DS-3GX
Archived
1
0
Fork 0
Code Issues Pull Requests Releases 2 Wiki Activity

rosalina: use kernel cached RAM mapping for pixel conversion

Browse Source 
Conversion time goes down by 90% with this
This commit is contained in:
TuxSH 2020-05-08 20:20:51 +01:00
parent 49c8888948
commit 037fae99d6
2 changed files with 30 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
sysmodules/rosalina/Makefile
View File

@ -130,6 +130,8 @@ $(OUTPUT).elf : $(OFILES)
@$(LD) $(LDFLAGS) $(OFILES) $(LIBPATHS) $(LIBS) -o $@
@$(NM) -CSn $@ > $(notdir $*.lst)
draw.o: CFLAGS += -O3
$(OFILES_SRC) : $(HFILES_BIN)
#---------------------------------------------------------------------------------

38
sysmodules/rosalina/source/draw.c
View File

@ -32,6 +32,9 @@
#include "memory.h"
#include "menu.h"
#include "utils.h"
#include "csvc.h"
#define KERNPA2VA(a) ((a) + (GET_VERSION_MINOR(osGetKernelVersion()) < 44 ? 0xD0000000 : 0xC0000000))
static u32 gpuSavedFramebufferAddr1, gpuSavedFramebufferAddr2, gpuSavedFramebufferFormat, gpuSavedFramebufferStride;
static u32 framebufferCacheSize;
@ -55,7 +58,7 @@ void Draw_Unlock(void)
void Draw_DrawCharacter(u32 posX, u32 posY, u32 color, char character)
{
volatile u16 *const fb = (volatile u16 *const)FB_BOTTOM_VRAM_ADDR;
u16 *const fb = (u16 *)FB_BOTTOM_VRAM_ADDR;
s32 y;
for(y = 0; y < 10; y++)
@ -308,19 +311,34 @@ static inline void Draw_ConvertPixelToBGR8(u8 *dst, const u8 *src, GSPGPU_Frameb
}
}
void Draw_ConvertFrameBufferLines(u8 *buf, u32 startingLine, u32 numLines, bool top, bool left)
typedef struct FrameBufferConvertArgs {
u8 *buf;
u8 startingLine;
u8 numLines;
bool top;
bool left;
} FrameBufferConvertArgs;
static void Draw_ConvertFrameBufferLinesKernel(const FrameBufferConvertArgs *args)
{
GSPGPU_FramebufferFormats fmt = top ? (GSPGPU_FramebufferFormats)(GPU_FB_TOP_FMT & 7) : (GSPGPU_FramebufferFormats)(GPU_FB_BOTTOM_FMT & 7);
u32 width = top ? 400 : 320;
u8 formatSizes[] = { 4, 3, 2, 2, 2 };
u32 stride = top ? GPU_FB_TOP_STRIDE : GPU_FB_BOTTOM_STRIDE;
static const u8 formatSizes[] = { 4, 3, 2, 2, 2 };
u32 pa = Draw_GetCurrentFramebufferAddress(top, left);
u8 *addr = (u8 *)PA_PTR(pa);
GSPGPU_FramebufferFormats fmt = args->top ? (GSPGPU_FramebufferFormats)(GPU_FB_TOP_FMT & 7) : (GSPGPU_FramebufferFormats)(GPU_FB_BOTTOM_FMT & 7);
u32 width = args->top ? 400 : 320;
u32 stride = args->top ? GPU_FB_TOP_STRIDE : GPU_FB_BOTTOM_STRIDE;
for (u32 y = startingLine; y < startingLine + numLines; y++)
u32 pa = Draw_GetCurrentFramebufferAddress(args->top, args->left);
u8 *addr = (u8 *)KERNPA2VA(pa);
for (u32 y = args->startingLine; y < args->startingLine + args->numLines; y++)
{
for(u32 x = 0; x < width; x++)
Draw_ConvertPixelToBGR8(buf + (x + width * y) * 3 , addr + x * stride + y * formatSizes[(u8)fmt], fmt);
Draw_ConvertPixelToBGR8(args->buf + (x + width * y) * 3 , addr + x * stride + y * formatSizes[fmt], fmt);
}
}
void Draw_ConvertFrameBufferLines(u8 *buf, u32 startingLine, u32 numLines, bool top, bool left)
{
FrameBufferConvertArgs args = { buf, (u8)startingLine, (u8)numLines, top, left };
svcCustomBackdoor(Draw_ConvertFrameBufferLinesKernel, &args);
}