diff options
author | trav90 <travawine@palemoon.org> | 2018-10-19 21:52:15 -0500 |
---|---|---|
committer | trav90 <travawine@palemoon.org> | 2018-10-19 21:52:20 -0500 |
commit | bbcc64772580c8a979288791afa02d30bc476d2e (patch) | |
tree | 437ce94c3fdd7497508e5b55de06c6d011678597 /third_party/aom/aom_dsp/x86/aom_subpixel_8t_sse2.asm | |
parent | 14805f6ddbfb173c327768fff9f81f40ce5e81b0 (diff) | |
download | UXP-bbcc64772580c8a979288791afa02d30bc476d2e.tar UXP-bbcc64772580c8a979288791afa02d30bc476d2e.tar.gz UXP-bbcc64772580c8a979288791afa02d30bc476d2e.tar.lz UXP-bbcc64772580c8a979288791afa02d30bc476d2e.tar.xz UXP-bbcc64772580c8a979288791afa02d30bc476d2e.zip |
Update aom to v1.0.0
Update aom to commit id d14c5bb4f336ef1842046089849dee4a301fbbf0.
Diffstat (limited to 'third_party/aom/aom_dsp/x86/aom_subpixel_8t_sse2.asm')
-rw-r--r-- | third_party/aom/aom_dsp/x86/aom_subpixel_8t_sse2.asm | 379 |
1 files changed, 2 insertions, 377 deletions
diff --git a/third_party/aom/aom_dsp/x86/aom_subpixel_8t_sse2.asm b/third_party/aom/aom_dsp/x86/aom_subpixel_8t_sse2.asm index b946010d3..c88fc9ffb 100644 --- a/third_party/aom/aom_dsp/x86/aom_subpixel_8t_sse2.asm +++ b/third_party/aom/aom_dsp/x86/aom_subpixel_8t_sse2.asm @@ -179,6 +179,8 @@ movq [rdi + %2], xmm0 %endm +SECTION .text + ;void aom_filter_block1d4_v8_sse2 ;( ; unsigned char *src_ptr, @@ -368,166 +370,6 @@ sym(aom_filter_block1d16_v8_sse2): pop rbp ret -global sym(aom_filter_block1d4_v8_avg_sse2) PRIVATE -sym(aom_filter_block1d4_v8_avg_sse2): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 6 - SAVE_XMM 7 - push rsi - push rdi - push rbx - ; end prolog - - ALIGN_STACK 16, rax - sub rsp, 16 * 6 - %define k0k1 [rsp + 16 * 0] - %define k2k3 [rsp + 16 * 1] - %define k5k4 [rsp + 16 * 2] - %define k6k7 [rsp + 16 * 3] - %define krd [rsp + 16 * 4] - %define zero [rsp + 16 * 5] - - GET_FILTERS_4 - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;output_ptr - - movsxd rax, DWORD PTR arg(1) ;pixels_per_line - movsxd rbx, DWORD PTR arg(3) ;out_pitch - lea rdx, [rax + rax * 2] - movsxd rcx, DWORD PTR arg(4) ;output_height - -.loop: - movd xmm0, [rsi] ;load src: row 0 - movd xmm1, [rsi + rax] ;1 - movd xmm6, [rsi + rdx * 2] ;6 - lea rsi, [rsi + rax] - movd xmm7, [rsi + rdx * 2] ;7 - movd xmm2, [rsi + rax] ;2 - movd xmm3, [rsi + rax * 2] ;3 - movd xmm4, [rsi + rdx] ;4 - movd xmm5, [rsi + rax * 4] ;5 - - APPLY_FILTER_4 1 - - lea rdi, [rdi + rbx] - dec rcx - jnz .loop - - add rsp, 16 * 6 - pop rsp - pop rbx - ; begin epilog - pop rdi - pop rsi - RESTORE_XMM - UNSHADOW_ARGS - pop rbp - ret - -global sym(aom_filter_block1d8_v8_avg_sse2) PRIVATE -sym(aom_filter_block1d8_v8_avg_sse2): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 6 - SAVE_XMM 7 - push rsi - push rdi - push rbx - ; end prolog - - ALIGN_STACK 16, rax - sub rsp, 16 * 10 - %define