generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Add SSE3 versions for sad{32x32,64x64}x4d functions.

Browse Source 
Overall encoding about 15% faster.

Change-Id: I176a775c704317509e32eee83739721804120ff2
This commit is contained in:
Ronald S. Bultje
2013-02-05 15:21:47 -08:00
parent b499c24c2f
commit 58c983d109
2 changed files with 131 additions and 56 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
vp9/common/vp9_rtcd_defs.sh
View File

@@ -601,10 +601,10 @@ prototype void vp9_sad4x4x8 "const uint8_t *src_ptr, int src_stride, const uint
specialize vp9_sad4x4x8 sse4 specialize vp9_sad4x4x8 sse4
prototype void vp9_sad64x64x4d "const uint8_t *src_ptr, int src_stride, const uint8_t **ref_ptr, int ref_stride, unsigned int *sad_array" prototype void vp9_sad64x64x4d "const uint8_t *src_ptr, int src_stride, const uint8_t **ref_ptr, int ref_stride, unsigned int *sad_array"
specialize vp9_sad64x64x4d specialize vp9_sad64x64x4d sse3
prototype void vp9_sad32x32x4d "const uint8_t *src_ptr, int src_stride, const uint8_t **ref_ptr, int ref_stride, unsigned int *sad_array" prototype void vp9_sad32x32x4d "const uint8_t *src_ptr, int src_stride, const uint8_t **ref_ptr, int ref_stride, unsigned int *sad_array"
specialize vp9_sad32x32x4d specialize vp9_sad32x32x4d sse3
prototype void vp9_sad16x16x4d "const uint8_t *src_ptr, int src_stride, const uint8_t **ref_ptr, int ref_stride, unsigned int *sad_array" prototype void vp9_sad16x16x4d "const uint8_t *src_ptr, int src_stride, const uint8_t **ref_ptr, int ref_stride, unsigned int *sad_array"
specialize vp9_sad16x16x4d sse3 specialize vp9_sad16x16x4d sse3

