Unified · Split
| @@ -85,10 +85,9 @@ sym(vp8_get16x16var_sse2): · 84 unmodified lines | |||
| 85 | push rbp | 85 | push rbp |
| 86 | mov rbp, rsp | 86 | mov rbp, rsp |
| 87 | SHADOW_ARGS_TO_STACK 6 | 87 | SHADOW_ARGS_TO_STACK 6 |
| 88 | GET_GOT rbx | 88 | push rbx |
| 89 | push rsi | 89 | push rsi |
| 90 | push rdi | 90 | push rdi |
| 91 | sub rsp, 16 | ||
| 92 | ; end prolog | 91 | ; end prolog |
| 93 | 92 | ||
| 94 | mov rsi, arg(0) ;[src_ptr] | 93 | mov rsi, arg(0) ;[src_ptr] |
| @@ -97,6 +96,29 @@ sym(vp8_get16x16var_sse2): · 2 unmodified lines | |||
| 97 | movsxd rax, DWORD PTR arg(1) ;[source_stride] | 96 | movsxd rax, DWORD PTR arg(1) ;[source_stride] |
| 98 | movsxd rdx, DWORD PTR arg(3) ;[recon_stride] | 97 | movsxd rdx, DWORD PTR arg(3) ;[recon_stride] |
| 99 | 98 | ||
| 99 | ; Prefetch data | ||
| 100 | lea rcx, [rax+rax*2] | ||
| 101 | prefetcht0 [rsi] | ||
| 102 | prefetcht0 [rsi+rax] | ||
| 103 | prefetcht0 [rsi+rax*2] | ||
| 104 | prefetcht0 [rsi+rcx] | ||
| 105 | lea rbx, [rsi+rax*4] | ||
| 106 | prefetcht0 [rbx] | ||
| 107 | prefetcht0 [rbx+rax] | ||
| 108 | prefetcht0 [rbx+rax*2] | ||
| 109 | prefetcht0 [rbx+rcx] | ||
| 110 | |||
| 111 | lea rcx, [rdx+rdx*2] | ||
| 112 | prefetcht0 [rdi] | ||
| 113 | prefetcht0 [rdi+rdx] | ||
| 114 | prefetcht0 [rdi+rdx*2] | ||
| 115 | prefetcht0 [rdi+rcx] | ||
| 116 | lea rbx, [rdi+rdx*4] | ||
| 117 | prefetcht0 [rbx] | ||
| 118 | prefetcht0 [rbx+rdx] | ||
| 119 | prefetcht0 [rbx+rdx*2] | ||
| 120 | prefetcht0 [rbx+rcx] | ||
| 121 | |||
| 100 | pxor xmm0, xmm0 ; clear xmm0 for unpack | 122 | pxor xmm0, xmm0 ; clear xmm0 for unpack |
| 101 | pxor xmm7, xmm7 ; clear xmm7 for accumulating diffs | 123 | pxor xmm7, xmm7 ; clear xmm7 for accumulating diffs |
| 102 | 124 | ||
| @@ -107,6 +129,9 @@ var16loop: · 4 unmodified lines | |||
| 107 | movdqu xmm1, XMMWORD PTR [rsi] | 129 | movdqu xmm1, XMMWORD PTR [rsi] |
| 108 | movdqu xmm2, XMMWORD PTR [rdi] | 130 | movdqu xmm2, XMMWORD PTR [rdi] |
| 109 | 131 | ||
| 132 | prefetcht0 [rsi+rax*8] | ||
| 133 | prefetcht0 [rdi+rdx*8] | ||
| 134 | |||
| 110 | movdqa xmm3, xmm1 | 135 | movdqa xmm3, xmm1 |
| 111 | movdqa xmm4, xmm2 | 136 | movdqa xmm4, xmm2 |
| 112 | 137 | ||
| @@ -178,10 +203,9 @@ var16loop: · 65 unmodified lines | |||
| 178 | 203 | ||
| 179 | 204 | ||
| 180 | ; begin epilog | 205 | ; begin epilog |
| 181 | add rsp, 16 | ||
| 182 | pop rdi | 206 | pop rdi |
| 183 | pop rsi | 207 | pop rsi |
| 184 | RESTORE_GOT | 208 | pop rbx |
| 185 | UNSHADOW_ARGS | 209 | UNSHADOW_ARGS |
| 186 | pop rbp | 210 | pop rbp |
| 187 | ret | 211 | ret |