diff options
Diffstat (limited to 'vp8/common/x86')
| -rw-r--r-- | vp8/common/x86/copy_sse2.asm | 93 | ||||
| -rw-r--r-- | vp8/common/x86/copy_sse3.asm | 146 | ||||
| -rw-r--r-- | vp8/common/x86/sad_mmx.asm | 427 | ||||
| -rw-r--r-- | vp8/common/x86/sad_sse2.asm | 410 | ||||
| -rw-r--r-- | vp8/common/x86/sad_sse3.asm | 960 | ||||
| -rw-r--r-- | vp8/common/x86/sad_sse4.asm | 353 | ||||
| -rw-r--r-- | vp8/common/x86/sad_ssse3.asm | 370 |
7 files changed, 239 insertions, 2520 deletions
diff --git a/vp8/common/x86/copy_sse2.asm b/vp8/common/x86/copy_sse2.asm new file mode 100644 index 000000000..86fae2695 --- /dev/null +++ b/vp8/common/x86/copy_sse2.asm @@ -0,0 +1,93 @@ +; +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. +; +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. +; + + +%include "vpx_ports/x86_abi_support.asm" + + +;void vp8_copy32xn_sse2( +; unsigned char *src_ptr, +; int src_stride, +; unsigned char *dst_ptr, +; int dst_stride, +; int height); +global sym(vp8_copy32xn_sse2) PRIVATE +sym(vp8_copy32xn_sse2): + push rbp + mov rbp, rsp + SHADOW_ARGS_TO_STACK 5 + SAVE_XMM 7 + push rsi + push rdi + ; end prolog + + mov rsi, arg(0) ;src_ptr + mov rdi, arg(2) ;dst_ptr + + movsxd rax, dword ptr arg(1) ;src_stride + movsxd rdx, dword ptr arg(3) ;dst_stride + movsxd rcx, dword ptr arg(4) ;height + +.block_copy_sse2_loopx4: + movdqu xmm0, XMMWORD PTR [rsi] + movdqu xmm1, XMMWORD PTR [rsi + 16] + movdqu xmm2, XMMWORD PTR [rsi + rax] + movdqu xmm3, XMMWORD PTR [rsi + rax + 16] + + lea rsi, [rsi+rax*2] + + movdqu xmm4, XMMWORD PTR [rsi] + movdqu xmm5, XMMWORD PTR [rsi + 16] + movdqu xmm6, XMMWORD PTR [rsi + rax] + movdqu xmm7, XMMWORD PTR [rsi + rax + 16] + + lea rsi, [rsi+rax*2] + + movdqa XMMWORD PTR [rdi], xmm0 + movdqa XMMWORD PTR [rdi + 16], xmm1 + movdqa XMMWORD PTR [rdi + rdx], xmm2 + movdqa XMMWORD PTR [rdi + rdx + 16], xmm3 + + lea rdi, [rdi+rdx*2] + + movdqa XMMWORD PTR [rdi], xmm4 + movdqa XMMWORD PTR [rdi + 16], xmm5 + movdqa XMMWORD PTR [rdi + rdx], xmm6 + movdqa XMMWORD PTR [rdi + rdx + 16], xmm7 + + lea rdi, [rdi+rdx*2] + + sub rcx, 4 + cmp rcx, 4 + jge .block_copy_sse2_loopx4 + + cmp rcx, 0 + je .copy_is_done + +.block_copy_sse2_loop: + movdqu xmm0, XMMWORD PTR [rsi] + movdqu xmm1, XMMWORD PTR [rsi + 16] + lea rsi, [rsi+rax] + + movdqa XMMWORD PTR [rdi], xmm0 + movdqa XMMWORD PTR [rdi + 16], xmm1 + lea rdi, [rdi+rdx] + + sub rcx, 1 + jne .block_copy_sse2_loop + +.copy_is_done: + ; begin epilog + pop rdi + pop rsi + RESTORE_XMM + UNSHADOW_ARGS + pop rbp + ret diff --git a/vp8/common/x86/copy_sse3.asm b/vp8/common/x86/copy_sse3.asm new file mode 100644 index 000000000..d789a40cc --- /dev/null +++ b/vp8/common/x86/copy_sse3.asm @@ -0,0 +1,146 @@ +; +; Copyright (c) 2010 The WebM project authors. All Rights Reserved. +; +; Use of this source code is governed by a BSD-style license +; that can be found in the LICENSE file in the root of the source +; tree. An additional intellectual property rights grant can be found +; in the file PATENTS. All contributing project authors may +; be found in the AUTHORS file in the root of the source tree. +; + +%include "vpx_ports/x86_abi_support.asm" + +%macro STACK_FRAME_CREATE_X3 0 +%if ABI_IS_32BIT + %define src_ptr rsi + %define src_stride rax + %define ref_ptr rdi + %define ref_stride rdx + %define end_ptr rcx + %define ret_var rbx + %define result_ptr arg(4) + %define max_sad arg(4) + %define height dword ptr arg(4) + push rbp + mov rbp, rsp + push rsi + push rdi + push rbx + + mov rsi, arg(0) ; src_ptr + mov rdi, arg(2) ; ref_ptr + + movsxd rax, dword ptr arg(1) ; src_stride + movsxd rdx, dword ptr arg(3) ; ref_stride +%else + %if LIBVPX_YASM_WIN64 + SAVE_XMM 7, u + %define src_ptr rcx + %define src_stride rdx + %define ref_ptr r8 + %define ref_stride r9 + %define end_ptr r10 + %define ret_var r11 + %define result_ptr [rsp+xmm_stack_space+8+4*8] + %define max_sad [rsp+xmm_stack_space+8+4*8] + %define height dword ptr [rsp+xmm_stack_space+8+4*8] + %else + %define src_ptr rdi + %define src_stride rsi + %define ref_ptr rdx + %define ref_stride rcx + %define end_ptr r9 + %define ret_var r10 + %define result_ptr r8 + %define max_sad r8 + %define height r8 + %endif +%endif + +%endmacro + +%macro STACK_FRAME_DESTROY_X3 0 + %define src_ptr + %define src_stride + %define ref_ptr + %define ref_stride + %define end_ptr + %define ret_var + %define result_ptr + %define max_sad + %define height + +%if ABI_IS_32BIT + pop rbx + pop rdi + pop rsi + pop rbp +%else + %if LIBVPX_YASM_WIN64 + RESTORE_XMM + %endif +%endif + ret +%endmacro + + +;void vp8_copy32xn_sse3( +; unsigned char *src_ptr, +; int src_stride, +; unsigned char *dst_ptr, +; int dst_stride, +; int height); +global sym(vp8_copy32xn_sse3) PRIVATE +sym(vp8_copy32xn_sse3): + + STACK_FRAME_CREATE_X3 + +.block_copy_sse3_loopx4: + lea end_ptr, [src_ptr+src_stride*2] + + movdqu xmm0, XMMWORD PTR [src_ptr] + movdqu xmm1, XMMWORD PTR [src_ptr + 16] + movdqu xmm2, XMMWORD PTR [src_ptr + src_stride] + movdqu xmm3, XMMWORD PTR [src_ptr + src_stride + 16] + movdqu xmm4, XMMWORD PTR [end_ptr] + movdqu xmm5, XMMWORD PTR [end_ptr + 16] + movdqu xmm6, XMMWORD PTR [end_ptr + src_stride] + movdqu xmm7, XMMWORD PTR [end_ptr + src_stride + 16] + + lea src_ptr, [src_ptr+src_stride*4] + + lea end_ptr, [ref_ptr+ref_stride*2] + + movdqa XMMWORD PTR [ref_ptr], xmm0 + movdqa XMMWORD PTR [ref_ptr + 16], xmm1 + movdqa XMMWORD PTR [ref_ptr + ref_stride], xmm2 + movdqa XMMWORD PTR [ref_ptr + ref_stride + 16], xmm3 + movdqa XMMWORD PTR [end_ptr], xmm4 + movdqa XMMWORD PTR [end_ptr + 16], xmm5 + movdqa XMMWORD PTR [end_ptr + ref_stride], xmm6 + movdqa XMMWORD PTR [end_ptr + ref_stride + 16], xmm7 + + lea ref_ptr, [ref_ptr+ref_stride*4] + + sub height, 4 + cmp height, 4 + jge .block_copy_sse3_loopx4 + + ;Check to see if there is more rows need to be copied. + cmp height, 0 + je .copy_is_done + +.block_copy_sse3_loop: + movdqu xmm0, XMMWORD PTR [src_ptr] + movdqu xmm1, XMMWORD PTR [src_ptr + 16] + lea src_ptr, [src_ptr+src_stride] + + movdqa XMMWORD PTR [ref_ptr], xmm0 + movdqa XMMWORD PTR [ref_ptr + 16], xmm1 + lea ref_ptr, [ref_ptr+ref_stride] + + sub height, 1 + jne .block_copy_sse3_loop + +.copy_is_done: + STACK_FRAME_DESTROY_X3 diff --git a/vp8/common/x86/sad_mmx.asm b/vp8/common/x86/sad_mmx.asm deleted file mode 100644 index 592112fa9..000000000 --- a/vp8/common/x86/sad_mmx.asm +++ /dev/null @@ -1,427 +0,0 @@ -; -; Copyright (c) 2010 The WebM project authors. All Rights Reserved. -; -; Use of this source code is governed by a BSD-style license -; that can be found in the LICENSE file in the root of the source -; tree. An additional intellectual property rights grant can be found -; in the file PATENTS. All contributing project authors may -; be found in the AUTHORS file in the root of the source tree. -; - - -%include "vpx_ports/x86_abi_support.asm" - -global sym(vp8_sad16x16_mmx) PRIVATE -global sym(vp8_sad8x16_mmx) PRIVATE -global sym(vp8_sad8x8_mmx) PRIVATE -global sym(vp8_sad4x4_mmx) PRIVATE -global sym(vp8_sad16x8_mmx) PRIVATE - -;unsigned int vp8_sad16x16_mmx( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride) -sym(vp8_sad16x16_mmx): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 4 - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - lea rcx, [rsi+rax*8] - - lea rcx, [rcx+rax*8] - pxor mm7, mm7 - - pxor mm6, mm6 - -.x16x16sad_mmx_loop: - - movq mm0, QWORD PTR [rsi] - movq mm2, QWORD PTR [rsi+8] - - movq mm1, QWORD PTR [rdi] - movq mm3, QWORD PTR [rdi+8] - - movq mm4, mm0 - movq mm5, mm2 - - psubusb mm0, mm1 - psubusb mm1, mm4 - - psubusb mm2, mm3 - psubusb mm3, mm5 - - por mm0, mm1 - por mm2, mm3 - - movq mm1, mm0 - movq mm3, mm2 - - punpcklbw mm0, mm6 - punpcklbw mm2, mm6 - - punpckhbw mm1, mm6 - punpckhbw mm3, mm6 - - paddw mm0, mm2 - paddw mm1, mm3 - - - lea rsi, [rsi+rax] - add rdi, rdx - - paddw mm7, mm0 - paddw mm7, mm1 - - cmp rsi, rcx - jne .x16x16sad_mmx_loop - - - movq mm0, mm7 - - punpcklwd mm0, mm6 - punpckhwd mm7, mm6 - - paddw mm0, mm7 - movq mm7, mm0 - - - psrlq mm0, 32 - paddw mm7, mm0 - - movq rax, mm7 - - pop rdi - pop rsi - mov rsp, rbp - ; begin epilog - UNSHADOW_ARGS - pop rbp - ret - - -;unsigned int vp8_sad8x16_mmx( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride) -sym(vp8_sad8x16_mmx): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 4 - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - lea rcx, [rsi+rax*8] - - lea rcx, [rcx+rax*8] - pxor mm7, mm7 - - pxor mm6, mm6 - -.x8x16sad_mmx_loop: - - movq mm0, QWORD PTR [rsi] - movq mm1, QWORD PTR [rdi] - - movq mm2, mm0 - psubusb mm0, mm1 - - psubusb mm1, mm2 - por mm0, mm1 - - movq mm2, mm0 - punpcklbw mm0, mm6 - - punpckhbw mm2, mm6 - lea rsi, [rsi+rax] - - add rdi, rdx - paddw mm7, mm0 - - paddw mm7, mm2 - cmp rsi, rcx - - jne .x8x16sad_mmx_loop - - movq mm0, mm7 - punpcklwd mm0, mm6 - - punpckhwd mm7, mm6 - paddw mm0, mm7 - - movq mm7, mm0 - psrlq mm0, 32 - - paddw mm7, mm0 - movq rax, mm7 - - pop rdi - pop rsi - mov rsp, rbp - ; begin epilog - UNSHADOW_ARGS - pop rbp - ret - - -;unsigned int vp8_sad8x8_mmx( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride) -sym(vp8_sad8x8_mmx): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 4 - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - lea rcx, [rsi+rax*8] - pxor mm7, mm7 - - pxor mm6, mm6 - -.x8x8sad_mmx_loop: - - movq mm0, QWORD PTR [rsi] - movq mm1, QWORD PTR [rdi] - - movq mm2, mm0 - psubusb mm0, mm1 - - psubusb mm1, mm2 - por mm0, mm1 - - movq mm2, mm0 - punpcklbw mm0, mm6 - - punpckhbw mm2, mm6 - paddw mm0, mm2 - - lea rsi, [rsi+rax] - add rdi, rdx - - paddw mm7, mm0 - cmp rsi, rcx - - jne .x8x8sad_mmx_loop - - movq mm0, mm7 - punpcklwd mm0, mm6 - - punpckhwd mm7, mm6 - paddw mm0, mm7 - - movq mm7, mm0 - psrlq mm0, 32 - - paddw mm7, mm0 - movq rax, mm7 - - pop rdi - pop rsi - mov rsp, rbp - ; begin epilog - UNSHADOW_ARGS - pop rbp - ret - - -;unsigned int vp8_sad4x4_mmx( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride) -sym(vp8_sad4x4_mmx): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 4 - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - movd mm0, DWORD PTR [rsi] - movd mm1, DWORD PTR [rdi] - - movd mm2, DWORD PTR [rsi+rax] - movd mm3, DWORD PTR [rdi+rdx] - - punpcklbw mm0, mm2 - punpcklbw mm1, mm3 - - movq mm2, mm0 - psubusb mm0, mm1 - - psubusb mm1, mm2 - por mm0, mm1 - - movq mm2, mm0 - pxor mm3, mm3 - - punpcklbw mm0, mm3 - punpckhbw mm2, mm3 - - paddw mm0, mm2 - - lea rsi, [rsi+rax*2] - lea rdi, [rdi+rdx*2] - - movd mm4, DWORD PTR [rsi] - movd mm5, DWORD PTR [rdi] - - movd mm6, DWORD PTR [rsi+rax] - movd mm7, DWORD PTR [rdi+rdx] - - punpcklbw mm4, mm6 - punpcklbw mm5, mm7 - - movq mm6, mm4 - psubusb mm4, mm5 - - psubusb mm5, mm6 - por mm4, mm5 - - movq mm5, mm4 - punpcklbw mm4, mm3 - - punpckhbw mm5, mm3 - paddw mm4, mm5 - - paddw mm0, mm4 - movq mm1, mm0 - - punpcklwd mm0, mm3 - punpckhwd mm1, mm3 - - paddw mm0, mm1 - movq mm1, mm0 - - psrlq mm0, 32 - paddw mm0, mm1 - - movq rax, mm0 - - pop rdi - pop rsi - mov rsp, rbp - ; begin epilog - UNSHADOW_ARGS - pop rbp - ret - - -;unsigned int vp8_sad16x8_mmx( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride) -sym(vp8_sad16x8_mmx): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 4 - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - lea rcx, [rsi+rax*8] - pxor mm7, mm7 - - pxor mm6, mm6 - -.x16x8sad_mmx_loop: - - movq mm0, [rsi] - movq mm1, [rdi] - - movq mm2, [rsi+8] - movq mm3, [rdi+8] - - movq mm4, mm0 - movq mm5, mm2 - - psubusb mm0, mm1 - psubusb mm1, mm4 - - psubusb mm2, mm3 - psubusb mm3, mm5 - - por mm0, mm1 - por mm2, mm3 - - movq mm1, mm0 - movq mm3, mm2 - - punpcklbw mm0, mm6 - punpckhbw mm1, mm6 - - punpcklbw mm2, mm6 - punpckhbw mm3, mm6 - - - paddw mm0, mm2 - paddw mm1, mm3 - - paddw mm0, mm1 - lea rsi, [rsi+rax] - - add rdi, rdx - paddw mm7, mm0 - - cmp rsi, rcx - jne .x16x8sad_mmx_loop - - movq mm0, mm7 - punpcklwd mm0, mm6 - - punpckhwd mm7, mm6 - paddw mm0, mm7 - - movq mm7, mm0 - psrlq mm0, 32 - - paddw mm7, mm0 - movq rax, mm7 - - pop rdi - pop rsi - mov rsp, rbp - ; begin epilog - UNSHADOW_ARGS - pop rbp - ret diff --git a/vp8/common/x86/sad_sse2.asm b/vp8/common/x86/sad_sse2.asm deleted file mode 100644 index 8d86abc07..000000000 --- a/vp8/common/x86/sad_sse2.asm +++ /dev/null @@ -1,410 +0,0 @@ -; -; Copyright (c) 2010 The WebM project authors. All Rights Reserved. -; -; Use of this source code is governed by a BSD-style license -; that can be found in the LICENSE file in the root of the source -; tree. An additional intellectual property rights grant can be found -; in the file PATENTS. All contributing project authors may -; be found in the AUTHORS file in the root of the source tree. -; - - -%include "vpx_ports/x86_abi_support.asm" - -;unsigned int vp8_sad16x16_wmt( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride) -global sym(vp8_sad16x16_wmt) PRIVATE -sym(vp8_sad16x16_wmt): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 4 - SAVE_XMM 6 - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - lea rcx, [rsi+rax*8] - - lea rcx, [rcx+rax*8] - pxor xmm6, xmm6 - -.x16x16sad_wmt_loop: - - movq xmm0, QWORD PTR [rsi] - movq xmm2, QWORD PTR [rsi+8] - - movq xmm1, QWORD PTR [rdi] - movq xmm3, QWORD PTR [rdi+8] - - movq xmm4, QWORD PTR [rsi+rax] - movq xmm5, QWORD PTR [rdi+rdx] - - - punpcklbw xmm0, xmm2 - punpcklbw xmm1, xmm3 - - psadbw xmm0, xmm1 - movq xmm2, QWORD PTR [rsi+rax+8] - - movq xmm3, QWORD PTR [rdi+rdx+8] - lea rsi, [rsi+rax*2] - - lea rdi, [rdi+rdx*2] - punpcklbw xmm4, xmm2 - - punpcklbw xmm5, xmm3 - psadbw xmm4, xmm5 - - paddw xmm6, xmm0 - paddw xmm6, xmm4 - - cmp rsi, rcx - jne .x16x16sad_wmt_loop - - movq xmm0, xmm6 - psrldq xmm6, 8 - - paddw xmm0, xmm6 - movq rax, xmm0 - - ; begin epilog - pop rdi - pop rsi - RESTORE_XMM - UNSHADOW_ARGS - pop rbp - ret - -;unsigned int vp8_sad8x16_wmt( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride, -; int max_sad) -global sym(vp8_sad8x16_wmt) PRIVATE -sym(vp8_sad8x16_wmt): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 - push rbx - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rbx, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - lea rcx, [rsi+rbx*8] - - lea rcx, [rcx+rbx*8] - pxor mm7, mm7 - -.x8x16sad_wmt_loop: - - movq rax, mm7 - cmp eax, arg(4) - ja .x8x16sad_wmt_early_exit - - movq mm0, QWORD PTR [rsi] - movq mm1, QWORD PTR [rdi] - - movq mm2, QWORD PTR [rsi+rbx] - movq mm3, QWORD PTR [rdi+rdx] - - psadbw mm0, mm1 - psadbw mm2, mm3 - - lea rsi, [rsi+rbx*2] - lea rdi, [rdi+rdx*2] - - paddw mm7, mm0 - paddw mm7, mm2 - - cmp rsi, rcx - jne .x8x16sad_wmt_loop - - movq rax, mm7 - -.x8x16sad_wmt_early_exit: - - ; begin epilog - pop rdi - pop rsi - pop rbx - UNSHADOW_ARGS - pop rbp - ret - - -;unsigned int vp8_sad8x8_wmt( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride) -global sym(vp8_sad8x8_wmt) PRIVATE -sym(vp8_sad8x8_wmt): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 - push rbx - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rbx, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - lea rcx, [rsi+rbx*8] - pxor mm7, mm7 - -.x8x8sad_wmt_loop: - - movq rax, mm7 - cmp eax, arg(4) - ja .x8x8sad_wmt_early_exit - - movq mm0, QWORD PTR [rsi] - movq mm1, QWORD PTR [rdi] - - psadbw mm0, mm1 - lea rsi, [rsi+rbx] - - add rdi, rdx - paddw mm7, mm0 - - cmp rsi, rcx - jne .x8x8sad_wmt_loop - - movq rax, mm7 -.x8x8sad_wmt_early_exit: - - ; begin epilog - pop rdi - pop rsi - pop rbx - UNSHADOW_ARGS - pop rbp - ret - -;unsigned int vp8_sad4x4_wmt( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride) -global sym(vp8_sad4x4_wmt) PRIVATE -sym(vp8_sad4x4_wmt): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 4 - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - movd mm0, DWORD PTR [rsi] - movd mm1, DWORD PTR [rdi] - - movd mm2, DWORD PTR [rsi+rax] - movd mm3, DWORD PTR [rdi+rdx] - - punpcklbw mm0, mm2 - punpcklbw mm1, mm3 - - psadbw mm0, mm1 - lea rsi, [rsi+rax*2] - - lea rdi, [rdi+rdx*2] - movd mm4, DWORD PTR [rsi] - - movd mm5, DWORD PTR [rdi] - movd mm6, DWORD PTR [rsi+rax] - - movd mm7, DWORD PTR [rdi+rdx] - punpcklbw mm4, mm6 - - punpcklbw mm5, mm7 - psadbw mm4, mm5 - - paddw mm0, mm4 - movq rax, mm0 - - ; begin epilog - pop rdi - pop rsi - UNSHADOW_ARGS - pop rbp - ret - - -;unsigned int vp8_sad16x8_wmt( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride) -global sym(vp8_sad16x8_wmt) PRIVATE -sym(vp8_sad16x8_wmt): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 - push rbx - push rsi - push rdi - ; end prolog - - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rbx, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - lea rcx, [rsi+rbx*8] - pxor mm7, mm7 - -.x16x8sad_wmt_loop: - - movq rax, mm7 - cmp eax, arg(4) - ja .x16x8sad_wmt_early_exit - - movq mm0, QWORD PTR [rsi] - movq mm2, QWORD PTR [rsi+8] - - movq mm1, QWORD PTR [rdi] - movq mm3, QWORD PTR [rdi+8] - - movq mm4, QWORD PTR [rsi+rbx] - movq mm5, QWORD PTR [rdi+rdx] - - psadbw mm0, mm1 - psadbw mm2, mm3 - - movq mm1, QWORD PTR [rsi+rbx+8] - movq mm3, QWORD PTR [rdi+rdx+8] - - psadbw mm4, mm5 - psadbw mm1, mm3 - - lea rsi, [rsi+rbx*2] - lea rdi, [rdi+rdx*2] - - paddw mm0, mm2 - paddw mm4, mm1 - - paddw mm7, mm0 - paddw mm7, mm4 - - cmp rsi, rcx - jne .x16x8sad_wmt_loop - - movq rax, mm7 - -.x16x8sad_wmt_early_exit: - - ; begin epilog - pop rdi - pop rsi - pop rbx - UNSHADOW_ARGS - pop rbp - ret - -;void vp8_copy32xn_sse2( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *dst_ptr, -; int dst_stride, -; int height); -global sym(vp8_copy32xn_sse2) PRIVATE -sym(vp8_copy32xn_sse2): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 - SAVE_XMM 7 - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;dst_ptr - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;dst_stride - movsxd rcx, dword ptr arg(4) ;height - -.block_copy_sse2_loopx4: - movdqu xmm0, XMMWORD PTR [rsi] - movdqu xmm1, XMMWORD PTR [rsi + 16] - movdqu xmm2, XMMWORD PTR [rsi + rax] - movdqu xmm3, XMMWORD PTR [rsi + rax + 16] - - lea rsi, [rsi+rax*2] - - movdqu xmm4, XMMWORD PTR [rsi] - movdqu xmm5, XMMWORD PTR [rsi + 16] - movdqu xmm6, XMMWORD PTR [rsi + rax] - movdqu xmm7, XMMWORD PTR [rsi + rax + 16] - - lea rsi, [rsi+rax*2] - - movdqa XMMWORD PTR [rdi], xmm0 - movdqa XMMWORD PTR [rdi + 16], xmm1 - movdqa XMMWORD PTR [rdi + rdx], xmm2 - movdqa XMMWORD PTR [rdi + rdx + 16], xmm3 - - lea rdi, [rdi+rdx*2] - - movdqa XMMWORD PTR [rdi], xmm4 - movdqa XMMWORD PTR [rdi + 16], xmm5 - movdqa XMMWORD PTR [rdi + rdx], xmm6 - movdqa XMMWORD PTR [rdi + rdx + 16], xmm7 - - lea rdi, [rdi+rdx*2] - - sub rcx, 4 - cmp rcx, 4 - jge .block_copy_sse2_loopx4 - - cmp rcx, 0 - je .copy_is_done - -.block_copy_sse2_loop: - movdqu xmm0, XMMWORD PTR [rsi] - movdqu xmm1, XMMWORD PTR [rsi + 16] - lea rsi, [rsi+rax] - - movdqa XMMWORD PTR [rdi], xmm0 - movdqa XMMWORD PTR [rdi + 16], xmm1 - lea rdi, [rdi+rdx] - - sub rcx, 1 - jne .block_copy_sse2_loop - -.copy_is_done: - ; begin epilog - pop rdi - pop rsi - RESTORE_XMM - UNSHADOW_ARGS - pop rbp - ret diff --git a/vp8/common/x86/sad_sse3.asm b/vp8/common/x86/sad_sse3.asm deleted file mode 100644 index 69c8d3769..000000000 --- a/vp8/common/x86/sad_sse3.asm +++ /dev/null @@ -1,960 +0,0 @@ -; -; Copyright (c) 2010 The WebM project authors. All Rights Reserved. -; -; Use of this source code is governed by a BSD-style license -; that can be found in the LICENSE file in the root of the source -; tree. An additional intellectual property rights grant can be found -; in the file PATENTS. All contributing project authors may -; be found in the AUTHORS file in the root of the source tree. -; - -%include "vpx_ports/x86_abi_support.asm" - -%macro STACK_FRAME_CREATE_X3 0 -%if ABI_IS_32BIT - %define src_ptr rsi - %define src_stride rax - %define ref_ptr