Skip to content

F
ffmpeg.wasm-core
Commit 2734ba78 authored by Ronald S. Bultje's avatar Ronald S. Bultje
Browse files
Options

vp56: port x86 simd to cpuflags.

parent 5361e10a
Hide whitespace changes
Inline Side-by-side
Showing with 15 additions and 19 deletions
libavcodec/x86/vp56dsp.asm
View file @ 2734ba78
...@@ -27,7 +27,8 @@ cextern pw_64 ...@@ -27,7 +27,8 @@ cextern pw_64
SECTION .text SECTION .text
%macro DIAG4_MMX 6 %macro DIAG4 6
%if mmsize == 8
movq m0, [%1+%2] movq m0, [%1+%2]
movq m1, [%1+%3] movq m1, [%1+%3]
movq m3, m0 movq m3, m0
...@@ -64,9 +65,7 @@ SECTION .text ...@@ -64,9 +65,7 @@ SECTION .text
psraw m3, 7 psraw m3, 7
packuswb m0, m3 packuswb m0, m3
movq [%6], m0 movq [%6], m0
%endmacro %else ; mmsize == 16
%macro DIAG4_SSE2 6
movq m0, [%1+%2] movq m0, [%1+%2]
movq m1, [%1+%3] movq m1, [%1+%3]
punpcklbw m0, m7 punpcklbw m0, m7
...@@ -86,9 +85,11 @@ SECTION .text ...@@ -86,9 +85,11 @@ SECTION .text
psraw m0, 7 psraw m0, 7
packuswb m0, m0 packuswb m0, m0
movq [%6], m0 movq [%6], m0
%endif ; mmsize == 8/16
%endmacro %endmacro
%macro SPLAT4REGS_MMX 0 %macro SPLAT4REGS 0
%if mmsize == 8
movq m5, m3 movq m5, m3
punpcklwd m3, m3 punpcklwd m3, m3
movq m4, m3 movq m4, m3
...@@ -102,9 +103,7 @@ SECTION .text ...@@ -102,9 +103,7 @@ SECTION .text
movq [rsp+8*12], m4 movq [rsp+8*12], m4
movq [rsp+8*13], m5 movq [rsp+8*13], m5
movq [rsp+8*14], m2 movq [rsp+8*14], m2
%endmacro %else ; mmsize == 16
%macro SPLAT4REGS_SSE2 0
pshuflw m4, m3, 0x0 pshuflw m4, m3, 0x0
pshuflw m5, m3, 0x55 pshuflw m5, m3, 0x55
pshuflw m6, m3, 0xAA pshuflw m6, m3, 0xAA
...@@ -113,15 +112,16 @@ SECTION .text ...@@ -113,15 +112,16 @@ SECTION .text
punpcklqdq m5, m5 punpcklqdq m5, m5
punpcklqdq m6, m6 punpcklqdq m6, m6
punpcklqdq m3, m3 punpcklqdq m3, m3
%endif ; mmsize == 8/16
%endmacro %endmacro
%macro vp6_filter_diag4 2 %macro vp6_filter_diag4 0
; void ff_vp6_filter_diag4_<opt>(uint8_t *dst, uint8_t *src, int stride, ; void ff_vp6_filter_diag4_<opt>(uint8_t *dst, uint8_t *src, int stride,
; const int16_t h_weight[4], const int16_t v_weights[4]) ; const int16_t h_weight[4], const int16_t v_weights[4])
cglobal vp6_filter_diag4_%1, 5, 7, %2 cglobal vp6_filter_diag4, 5, 7, 8
mov r5, rsp ; backup stack pointer mov r5, rsp ; backup stack pointer
and rsp, ~(mmsize-1) ; align stack and rsp, ~(mmsize-1) ; align stack
%ifidn %1, sse2 %if mmsize == 16
sub rsp, 8*11 sub rsp, 8*11
%else %else
sub rsp, 8*15 sub rsp, 8*15
...@@ -162,12 +162,8 @@ cglobal vp6_filter_diag4_%1, 5, 7, %2 ...@@ -162,12 +162,8 @@ cglobal vp6_filter_diag4_%1, 5, 7, %2
RET RET
%endmacro %endmacro
INIT_MMX INIT_MMX mmx
%define DIAG4 DIAG4_MMX vp6_filter_diag4
%define SPLAT4REGS SPLAT4REGS_MMX
vp6_filter_diag4 mmx, 0
INIT_XMM INIT_XMM sse2
%define DIAG4 DIAG4_SSE2 vp6_filter_diag4
%define SPLAT4REGS SPLAT4REGS_SSE2
vp6_filter_diag4 sse2, 8
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment