encounter/SDL
1
0
Fork 0
mirror of https://github.com/encounter/SDL.git synced 2025-12-15 08:06:26 +00:00
Code Issues Packages Projects Releases Wiki Activity

ARM: assembly optimization for SDL_FillRect

Browse Source
This commit is contained in:
Ben Avison
2019-10-24 21:15:35 -04:00
parent 7ac733f025
commit becc649ae2
2 changed files with 91 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

68
src/video/arm/pixman-arm-simd-asm.S
View File

@@ -47,6 +47,74 @@
/******************************************************************************/
.macro FillRect32_init
ldr SRC, [sp, #ARGS_STACK_OFFSET]
mov STRIDE_S, SRC
mov MASK, SRC
mov STRIDE_M, SRC
.endm
.macro FillRect16_init
ldrh SRC, [sp, #ARGS_STACK_OFFSET]
orr SRC, SRC, lsl #16
mov STRIDE_S, SRC
mov MASK, SRC
mov STRIDE_M, SRC
.endm
.macro FillRect8_init
ldrb SRC, [sp, #ARGS_STACK_OFFSET]
orr SRC, SRC, lsl #8
orr SRC, SRC, lsl #16
mov STRIDE_S, SRC
mov MASK, SRC
mov STRIDE_M, SRC
.endm
.macro FillRect_process_tail cond, numbytes, firstreg
WK4 .req SRC
WK5 .req STRIDE_S
WK6 .req MASK
WK7 .req STRIDE_M
pixst cond, numbytes, 4, DST
.unreq WK4
.unreq WK5
.unreq WK6
.unreq WK7
.endm
generate_composite_function \
FillRect32ARMSIMDAsm, 0, 0, 32, \
FLAG_DST_WRITEONLY | FLAG_COND_EXEC | FLAG_PROCESS_PRESERVES_PSR | FLAG_PROCESS_DOES_STORE | FLAG_PROCESS_PRESERVES_SCRATCH \
0, /* prefetch distance doesn't apply */ \
FillRect32_init \
nop_macro, /* newline */ \
nop_macro /* cleanup */ \
nop_macro /* process head */ \
FillRect_process_tail
generate_composite_function \
FillRect16ARMSIMDAsm, 0, 0, 16, \
FLAG_DST_WRITEONLY | FLAG_COND_EXEC | FLAG_PROCESS_PRESERVES_PSR | FLAG_PROCESS_DOES_STORE | FLAG_PROCESS_PRESERVES_SCRATCH \
0, /* prefetch distance doesn't apply */ \
FillRect16_init \
nop_macro, /* newline */ \
nop_macro /* cleanup */ \
nop_macro /* process head */ \
FillRect_process_tail
generate_composite_function \
FillRect8ARMSIMDAsm, 0, 0, 8, \
FLAG_DST_WRITEONLY | FLAG_COND_EXEC | FLAG_PROCESS_PRESERVES_PSR | FLAG_PROCESS_DOES_STORE | FLAG_PROCESS_PRESERVES_SCRATCH \
0, /* prefetch distance doesn't apply */ \
FillRect8_init \
nop_macro, /* newline */ \
nop_macro /* cleanup */ \
nop_macro /* process head */ \
FillRect_process_tail
/******************************************************************************/
/* This differs from the over_8888_8888 routine in Pixman in that the destination
* alpha component is always left unchanged, and RGB components are not
* premultiplied by alpha. It differs from BlitRGBtoRGBPixelAlpha in that