rdi - %define ref_stride rdx - %define end_ptr rcx - %define ret_var rbx - %define result_ptr arg(4) - %define max_sad arg(4) - %define height dword ptr arg(4) - push rbp - mov rbp, rsp - push rsi - push rdi - push rbx - - mov rsi, arg(0) ; src_ptr - mov rdi, arg(2) ; ref_ptr - - movsxd rax, dword ptr arg(1) ; src_stride - movsxd rdx, dword ptr arg(3) ; ref_stride -%else - %if LIBVPX_YASM_WIN64 - SAVE_XMM 7, u - %define src_ptr rcx - %define src_stride rdx - %define ref_ptr r8 - %define ref_stride r9 - %define end_ptr r10 - %define ret_var r11 - %define result_ptr [rsp+xmm_stack_space+8+4*8] - %define max_sad [rsp+xmm_stack_space+8+4*8] - %define height dword ptr [rsp+xmm_stack_space+8+4*8] - %else - %define src_ptr rdi - %define src_stride rsi - %define ref_ptr rdx - %define ref_stride rcx - %define end_ptr r9 - %define ret_var r10 - %define result_ptr r8 - %define max_sad r8 - %define height r8 - %endif -%endif - -%endmacro - -%macro STACK_FRAME_DESTROY_X3 0 - %define src_ptr - %define src_stride - %define ref_ptr - %define ref_stride - %define end_ptr - %define ret_var - %define result_ptr - %define max_sad - %define height - -%if ABI_IS_32BIT - pop rbx - pop rdi - pop rsi - pop rbp -%else - %if LIBVPX_YASM_WIN64 - RESTORE_XMM - %endif -%endif - ret -%endmacro - -%macro STACK_FRAME_CREATE_X4 0 -%if ABI_IS_32BIT - %define src_ptr rsi - %define src_stride rax - %define r0_ptr rcx - %define r1_ptr rdx - %define r2_ptr rbx - %define r3_ptr rdi - %define ref_stride rbp - %define result_ptr arg(4) - push rbp - mov rbp, rsp - push rsi - push rdi - push rbx - - push rbp - mov rdi, arg(2) ; ref_ptr_base - - LOAD_X4_ADDRESSES rdi, rcx, rdx, rax, rdi - - mov rsi, arg(0) ; src_ptr - - movsxd rbx, dword ptr arg(1) ; src_stride - movsxd rbp, dword ptr arg(3) ; ref_stride - - xchg rbx, rax -%else - %if LIBVPX_YASM_WIN64 - SAVE_XMM 7, u - %define src_ptr rcx - %define src_stride rdx - %define r0_ptr rsi - %define r1_ptr r10 - %define r2_ptr r11 - %define r3_ptr r8 - %define ref_stride r9 - %define result_ptr [rsp+xmm_stack_space+16+4*8] - push rsi - - LOAD_X4_ADDRESSES r8, r0_ptr, r1_ptr, r2_ptr, r3_ptr - %else - %define src_ptr rdi - %define src_stride rsi - %define r0_ptr r9 - %define r1_ptr r10 - %define r2_ptr r11 - %define r3_ptr rdx - %define ref_stride rcx - %define result_ptr r8 - - LOAD_X4_ADDRESSES rdx, r0_ptr, r1_ptr, r2_ptr, r3_ptr - - %endif -%endif -%endmacro - -%macro STACK_FRAME_DESTROY_X4 0 - %define src_ptr - %define src_stride - %define r0_ptr - %define r1_ptr - %define r2_ptr - %define r3_ptr - %define ref_stride - %define result_ptr - -%if ABI_IS_32BIT - pop rbx - pop rdi - pop rsi - pop rbp -%else - %if LIBVPX_YASM_WIN64 - pop rsi - RESTORE_XMM - %endif -%endif - ret -%endmacro - -%macro PROCESS_16X2X3 5 -%if %1==0 - movdqa xmm0, XMMWORD PTR [%2] - lddqu xmm5, XMMWORD PTR [%3] - lddqu xmm6, XMMWORD PTR [%3+1] - lddqu xmm7, XMMWORD PTR [%3+2] - - psadbw xmm5, xmm0 - psadbw xmm6, xmm0 - psadbw xmm7, xmm0 -%else - movdqa xmm0, XMMWORD PTR [%2] - lddqu xmm1, XMMWORD PTR [%3] - lddqu xmm2, XMMWORD PTR [%3+1] - lddqu xmm3, XMMWORD PTR [%3+2] - - psadbw xmm1, xmm0 - psadbw xmm2, xmm0 - psadbw xmm3, xmm0 - - paddw xmm5, xmm1 - paddw xmm6, xmm2 - paddw xmm7, xmm3 -%endif - movdqa xmm0, XMMWORD PTR [%2+%4] - lddqu xmm1, XMMWORD PTR [%3+%5] - lddqu xmm2, XMMWORD PTR [%3+%5+1] - lddqu xmm3, XMMWORD PTR [%3+%5+2] - -%if %1==0 || %1==1 - lea %2, [%2+%4*2] - lea %3, [%3+%5*2] -%endif - - psadbw xmm1, xmm0 - psadbw xmm2, xmm0 - psadbw xmm3, xmm0 - - paddw xmm5, xmm1 - paddw xmm6, xmm2 - paddw xmm7, xmm3 -%endmacro - -%macro PROCESS_8X2X3 5 -%if %1==0 - movq mm0, QWORD PTR [%2] - movq mm5, QWORD PTR [%3] - movq mm6, QWORD PTR [%3+1] - movq mm7, QWORD PTR [%3+2] - - psadbw mm5, mm0 - psadbw mm6, mm0 - psadbw mm7, mm0 -%else - movq mm0, QWORD PTR [%2] - movq mm1, QWORD PTR [%3] - movq mm2, QWORD PTR [%3+1] - movq mm3, QWORD PTR [%3+2] - - psadbw mm1, mm0 - psadbw mm2, mm0 - psadbw mm3, mm0 - - paddw mm5, mm1 - paddw mm6, mm2 - paddw mm7, mm3 -%endif - movq mm0, QWORD PTR [%2+%4] - movq mm1, QWORD PTR [%3+%5] - movq mm2, QWORD PTR [%3+%5+1] - movq mm3, QWORD PTR [%3+%5+2] - -%if %1==0 || %1==1 - lea %2, [%2+%4*2] - lea %3, [%3+%5*2] -%endif - - psadbw mm1, mm0 - psadbw mm2, mm0 - psadbw mm3, mm0 - - paddw mm5, mm1 - paddw mm6, mm2 - paddw mm7, mm3 -%endmacro - -%macro LOAD_X4_ADDRESSES 5 - mov %2, [%1+REG_SZ_BYTES*0] - mov %3, [%1+REG_SZ_BYTES*1] - - mov %4, [%1+REG_SZ_BYTES*2] - mov %5, [%1+REG_SZ_BYTES*3] -%endmacro - -%macro PROCESS_16X2X4 8 -%if %1==0 - movdqa xmm0, XMMWORD PTR [%2] - lddqu xmm4, XMMWORD PTR [%3] - lddqu xmm5, XMMWORD PTR [%4] - lddqu xmm6, XMMWORD PTR [%5] - lddqu xmm7, XMMWORD PTR [%6] - - psadbw xmm4, xmm0 - psadbw xmm5, xmm0 - psadbw xmm6, xmm0 - psadbw xmm7, xmm0 -%else - movdqa xmm0, XMMWORD PTR [%2] - lddqu xmm1, XMMWORD PTR [%3] - lddqu xmm2, XMMWORD PTR [%4] - lddqu xmm3, XMMWORD PTR [%5] - - psadbw xmm1, xmm0 - psadbw xmm2, xmm0 - psadbw xmm3, xmm0 - - paddw xmm4, xmm1 - lddqu xmm1, XMMWORD PTR [%6] - paddw xmm5, xmm2 - paddw xmm6, xmm3 - - psadbw xmm1, xmm0 - paddw xmm7, xmm1 -%endif - movdqa xmm0, XMMWORD PTR [%2+%7] - lddqu xmm1, XMMWORD PTR [%3+%8] - lddqu xmm2, XMMWORD PTR [%4+%8] - lddqu xmm3, XMMWORD PTR [%5+%8] - - psadbw xmm1, xmm0 - psadbw xmm2, xmm0 - psadbw xmm3, xmm0 - - paddw xmm4, xmm1 - lddqu xmm1, XMMWORD PTR [%6+%8] - paddw xmm5, xmm2 - paddw xmm6, xmm3 - -%if %1==0 || %1==1 - lea %2, [%2+%7*2] - lea %3, [%3+%8*2] - - lea %4, [%4+%8*2] - lea %5, [%5+%8*2] - - lea %6, [%6+%8*2] -%endif - psadbw xmm1, xmm0 - paddw xmm7, xmm1 - -%endmacro - -%macro PROCESS_8X2X4 8 -%if %1==0 - movq mm0, QWORD PTR [%2] - movq mm4, QWORD PTR [%3] - movq mm5, QWORD PTR [%4] - movq mm6, QWORD PTR [%5] - movq mm7, QWORD PTR [%6] - - psadbw mm4, mm0 - psadbw mm5, mm0 - psadbw mm6, mm0 - psadbw mm7, mm0 -%else - movq mm0, QWORD PTR [%2] - movq mm1, QWORD PTR [%3] - movq mm2, QWORD PTR [%4] - movq mm3, QWORD PTR [%5] - - psadbw mm1, mm0 - psadbw mm2, mm0 - psadbw mm3, mm0 - - paddw mm4, mm1 - movq mm1, QWORD PTR [%6] - paddw mm5, mm2 - paddw mm6, mm3 - - psadbw mm1, mm0 - paddw mm7, mm1 -%endif - movq mm0, QWORD PTR [%2+%7] - movq mm1, QWORD PTR [%3+%8] - movq mm2, QWORD PTR [%4+%8] - movq mm3, QWORD PTR [%5+%8] - - psadbw mm1, mm0 - psadbw mm2, mm0 - psadbw mm3, mm0 - - paddw mm4, mm1 - movq mm1, QWORD PTR [%6+%8] - paddw mm5, mm2 - paddw mm6, mm3 - -%if %1==0 || %1==1 - lea %2, [%2+%7*2] - lea %3, [%3+%8*2] - - lea %4, [%4+%8*2] - lea %5, [%5+%8*2] - - lea %6, [%6+%8*2] -%endif - psadbw mm1, mm0 - paddw mm7, mm1 - -%endmacro - -;void int