183
vp9/encoder/x86/vp9_sad_sse3.asm
View File

@@ -258,49 +258,49 @@
mov %5, [%1+REG_SZ_BYTES*3] mov %5, [%1+REG_SZ_BYTES*3]
%endmacro %endmacro
%macro PROCESS_16X2X4 8 %macro PROCESS_16X2X4 8-9 0
%if %1==0 %if %1==0 || %1==3
movdqa xmm0, XMMWORD PTR [%2] movdqa xmm0, XMMWORD PTR [%2+%9]
lddqu xmm4, XMMWORD PTR [%3] lddqu xmm4, XMMWORD PTR [%3+%9]
lddqu xmm5, XMMWORD PTR [%4] lddqu xmm5, XMMWORD PTR [%4+%9]
lddqu xmm6, XMMWORD PTR [%5] lddqu xmm6, XMMWORD PTR [%5+%9]
lddqu xmm7, XMMWORD PTR [%6] lddqu xmm7, XMMWORD PTR [%6+%9]
psadbw xmm4, xmm0 psadbw xmm4, xmm0
psadbw xmm5, xmm0 psadbw xmm5, xmm0
psadbw xmm6, xmm0 psadbw xmm6, xmm0
psadbw xmm7, xmm0 psadbw xmm7, xmm0
%else %else
movdqa xmm0, XMMWORD PTR [%2] movdqa xmm0, XMMWORD PTR [%2+%9]
lddqu xmm1, XMMWORD PTR [%3] lddqu xmm1, XMMWORD PTR [%3+%9]
lddqu xmm2, XMMWORD PTR [%4] lddqu xmm2, XMMWORD PTR [%4+%9]
lddqu xmm3, XMMWORD PTR [%5] lddqu xmm3, XMMWORD PTR [%5+%9]
psadbw xmm1, xmm0 psadbw xmm1, xmm0
psadbw xmm2, xmm0 psadbw xmm2, xmm0
psadbw xmm3, xmm0 psadbw xmm3, xmm0
paddw xmm4, xmm1 paddd xmm4, xmm1
lddqu xmm1, XMMWORD PTR [%6] lddqu xmm1, XMMWORD PTR [%6+%9]
paddw xmm5, xmm2 paddd xmm5, xmm2
paddw xmm6, xmm3 paddd xmm6, xmm3
psadbw xmm1, xmm0 psadbw xmm1, xmm0
paddw xmm7, xmm1 paddd xmm7, xmm1
%endif %endif
movdqa xmm0, XMMWORD PTR [%2+%7] movdqa xmm0, XMMWORD PTR [%2+%7+%9]
lddqu xmm1, XMMWORD PTR [%3+%8] lddqu xmm1, XMMWORD PTR [%3+%8+%9]
lddqu xmm2, XMMWORD PTR [%4+%8] lddqu xmm2, XMMWORD PTR [%4+%8+%9]
lddqu xmm3, XMMWORD PTR [%5+%8] lddqu xmm3, XMMWORD PTR [%5+%8+%9]
psadbw xmm1, xmm0 psadbw xmm1, xmm0
psadbw xmm2, xmm0 psadbw xmm2, xmm0
psadbw xmm3, xmm0 psadbw xmm3, xmm0
paddw xmm4, xmm1 paddd xmm4, xmm1
lddqu xmm1, XMMWORD PTR [%6+%8] lddqu xmm1, XMMWORD PTR [%6+%8+%9]
paddw xmm5, xmm2 paddd xmm5, xmm2
paddw xmm6, xmm3 paddd xmm6, xmm3
%if %1==0 || %1==1 %if %1==0 || %1==1
lea %2, [%2+%7*2] lea %2, [%2+%7*2]
@@ -312,7 +312,7 @@
lea %6, [%6+%8*2] lea %6, [%6+%8*2]
%endif %endif
psadbw xmm1, xmm0 psadbw xmm1, xmm0
paddw xmm7, xmm1 paddd xmm7, xmm1
%endmacro %endmacro
@@ -697,6 +697,109 @@ sym(vp9_copy32xn_sse3):
.copy_is_done: .copy_is_done:
STACK_FRAME_DESTROY_X3 STACK_FRAME_DESTROY_X3
;void vp9_sad64x64x4d_sse3(
; unsigned char *src_ptr,
; int src_stride,
; unsigned char *ref_ptr_base,
; int ref_stride,
; int *results)
global sym(vp9_sad64x64x4d_sse3) PRIVATE
sym(vp9_sad64x64x4d_sse3):
STACK_FRAME_CREATE_X4
%macro PROCESS_64X8X4 2-3+
PROCESS_16X2X4 %1, %3
PROCESS_16X2X4 2, %3, 16
PROCESS_16X2X4 2, %3, 32
PROCESS_16X2X4 1, %3, 48
PROCESS_16X2X4 2, %3
PROCESS_16X2X4 2, %3, 16
PROCESS_16X2X4 2, %3, 32
PROCESS_16X2X4 1, %3, 48
PROCESS_16X2X4 2, %3
PROCESS_16X2X4 2, %3, 16
PROCESS_16X2X4 2, %3, 32
PROCESS_16X2X4 1, %3, 48
PROCESS_16X2X4 2, %3
PROCESS_16X2X4 2, %3, 16
PROCESS_16X2X4 2, %3, 32
PROCESS_16X2X4 %2, %3, 48
%endmacro
PROCESS_64X8X4 3, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_64X8X4 2, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_64X8X4 2, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_64X8X4 2, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_64X8X4 2, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_64X8X4 2, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_64X8X4 2, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_64X8X4 2, 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
%macro STORE_4D_RESULTS 0
%if ABI_IS_32BIT
pop rbp
%endif
mov rcx, result_ptr
movq xmm0, xmm4
psrldq xmm4, 8
paddd xmm0, xmm4
movd [rcx], xmm0
;-
movq xmm0, xmm5
psrldq xmm5, 8
paddd xmm0, xmm5
movd [rcx+4], xmm0
;-
movq xmm0, xmm6
psrldq xmm6, 8
paddd xmm0, xmm6
movd [rcx+8], xmm0
;-
movq xmm0, xmm7
psrldq xmm7, 8
paddd xmm0, xmm7
movd [rcx+12], xmm0
%endmacro
STORE_4D_RESULTS
STACK_FRAME_DESTROY_X4
;void vp9_sad32x32x4d_sse3(
; unsigned char *src_ptr,
; int src_stride,
; unsigned char *ref_ptr_base,
; int ref_stride,
; int *results)
global sym(vp9_sad32x32x4d_sse3) PRIVATE
sym(vp9_sad32x32x4d_sse3):
STACK_FRAME_CREATE_X4
%macro PROCESS_32X4X4 2-3+
PROCESS_16X2X4 %1, %3
PROCESS_16X2X4 1, %3, 16
PROCESS_16X2X4 2, %3
PROCESS_16X2X4 %2, %3, 16
%endmacro
PROCESS_32X4X4 3, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_32X4X4 2, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_32X4X4 2, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_32X4X4 2, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_32X4X4 2, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_32X4X4 2, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_32X4X4 2, 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_32X4X4 2, 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
STORE_4D_RESULTS
STACK_FRAME_DESTROY_X4
;void vp9_sad16x16x4d_sse3( ;void vp9_sad16x16x4d_sse3(
; unsigned char *src_ptr, ; unsigned char *src_ptr,
; int src_stride, ; int src_stride,
@@ -717,35 +820,7 @@ sym(vp9_sad16x16x4d_sse3):
PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride PROCESS_16X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
PROCESS_16X2X4 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride PROCESS_16X2X4 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride
%if ABI_IS_32BIT STORE_4D_RESULTS
pop rbp
%endif
mov rcx, result_ptr
movq xmm0, xmm4
psrldq xmm4, 8
paddw xmm0, xmm4
movd [rcx], xmm0
;-
movq xmm0, xmm5
psrldq xmm5, 8
paddw xmm0, xmm5
movd [rcx+4], xmm0
;-
movq xmm0, xmm6
psrldq xmm6, 8
paddw xmm0, xmm6
movd [rcx+8], xmm0
;-
movq xmm0, xmm7
psrldq xmm7, 8
paddw xmm0, xmm7
movd [rcx+12], xmm0
STACK_FRAME_DESTROY_X4 STACK_FRAME_DESTROY_X4
;void vp9_sad16x8x4d_sse3( ;void vp9_sad16x8x4d_sse3(