k0 [rsp + 16 * 0] - %define k1 [rsp + 16 * 1] - %define k2 [rsp + 16 * 2] - %define k3 [rsp + 16 * 3] - %define k4 [rsp + 16 * 4] - %define k5 [rsp + 16 * 5] - %define k6 [rsp + 16 * 6] - %define k7 [rsp + 16 * 7] - %define krd [rsp + 16 * 8] - %define zero [rsp + 16 * 9] - - GET_FILTERS - - movsxd rax, DWORD PTR arg(1) ;pixels_per_line - movsxd rbx, DWORD PTR arg(3) ;out_pitch - lea rdx, [rax + rax * 2] - movsxd rcx, DWORD PTR arg(4) ;output_height -.loop: - LOAD_VERT_8 0 - APPLY_FILTER_8 1, 0 - - lea rdi, [rdi + rbx] - dec rcx - jnz .loop - - add rsp, 16 * 10 - pop rsp - pop rbx - ; begin epilog - pop rdi - pop rsi - RESTORE_XMM - UNSHADOW_ARGS - pop rbp - ret - -global sym(aom_filter_block1d16_v8_avg_sse2) PRIVATE -sym(aom_filter_block1d16_v8_avg_sse2): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 6 - SAVE_XMM 7 - push rsi - push rdi - push rbx - ; end prolog - - ALIGN_STACK 16, rax - sub rsp, 16 * 10 - %define k0 [rsp + 16 * 0] - %define k1 [rsp + 16 * 1] - %define k2 [rsp + 16 * 2] - %define k3 [rsp + 16 * 3] - %define k4 [rsp + 16 * 4] - %define k5 [rsp + 16 * 5] - %define k6 [rsp + 16 * 6] - %define k7 [rsp + 16 * 7] - %define krd [rsp + 16 * 8] - %define zero [rsp + 16 * 9] - - GET_FILTERS - - movsxd rax, DWORD PTR arg(1) ;pixels_per_line - movsxd rbx, DWORD PTR arg(3) ;out_pitch - lea rdx, [rax + rax * 2] - movsxd rcx, DWORD PTR arg(4) ;output_height -.loop: - LOAD_VERT_8 0 - APPLY_FILTER_8 1, 0 - sub rsi, rax - - LOAD_VERT_8 8 - APPLY_FILTER_8 1, 8 - add rdi, rbx - - dec rcx - jnz .loop - - add rsp, 16 * 10 - pop rsp - pop rbx - ; begin epilog - pop rdi - pop rsi - RESTORE_XMM - UNSHADOW_ARGS - pop rbp - ret - ;void aom_filter_block1d4_h8_sse2 ;( ; unsigned char *src_ptr, @@ -771,220 +613,3 @@ sym(aom_filter_block1d16_h8_sse2): UNSHADOW_ARGS pop rbp ret - -global sym(aom_filter_block1d4_h8_avg_sse2) PRIVATE -sym(aom_filter_block1d4_h8_avg_sse2): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 6 - SAVE_XMM 7 - push rsi - push rdi - ; end prolog - - ALIGN_STACK 16, rax - sub rsp, 16 * 6 - %define k0k1 [rsp + 16 * 0] - %define k2k3 [rsp + 16 * 1] - %define k5k4 [rsp + 16 * 2] - %define k6k7 [rsp + 16 * 3] - %define krd [rsp + 16 * 4] - %define zero [rsp + 16 * 5] - - GET_FILTERS_4 - - mov rsi, arg(0) ;src_ptr - mov rdi, arg(2) ;output_ptr - - movsxd rax, DWORD PTR arg(1) ;pixels_per_line - movsxd rdx, DWORD PTR arg(3) ;out_pitch - movsxd rcx, DWORD PTR arg(4) ;output_height - -.loop: - movdqu xmm0, [rsi - 3] ;load src - - movdqa xmm1, xmm0 - movdqa xmm6, xmm0 - movdqa xmm7, xmm0 - movdqa xmm2, xmm0 - movdqa xmm3, xmm0 - movdqa xmm5, xmm0 - movdqa xmm4, xmm0 - - psrldq xmm1, 1 - psrldq xmm6, 6 - psrldq xmm7, 7 - psrldq xmm2, 2 - psrldq xmm3, 3 - psrldq xmm5, 5 - psrldq xmm4, 4 - - APPLY_FILTER_4 1 - - lea rsi, [rsi + rax] - lea rdi, [rdi + rdx] - dec rcx - jnz .loop - - add rsp, 16 * 6 - pop rsp - - ; begin epilog - pop rdi - pop rsi - RESTORE_XMM - UNSHADOW_ARGS - pop rbp - ret - -global sym(aom_filter_block1d8_h8_avg_sse2) PRIVATE -sym(aom_filter_block1d8_h8_avg_sse2): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 6 - SAVE_XMM 7 - push rsi - push rdi - ; end prolog - - ALIGN_STACK 16, rax - sub rsp, 16 * 10 - %define k0 [rsp + 16 * 0] - %define k1 [rsp + 16 * 1] - %define k2 [rsp + 16 * 2] - %define k3 [rsp + 16 * 3] - %define k4 [rsp + 16 * 4] - %define k5 [rsp + 16 * 5] - %define k6 [rsp + 16 * 6] - %define k7 [rsp + 16 * 7] - %define krd [rsp + 16 * 8] - %define zero [rsp + 16 * 9] - - GET_FILTERS - - movsxd rax, DWORD PTR arg(1) ;pixels_per_line - movsxd rdx, DWORD PTR arg(3) ;out_pitch - movsxd rcx, DWORD PTR arg(4) ;output_height - -.loop: - movdqu xmm0, [rsi - 3] ;load src - - movdqa xmm1, xmm0 - movdqa xmm6, xmm0 - movdqa xmm7, xmm0 - movdqa xmm2, xmm0 - movdqa xmm5, xmm0 - movdqa xmm3, xmm0 - movdqa xmm4, xmm0 - - psrldq xmm1, 1 - psrldq xmm6, 6 - psrldq xmm7, 7 - psrldq xmm2, 2 - psrldq xmm5, 5 - psrldq xmm3, 3 - psrldq xmm4, 4 - - APPLY_FILTER_8 1, 0 - - lea rsi, [rsi + rax] - lea rdi, [rdi + rdx] - dec rcx - jnz .loop - - add rsp, 16 * 10 - pop rsp - - ; begin epilog - pop rdi - pop rsi - RESTORE_XMM - UNSHADOW_ARGS - pop rbp - ret - -global sym(aom_filter_block1d16_h8_avg_sse2) PRIVATE -sym(aom_filter_block1d16_h8_avg_sse2): - push rbp - mov rbp, rsp - SHADOW_ARGS_TO_STACK 6 - SAVE_XMM 7 - push rsi - push rdi - ; end prolog - - ALIGN_STACK 16, rax - sub rsp, 16 * 10 - %define k0 [rsp + 16 * 0] - %define k1 [rsp + 16 * 1] - %define k2 [rsp + 16 * 2] - %define k3 [rsp + 16 * 3] - %define k4 [rsp + 16 * 4] - %define k5 [rsp + 16 * 5] - %define k6 [rsp + 16 * 6] - %define k7 [rsp + 16 * 7] - %define krd [rsp + 16 * 8] - %define zero [rsp + 16 * 9] - - GET_FILTERS - - movsxd rax, DWORD PTR arg(1) ;pixels_per_line - movsxd rdx, DWORD PTR arg(3) ;out_pitch - movsxd rcx, DWORD PTR arg(4) ;output_height - -.loop: - movdqu xmm0, [rsi - 3] ;load src - - movdqa xmm1, xmm0 - movdqa xmm6, xmm0 - movdqa xmm7, xmm0 - movdqa xmm2, xmm0 - movdqa xmm5, xmm0 - movdqa xmm3, xmm0 - movdqa xmm4, xmm0 - - psrldq xmm1, 1 - psrldq xmm6, 6 - psrldq xmm7, 7 - psrldq xmm2, 2 - psrldq xmm5, 5 - psrldq xmm3, 3 - psrldq xmm4, 4 - - APPLY_FILTER_8 1, 0 - - movdqu xmm0, [rsi + 5] ;load src - - movdqa xmm1, xmm0 - movdqa xmm6, xmm0 - movdqa xmm7, xmm0 - movdqa xmm2, xmm0 - movdqa xmm5, xmm0 - movdqa xmm3, xmm0 - movdqa xmm4, xmm0 - - psrldq xmm1, 1 - psrldq xmm6, 6 - psrldq xmm7, 7 - psrldq xmm2, 2 - psrldq xmm5, 5 - psrldq xmm3, 3 - psrldq xmm4, 4 - - APPLY_FILTER_8 1, 8 - - lea rsi, [rsi + rax] - lea rdi, [rdi + rdx] - dec rcx - jnz .loop - - add rsp, 16 * 10 - pop rsp - - ; begin epilog - pop rdi - pop rsi - RESTORE_XMM - UNSHADOW_ARGS - pop rbp - ret |