vp8_sad16x16x3_sse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride, -; int *results) -global sym(vp8_sad16x16x3_sse3) PRIVATE -sym(vp8_sad16x16x3_sse3): - - STACK_FRAME_CREATE_X3 - - PROCESS_16X2X3 0, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_16X2X3 2, src_ptr, ref_ptr, src_stride, ref_stride - - mov rcx, result_ptr - - movq xmm0, xmm5 - psrldq xmm5, 8 - - paddw xmm0, xmm5 - movd [rcx], xmm0 -;- - movq xmm0, xmm6 - psrldq xmm6, 8 - - paddw xmm0, xmm6 - movd [rcx+4], xmm0 -;- - movq xmm0, xmm7 - psrldq xmm7, 8 - - paddw xmm0, xmm7 - movd [rcx+8], xmm0 - - STACK_FRAME_DESTROY_X3 - -;void int vp8_sad16x8x3_sse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride, -; int *results) -global sym(vp8_sad16x8x3_sse3) PRIVATE -sym(vp8_sad16x8x3_sse3): - - STACK_FRAME_CREATE_X3 - - PROCESS_16X2X3 0, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_16X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_16X2X3 2, src_ptr, ref_ptr, src_stride, ref_stride - - mov rcx, result_ptr - - movq xmm0, xmm5 - psrldq xmm5, 8 - - paddw xmm0, xmm5 - movd [rcx], xmm0 -;- - movq xmm0, xmm6 - psrldq xmm6, 8 - - paddw xmm0, xmm6 - movd [rcx+4], xmm0 -;- - movq xmm0, xmm7 - psrldq xmm7, 8 - - paddw xmm0, xmm7 - movd [rcx+8], xmm0 - - STACK_FRAME_DESTROY_X3 - -;void int vp8_sad8x16x3_sse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride, -; int *results) -global sym(vp8_sad8x16x3_sse3) PRIVATE -sym(vp8_sad8x16x3_sse3): - - STACK_FRAME_CREATE_X3 - - PROCESS_8X2X3 0, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_8X2X3 2, src_ptr, ref_ptr, src_stride, ref_stride - - mov rcx, result_ptr - - punpckldq mm5, mm6 - - movq [rcx], mm5 - movd [rcx+8], mm7 - - STACK_FRAME_DESTROY_X3 - -;void int vp8_sad8x8x3_sse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride, -; int *results) -global sym(vp8_sad8x8x3_sse3) PRIVATE -sym(vp8_sad8x8x3_sse3): - - STACK_FRAME_CREATE_X3 - - PROCESS_8X2X3 0, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_8X2X3 1, src_ptr, ref_ptr, src_stride, ref_stride - PROCESS_8X2X3 2, src_ptr, ref_ptr, src_stride, ref_stride - - mov rcx, result_ptr - - punpckldq mm5, mm6 - - movq [rcx], mm5 - movd [rcx+8], mm7 - - STACK_FRAME_DESTROY_X3 - -;void int vp8_sad4x4x3_sse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride, -; int *results) -global sym(vp8_sad4x4x3_sse3) PRIVATE -sym(vp8_sad4x4x3_sse3): - - STACK_FRAME_CREATE_X3 - - movd mm0, DWORD PTR [src_ptr] - movd mm1, DWORD PTR [ref_ptr] - - movd mm2, DWORD PTR [src_ptr+src_stride] - movd mm3, DWORD PTR [ref_ptr+ref_stride] - - punpcklbw mm0, mm2 - punpcklbw mm1, mm3 - - movd mm4, DWORD PTR [ref_ptr+1] - movd mm5, DWORD PTR [ref_ptr+2] - - movd mm2, DWORD PTR [ref_ptr+ref_stride+1] - movd mm3, DWORD PTR [ref_ptr+ref_stride+2] - - psadbw mm1, mm0 - - punpcklbw mm4, mm2 - punpcklbw mm5, mm3 - - psadbw mm4, mm0 - psadbw mm5, mm0 - - lea src_ptr, [src_ptr+src_stride*2] - lea ref_ptr, [ref_ptr+ref_stride*2] - - movd mm0, DWORD PTR [src_ptr] - movd mm2, DWORD PTR [ref_ptr] - - movd mm3, DWORD PTR [src_ptr+src_stride] - movd mm6, DWORD PTR [ref_ptr+ref_stride] - - punpcklbw mm0, mm3 - punpcklbw mm2, mm6 - - movd mm3, DWORD PTR [ref_ptr+1] - movd mm7, DWORD PTR [ref_ptr+2] - - psadbw mm2, mm0 - - paddw mm1, mm2 - - movd mm2, DWORD PTR [ref_ptr+ref_stride+1] - movd mm6, DWORD PTR [ref_ptr+ref_stride+2] - - punpcklbw mm3, mm2 - punpcklbw mm7, mm6 - - psadbw mm3, mm0 - psadbw mm7, mm0 - - paddw mm3, mm4 - paddw mm7, mm5 - - mov rcx, result_ptr - - punpckldq mm1, mm3 - - movq [rcx], mm1 - movd [rcx+8], mm7 - - STACK_FRAME_DESTROY_X3 - -;unsigned int vp8_sad16x16_sse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride, -; int max_sad) -;%define lddqu movdqu -global sym(vp8_sad16x16_sse3) PRIVATE -sym(vp8_sad16x16_sse3): - - STACK_FRAME_CREATE_X3 - - mov end_ptr, 4 - pxor xmm7, xmm7 - -.vp8_sad16x16_sse3_loop: - movdqa xmm0, XMMWORD PTR [src_ptr] - movdqu xmm1, XMMWORD PTR [ref_ptr] - movdqa xmm2, XMMWORD PTR [src_ptr+src_stride] - movdqu xmm3, XMMWORD PTR [ref_ptr+ref_stride] - - lea src_ptr, [src_ptr+src_stride*2] - lea ref_ptr, [ref_ptr+ref_stride*2] - - movdqa xmm4, XMMWORD PTR [src_ptr] - movdqu xmm5, XMMWORD PTR [ref_ptr] - movdqa xmm6, XMMWORD PTR [src_ptr+src_stride] - - psadbw xmm0, xmm1 - - movdqu xmm1, XMMWORD PTR [ref_ptr+ref_stride] - - psadbw xmm2, xmm3 - psadbw xmm4, xmm5 - psadbw xmm6, xmm1 - - lea src_ptr, [src_ptr+src_stride*2] - lea ref_ptr, [ref_ptr+ref_stride*2] - - paddw xmm7, xmm0 - paddw xmm7, xmm2 - paddw xmm7, xmm4 - paddw xmm7, xmm6 - - sub end_ptr, 1 - jne .vp8_sad16x16_sse3_loop - - movq xmm0, xmm7 - psrldq xmm7, 8 - paddw xmm0, xmm7 - movq rax, xmm0 - - STACK_FRAME_DESTROY_X3 - -;void vp8_copy32xn_sse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *dst_ptr, -; int dst_stride, -; int height); -global sym(vp8_copy32xn_sse3) PRIVATE -sym(vp8_copy32xn_sse3): - - STACK_FRAME_CREATE_X3 - -.block_copy_sse3_loopx4: - lea end_ptr, [src_ptr+src_stride*2] - - movdqu xmm0, XMMWORD PTR [src_ptr] - movdqu xmm1, XMMWORD PTR [src_ptr + 16] - movdqu xmm2, XMMWORD PTR [src_ptr + src_stride] - movdqu xmm3, XMMWORD PTR [src_ptr + src_stride + 16] - movdqu xmm4, XMMWORD PTR [end_ptr] - movdqu xmm5, XMMWORD PTR [end_ptr + 16] - movdqu xmm6, XMMWORD PTR [end_ptr + src_stride] - movdqu xmm7, XMMWORD PTR [end_ptr + src_stride + 16] - - lea src_ptr, [src_ptr+src_stride*4] - - lea end_ptr, [ref_ptr+ref_stride*2] - - movdqa XMMWORD PTR [ref_ptr], xmm0 - movdqa XMMWORD PTR [ref_ptr + 16], xmm1 - movdqa XMMWORD PTR [ref_ptr + ref_stride], xmm2 - movdqa XMMWORD PTR [ref_ptr + ref_stride + 16], xmm3 - movdqa XMMWORD PTR [end_ptr], xmm4 - movdqa XMMWORD PTR [end_ptr + 16], xmm5 - movdqa XMMWORD PTR [end_ptr + ref_stride], xmm6 - movdqa XMMWORD PTR [end_ptr + ref_stride + 16], xmm7 - - lea ref_ptr, [ref_ptr+ref_stride*4] - - sub height, 4 - cmp height, 4 - jge .block_copy_sse3_loopx4 - - ;Check to see if there is more rows need to be copied. - cmp height, 0 - je .copy_is_done - -.block_copy_sse3_loop: - movdqu xmm0, XMMWORD PTR [src_ptr] - movdqu xmm1, XMMWORD PTR [src_ptr + 16] - lea src_ptr, [src_ptr+src_stride] - - movdqa XMMWORD PTR [ref_ptr], xmm0 - movdqa XMMWORD PTR [ref_ptr + 16], xmm1 - lea ref_ptr, [ref_ptr+ref_stride] - - sub height, 1 - jne .block_copy_sse3_loop - -.copy_is_done: - STACK_FRAME_DESTROY_X3 - -;void vp8_sad16x16x4d_sse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr_base, -; int ref_stride, -; int *results) -global sym(vp8_sad16x16x4d_sse3) PRIVATE -sym(vp8_sad16x16x4d_sse3): - - STACK_FRAME_CREATE_X4 - - PROCESS_16X2X4 0, 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 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 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 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 - -%if ABI_IS_32BIT - 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 - -;void vp8_sad16x8x4d_sse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr_base, -; int ref_stride, -; int *results) -global sym(vp8_sad16x8x4d_sse3) PRIVATE -sym(vp8_sad16x8x4d_sse3): - - STACK_FRAME_CREATE_X4 - - PROCESS_16X2X4 0, 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 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 - -%if ABI_IS_32BIT - 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 - -;void int vp8_sad8x16x4d_sse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride, -; int *results) -global sym(vp8_sad8x16x4d_sse3) PRIVATE -sym(vp8_sad8x16x4d_sse3): - - STACK_FRAME_CREATE_X4 - - PROCESS_8X2X4 0, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride - PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride - PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride - PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride - PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride - PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride - PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride - PROCESS_8X2X4 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride - -%if ABI_IS_32BIT - pop rbp -%endif - mov rcx, result_ptr - - punpckldq mm4, mm5 - punpckldq mm6, mm7 - - movq [rcx], mm4 - movq [rcx+8], mm6 - - STACK_FRAME_DESTROY_X4 - -;void int vp8_sad8x8x4d_sse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride, -; int *results) -global sym(vp8_sad8x8x4d_sse3) PRIVATE -sym(vp8_sad8x8x4d_sse3): - - STACK_FRAME_CREATE_X4 - - PROCESS_8X2X4 0, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride - PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride - PROCESS_8X2X4 1, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride - PROCESS_8X2X4 2, src_ptr, r0_ptr, r1_ptr, r2_ptr, r3_ptr, src_stride, ref_stride - -%if ABI_IS_32BIT - pop rbp -%endif - mov rcx, result_ptr - - punpckldq mm4, mm5 - punpckldq mm6, mm7 - - movq [rcx], mm4 - movq [rcx+8], mm6 - - STACK_FRAME_DESTROY_X4 - -;void int vp8_sad4x4x4d_sse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride, -; int *results) -global sym(vp8_sad4x4x4d_sse3) PRIVATE -sym(vp8_sad4x4x4d_sse3): - - STACK_FRAME_CREATE_X4 - - movd mm0, DWORD PTR [src_ptr] - movd mm1, DWORD PTR [r0_ptr] - - movd mm2, DWORD PTR [src_ptr+src_stride] - movd mm3, DWORD PTR [r0_ptr+ref_stride] - - punpcklbw mm0, mm2 - punpcklbw mm1, mm3 - - movd mm4, DWORD PTR [r1_ptr] - movd mm5, DWORD PTR [r2_ptr] - - movd mm6, DWORD PTR [r3_ptr] - movd mm2, DWORD PTR [r1_ptr+ref_stride] - - movd mm3, DWORD PTR [r2_ptr+ref_stride] - movd mm7, DWORD PTR [r3_ptr+ref_stride] - - psadbw mm1, mm0 - - punpcklbw mm4, mm2 - punpcklbw mm5, mm3 - - punpcklbw mm6, mm7 - psadbw mm4, mm0 - - psadbw mm5, mm0 - psadbw mm6, mm0 - - - - lea src_ptr, [src_ptr+src_stride*2] - lea r0_ptr, [r0_ptr+ref_stride*2] - - lea r1_ptr, [r1_ptr+ref_stride*2] - lea r2_ptr, [r2_ptr+ref_stride*2] - - lea r3_ptr, [r3_ptr+ref_stride*2] - - movd mm0, DWORD PTR [src_ptr] - movd mm2, DWORD PTR [r0_ptr] - - movd mm3, DWORD PTR [src_ptr+src_stride] - movd mm7, DWORD PTR [r0_ptr+ref_stride] - - punpcklbw mm0, mm3 - punpcklbw mm2, mm7 - - movd mm3, DWORD PTR [r1_ptr] - movd mm7, DWORD PTR [r2_ptr] - - psadbw mm2, mm0 -%if ABI_IS_32BIT - mov rax, rbp - - pop rbp -%define ref_stride rax -%endif - mov rsi, result_ptr - - paddw mm1, mm2 - movd [rsi], mm1 - - movd mm2, DWORD PTR [r1_ptr+ref_stride] - movd mm1, DWORD PTR [r2_ptr+ref_stride] - - punpcklbw mm3, mm2 - punpcklbw mm7, mm1 - - psadbw mm3, mm0 - psadbw mm7, mm0 - - movd mm2, DWORD PTR [r3_ptr] - movd mm1, DWORD PTR [r3_ptr+ref_stride] - - paddw mm3, mm4 - paddw mm7, mm5 - - movd [rsi+4], mm3 - punpcklbw mm2, mm1 - - movd [rsi+8], mm7 - psadbw mm2, mm0 - - paddw mm2, mm6 - movd [rsi+12], mm2 - - - STACK_FRAME_DESTROY_X4 - diff --git a/vp8/common/x86/sad_sse4.asm b/vp8/common/x86/sad_sse4.asm deleted file mode 100644 index f7fccd77c..000000000 --- a/vp8/common/x86/sad_sse4.asm +++ /dev/null @@ -1,353 +0,0 @@ -; -; Copyright (c) 2010 The WebM project authors. All Rights Reserved. -; -; Use of this source code is governed by a BSD-style license -; that can be found in the LICENSE file in the root of the source -; tree. An additional intellectual property rights grant can be found -; in the file PATENTS. All contributing project authors may -; be found in the AUTHORS file in the root of the source tree. -; - - -%include "vpx_ports/x86_abi_support.asm" - -%macro PROCESS_16X2X8 1 -%if %1 - movdqa xmm0, XMMWORD PTR [rsi] - movq xmm1, MMWORD PTR [rdi] - movq xmm3, MMWORD PTR [rdi+8] - movq xmm2, MMWORD PTR [rdi+16] - punpcklqdq xmm1, xmm3 - punpcklqdq xmm3, xmm2 - - movdqa xmm2, xmm1 - mpsadbw xmm1, xmm0, 0x0 - mpsadbw xmm2, xmm0, 0x5 - - psrldq xmm0, 8 - - movdqa xmm4, xmm3 - mpsadbw xmm3, xmm0, 0x0 - mpsadbw xmm4, xmm0, 0x5 - - paddw xmm1, xmm2 - paddw xmm1, xmm3 - paddw xmm1, xmm4 -%else - movdqa xmm0, XMMWORD PTR [rsi] - movq xmm5, MMWORD PTR [rdi] - movq xmm3, MMWORD PTR [rdi+8] - movq xmm2, MMWORD PTR [rdi+16] - punpcklqdq xmm5, xmm3 - punpcklqdq xmm3, xmm2 - - movdqa xmm2, xmm5 - mpsadbw xmm5, xmm0, 0x0 - mpsadbw xmm2, xmm0, 0x5 - - psrldq xmm0, 8 - - movdqa xmm4, xmm3 - mpsadbw xmm3, xmm0, 0x0 - mpsadbw xmm4, xmm0, 0x5 - - paddw xmm5, xmm2 - paddw xmm5, xmm3 - paddw xmm5, xmm4 - - paddw xmm1, xmm5 -%endif - movdqa xmm0, XMMWORD PTR [rsi + rax] - movq xmm5, MMWORD PTR [rdi+ rdx] - movq xmm3, MMWORD PTR [rdi+ rdx+8] - movq xmm2, MMWORD PTR [rdi+ rdx+16] - punpcklqdq xmm5, xmm3 - punpcklqdq xmm3, xmm2 - - lea rsi, [rsi+rax*2] - lea rdi, [rdi+rdx*2] - - movdqa xmm2, xmm5 - mpsadbw xmm5, xmm0, 0x0 - mpsadbw xmm2, xmm0, 0x5 - - psrldq xmm0, 8 - movdqa xmm4, xmm3 - mpsadbw xmm3, xmm0, 0x0 - mpsadbw xmm4, xmm0, 0x5 - - paddw xmm5, xmm2 - paddw xmm5, xmm3 - paddw xmm5, xmm4 - - paddw xmm1, xmm5 -%endmacro - -%macro PROCESS_8X2X8 1 -%if %1 - movq xmm0, MMWORD PTR [rsi] - movq xmm1, MMWORD PTR [rdi] - movq xmm3, MMWORD PTR [rdi+8] - punpcklqdq xmm1, xmm3 - - movdqa xmm2, xmm1 - mpsadbw xmm1, xmm0, 0x0 - mpsadbw xmm2, xmm0, 0x5 - paddw xmm1, xmm2 -%else - movq xmm0, MMWORD PTR [rsi] - movq xmm5, MMWORD PTR [rdi] - movq xmm3, MMWORD PTR [rdi+8] - punpcklqdq xmm5, xmm3 - - movdqa xmm2, xmm5 - mpsadbw xmm5, xmm0, 0x0 - mpsadbw xmm2, xmm0, 0x5 - paddw xmm5, xmm2 - - paddw xmm1, xmm5 -%endif - movq xmm0, MMWORD PTR [rsi + rax] - movq xmm5, MMWORD PTR [rdi+ rdx] - movq xmm3, MMWORD PTR [rdi+ rdx+8] - punpcklqdq xmm5, xmm3 - - lea rsi, [rsi+rax*2] - lea rdi, [rdi+rdx*2] - - movdqa xmm2, xmm5 - mpsadbw xmm5, xmm0, 0x0 - mpsadbw xmm2, xmm0, 0x5 - paddw xmm5, xmm2 - - paddw xmm1, xmm5 -%endmacro - -%macro PROCESS_4X2X8 1 -%if %1 - movd xmm0, [rsi] - movq xmm1, MMWORD PTR [rdi] - movq xmm3, MMWORD PTR [rdi+8] - punpcklqdq xmm1, xmm3 - - mpsadbw xmm1, xmm0, 0x0 -%else - movd xmm0, [rsi] - movq xmm5, MMWORD PTR [rdi] - movq xmm3, MMWORD PTR [rdi+8] - punpcklqdq xmm5, xmm3 - - mpsadbw xmm5, xmm0, 0x0 - - paddw xmm1, xmm5 -%endif - movd xmm0, [rsi + rax] - movq xmm5, MMWORD PTR [rdi+ rdx] - movq xmm3, MMWORD PTR [rdi+ rdx+8] - punpcklqdq xmm5, xmm3 - - lea rsi, [rsi+rax*2] - lea rdi, [rdi+rdx*2] - - mpsadbw xmm5, xmm0, 0x0 - - paddw xmm1, xmm5 -%endmacro - - -;void vp8_sad16x16x8_sse4( -; const unsigned char *src_ptr, -; int src_stride, -; const unsigned char *ref_ptr, -; int ref_stride, -; unsigned short *sad_array); -global sym(vp8_sad16x16x8_sse4) PRIVATE -sym(vp8_sad16x16x8_sse4): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - PROCESS_16X2X8 1 - PROCESS_16X2X8 0 - PROCESS_16X2X8 0 - PROCESS_16X2X8 0 - PROCESS_16X2X8 0 - PROCESS_16X2X8 0 - PROCESS_16X2X8 0 - PROCESS_16X2X8 0 - - mov rdi, arg(4) ;Results - movdqa XMMWORD PTR [rdi], xmm1 - - ; begin epilog - pop rdi - pop rsi - UNSHADOW_ARGS - pop rbp - ret - - -;void vp8_sad16x8x8_sse4( -; const unsigned char *src_ptr, -; int src_stride, -; const unsigned char *ref_ptr, -; int ref_stride, -; unsigned short *sad_array -;); -global sym(vp8_sad16x8x8_sse4) PRIVATE -sym(vp8_sad16x8x8_sse4): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - PROCESS_16X2X8 1 - PROCESS_16X2X8 0 - PROCESS_16X2X8 0 - PROCESS_16X2X8 0 - - mov rdi, arg(4) ;Results - movdqa XMMWORD PTR [rdi], xmm1 - - ; begin epilog - pop rdi - pop rsi - UNSHADOW_ARGS - pop rbp - ret - - -;void vp8_sad8x8x8_sse4( -; const unsigned char *src_ptr, -; int src_stride, -; const unsigned char *ref_ptr, -; int ref_stride, -; unsigned short *sad_array -;); -global sym(vp8_sad8x8x8_sse4) PRIVATE -sym(vp8_sad8x8x8_sse4): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - PROCESS_8X2X8 1 - PROCESS_8X2X8 0 - PROCESS_8X2X8 0 - PROCESS_8X2X8 0 - - mov rdi, arg(4) ;Results - movdqa XMMWORD PTR [rdi], xmm1 - - ; begin epilog - pop rdi - pop rsi - UNSHADOW_ARGS - pop rbp - ret - - -;void vp8_sad8x16x8_sse4( -; const unsigned char *src_ptr, -; int src_stride, -; const unsigned char *ref_ptr, -; int ref_stride, -; unsigned short *sad_array -;); -global sym(vp8_sad8x16x8_sse4) PRIVATE -sym(vp8_sad8x16x8_sse4): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - PROCESS_8X2X8 1 - PROCESS_8X2X8 0 - PROCESS_8X2X8 0 - PROCESS_8X2X8 0 - PROCESS_8X2X8 0 - PROCESS_8X2X8 0 - PROCESS_8X2X8 0 - PROCESS_8X2X8 0 - mov rdi, arg(4) ;Results - movdqa XMMWORD PTR [rdi], xmm1 - - ; begin epilog - pop rdi - pop rsi - UNSHADOW_ARGS - pop rbp - ret - - -;void vp8_sad4x4x8_c( -; const unsigned char *src_ptr, -; int src_stride, -; const unsigned char *ref_ptr, -; int ref_stride, -; unsigned short *sad_array -;); -global sym(vp8_sad4x4x8_sse4) PRIVATE -sym(vp8_sad4x4x8_sse4): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 - push rsi - push rdi - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - PROCESS_4X2X8 1 - PROCESS_4X2X8 0 - - mov rdi, arg(4) ;Results - movdqa XMMWORD PTR [rdi], xmm1 - - ; begin epilog - pop rdi - pop rsi - UNSHADOW_ARGS - pop rbp - ret - - - - diff --git a/vp8/common/x86/sad_ssse3.asm b/vp8/common/x86/sad_ssse3.asm deleted file mode 100644 index 278fc0640..000000000 --- a/vp8/common/x86/sad_ssse3.asm +++ /dev/null @@ -1,370 +0,0 @@ -; -; Copyright (c) 2010 The WebM project authors. All Rights Reserved. -; -; Use of this source code is governed by a BSD-style license -; that can be found in the LICENSE file in the root of the source -; tree. An additional intellectual property rights grant can be found -; in the file PATENTS. All contributing project authors may -; be found in the AUTHORS file in the root of the source tree. -; - - -%include "vpx_ports/x86_abi_support.asm" - -%macro PROCESS_16X2X3 1 -%if %1 - movdqa xmm0, XMMWORD PTR [rsi] - lddqu xmm5, XMMWORD PTR [rdi] - lddqu xmm6, XMMWORD PTR [rdi+1] - lddqu xmm7, XMMWORD PTR [rdi+2] - - psadbw xmm5, xmm0 - psadbw xmm6, xmm0 - psadbw xmm7, xmm0 -%else - movdqa xmm0, XMMWORD PTR [rsi] - lddqu xmm1, XMMWORD PTR [rdi] - lddqu xmm2, XMMWORD PTR [rdi+1] - lddqu xmm3, XMMWORD PTR [rdi+2] - - psadbw xmm1, xmm0 - psadbw xmm2, xmm0 - psadbw xmm3, xmm0 - - paddw xmm5, xmm1 - paddw xmm6, xmm2 - paddw xmm7, xmm3 -%endif - movdqa xmm0, XMMWORD PTR [rsi+rax] - lddqu xmm1, XMMWORD PTR [rdi+rdx] - lddqu xmm2, XMMWORD PTR [rdi+rdx+1] - lddqu xmm3, XMMWORD PTR [rdi+rdx+2] - - lea rsi, [rsi+rax*2] - lea rdi, [rdi+rdx*2] - - psadbw xmm1, xmm0 - psadbw xmm2, xmm0 - psadbw xmm3, xmm0 - - paddw xmm5, xmm1 - paddw xmm6, xmm2 - paddw xmm7, xmm3 -%endmacro - -%macro PROCESS_16X2X3_OFFSET 2 -%if %1 - movdqa xmm0, XMMWORD PTR [rsi] - movdqa xmm4, XMMWORD PTR [rdi] - movdqa xmm7, XMMWORD PTR [rdi+16] - - movdqa xmm5, xmm7 - palignr xmm5, xmm4, %2 - - movdqa xmm6, xmm7 - palignr xmm6, xmm4, (%2+1) - - palignr xmm7, xmm4, (%2+2) - - psadbw xmm5, xmm0 - psadbw xmm6, xmm0 - psadbw xmm7, xmm0 -%else - movdqa xmm0, XMMWORD PTR [rsi] - movdqa xmm4, XMMWORD PTR [rdi] - movdqa xmm3, XMMWORD PTR [rdi+16] - - movdqa xmm1, xmm3 - palignr xmm1, xmm4, %2 - - movdqa xmm2, xmm3 - palignr xmm2, xmm4, (%2+1) - - palignr xmm3, xmm4, (%2+2) - - psadbw xmm1, xmm0 - psadbw xmm2, xmm0 - psadbw xmm3, xmm0 - - paddw xmm5, xmm1 - paddw xmm6, xmm2 - paddw xmm7, xmm3 -%endif - movdqa xmm0, XMMWORD PTR [rsi+rax] - movdqa xmm4, XMMWORD PTR [rdi+rdx] - movdqa xmm3, XMMWORD PTR [rdi+rdx+16] - - movdqa xmm1, xmm3 - palignr xmm1, xmm4, %2 - - movdqa xmm2, xmm3 - palignr xmm2, xmm4, (%2+1) - - palignr xmm3, xmm4, (%2+2) - - lea rsi, [rsi+rax*2] - lea rdi, [rdi+rdx*2] - - psadbw xmm1, xmm0 - psadbw xmm2, xmm0 - psadbw xmm3, xmm0 - - paddw xmm5, xmm1 - paddw xmm6, xmm2 - paddw xmm7, xmm3 -%endmacro - -%macro PROCESS_16X16X3_OFFSET 2 -%2_aligned_by_%1: - - sub rdi, %1 - - PROCESS_16X2X3_OFFSET 1, %1 - PROCESS_16X2X3_OFFSET 0, %1 - PROCESS_16X2X3_OFFSET 0, %1 - PROCESS_16X2X3_OFFSET 0, %1 - PROCESS_16X2X3_OFFSET 0, %1 - PROCESS_16X2X3_OFFSET 0, %1 - PROCESS_16X2X3_OFFSET 0, %1 - PROCESS_16X2X3_OFFSET 0, %1 - - jmp %2_store_off - -%endmacro - -%macro PROCESS_16X8X3_OFFSET 2 -%2_aligned_by_%1: - - sub rdi, %1 - - PROCESS_16X2X3_OFFSET 1, %1 - PROCESS_16X2X3_OFFSET 0, %1 - PROCESS_16X2X3_OFFSET 0, %1 - PROCESS_16X2X3_OFFSET 0, %1 - - jmp %2_store_off - -%endmacro - -;void int vp8_sad16x16x3_ssse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride, -; int *results) -global sym(vp8_sad16x16x3_ssse3) PRIVATE -sym(vp8_sad16x16x3_ssse3): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 - SAVE_XMM 7 - push rsi - push rdi - push rcx - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - mov rdx, 0xf - and rdx, rdi - - jmp .vp8_sad16x16x3_ssse3_skiptable -.vp8_sad16x16x3_ssse3_jumptable: - dd .vp8_sad16x16x3_ssse3_aligned_by_0 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_1 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_2 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_3 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_4 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_5 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_6 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_7 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_8 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_9 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_10 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_11 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_12 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_13 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_14 - .vp8_sad16x16x3_ssse3_do_jump - dd .vp8_sad16x16x3_ssse3_aligned_by_15 - .vp8_sad16x16x3_ssse3_do_jump -.vp8_sad16x16x3_ssse3_skiptable: - - call .vp8_sad16x16x3_ssse3_do_jump -.vp8_sad16x16x3_ssse3_do_jump: - pop rcx ; get the address of do_jump - mov rax, .vp8_sad16x16x3_ssse3_jumptable - .vp8_sad16x16x3_ssse3_do_jump - add rax, rcx ; get the absolute address of vp8_sad16x16x3_ssse3_jumptable - - movsxd rax, dword [rax + 4*rdx] ; get the 32 bit offset from the jumptable - add rcx, rax - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - jmp rcx - - PROCESS_16X16X3_OFFSET 0, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 1, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 2, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 3, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 4, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 5, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 6, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 7, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 8, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 9, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 10, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 11, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 12, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 13, .vp8_sad16x16x3_ssse3 - PROCESS_16X16X3_OFFSET 14, .vp8_sad16x16x3_ssse3 - -.vp8_sad16x16x3_ssse3_aligned_by_15: - PROCESS_16X2X3 1 - PROCESS_16X2X3 0 - PROCESS_16X2X3 0 - PROCESS_16X2X3 0 - PROCESS_16X2X3 0 - PROCESS_16X2X3 0 - PROCESS_16X2X3 0 - PROCESS_16X2X3 0 - -.vp8_sad16x16x3_ssse3_store_off: - mov rdi, arg(4) ;Results - - movq xmm0, xmm5 - psrldq xmm5, 8 - - paddw xmm0, xmm5 - movd [rdi], xmm0 -;- - movq xmm0, xmm6 - psrldq xmm6, 8 - - paddw xmm0, xmm6 - movd [rdi+4], xmm0 -;- - movq xmm0, xmm7 - psrldq xmm7, 8 - - paddw xmm0, xmm7 - movd [rdi+8], xmm0 - - ; begin epilog - pop rcx - pop rdi - pop rsi - RESTORE_XMM - UNSHADOW_ARGS - pop rbp - ret - -;void int vp8_sad16x8x3_ssse3( -; unsigned char *src_ptr, -; int src_stride, -; unsigned char *ref_ptr, -; int ref_stride, -; int *results) -global sym(vp8_sad16x8x3_ssse3) PRIVATE -sym(vp8_sad16x8x3_ssse3): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 5 - SAVE_XMM 7 - push rsi - push rdi - push rcx - ; end prolog - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;ref_ptr - - mov rdx, 0xf - and rdx, rdi - - jmp .vp8_sad16x8x3_ssse3_skiptable -.vp8_sad16x8x3_ssse3_jumptable: - dd .vp8_sad16x8x3_ssse3_aligned_by_0 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_1 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_2 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_3 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_4 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_5 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_6 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_7 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_8 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_9 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_10 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_11 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_12 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_13 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_14 - .vp8_sad16x8x3_ssse3_do_jump - dd .vp8_sad16x8x3_ssse3_aligned_by_15 - .vp8_sad16x8x3_ssse3_do_jump -.vp8_sad16x8x3_ssse3_skiptable: - - call .vp8_sad16x8x3_ssse3_do_jump -.vp8_sad16x8x3_ssse3_do_jump: - pop rcx ; get the address of do_jump - mov rax, .vp8_sad16x8x3_ssse3_jumptable - .vp8_sad16x8x3_ssse3_do_jump - add rax, rcx ; get the absolute address of vp8_sad16x8x3_ssse3_jumptable - - movsxd rax, dword [rax + 4*rdx] ; get the 32 bit offset from the jumptable - add rcx, rax - - movsxd rax, dword ptr arg(1) ;src_stride - movsxd rdx, dword ptr arg(3) ;ref_stride - - jmp rcx - - PROCESS_16X8X3_OFFSET 0, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 1, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 2, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 3, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 4, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 5, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 6, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 7, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 8, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 9, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 10, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 11, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 12, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 13, .vp8_sad16x8x3_ssse3 - PROCESS_16X8X3_OFFSET 14, .vp8_sad16x8x3_ssse3 - -.vp8_sad16x8x3_ssse3_aligned_by_15: - - PROCESS_16X2X3 1 - PROCESS_16X2X3 0 - PROCESS_16X2X3 0 - PROCESS_16X2X3 0 - -.vp8_sad16x8x3_ssse3_store_off: - mov rdi, arg(4) ;Results - - movq xmm0, xmm5 - psrldq xmm5, 8 - - paddw xmm0, xmm5 - movd [rdi], xmm0 -;- - movq xmm0, xmm6 - psrldq xmm6, 8 - - paddw xmm0, xmm6 - movd [rdi+4], xmm0 -;- - movq xmm0, xmm7 - psrldq xmm7, 8 - - paddw xmm0, xmm7 - movd [rdi+8], xmm0 - - ; begin epilog - pop rcx - pop rdi - pop rsi - RESTORE_XMM - UNSHADOW_ARGS - pop rbp - ret |