[x264-devel] [Patch] zigzag SSE2 Version 2

Axel Zeuner Axel.Zeuner at gmx.de
Tue Aug 12 07:15:10 CEST 2008


On Sunday 10 August 2008 17:17:36 Jason Garrett-Glaser wrote:
> For your info, Axel, check out
> http://git.videolan.org/?p=x264-holger.git;a=summary ; it has Holger's
> zigzag additions.  I don't think he did as many as you did, so both
> will likely be useful.
Thank you for the hint, I checked them out and had a look at it, here are some 
results on my hardware

k10 results:

.../x264$ ./checkasm --bench=zig 2>/dev/null
nop: 672
zigzag_scan_4x4_field_c: 121
zigzag_scan_4x4_field_mmx: 74
zigzag_scan_4x4_field_sse2: 68
zigzag_scan_4x4_frame_c: 230
zigzag_scan_4x4_frame_sse2: 85
zigzag_scan_8x8_field_c: 754
zigzag_scan_8x8_field_sse2: 274
zigzag_scan_8x8_frame_c: 779
zigzag_scan_8x8_frame_sse2: 347
zigzag_sub_4x4_field_c: 388
zigzag_sub_4x4_field_sse2: 166
zigzag_sub_4x4_frame_c: 402
zigzag_sub_4x4_frame_sse2: 170

.../x264-holger$ ./checkasm --bench=zig 2>/dev/null
nop: 671
zigzag_scan_4x4_field_c: 118
zigzag_scan_4x4_field_mmx: 74
zigzag_scan_4x4_frame_c: 234
zigzag_scan_4x4_frame_mmx: 121
zigzag_scan_8x8_frame_c: 763
zigzag_scan_8x8_frame_mmx: 367
zigzag_scan_8x8_frame_sse2: 408

k8 results:
../x264$ ./checkasm --bench=zig 2>/dev/null
nop: 80
zigzag_scan_4x4_field_c: 138
zigzag_scan_4x4_field_mmx: 108
zigzag_scan_4x4_field_sse2: 105
zigzag_scan_4x4_frame_c: 223
zigzag_scan_4x4_frame_sse2: 130
zigzag_scan_8x8_field_c: 692
zigzag_scan_8x8_field_sse2: 447
zigzag_scan_8x8_frame_c: 756
zigzag_scan_8x8_frame_sse2: 550
zigzag_sub_4x4_field_c: 403
zigzag_sub_4x4_field_sse2: 240
zigzag_sub_4x4_frame_c: 407
zigzag_sub_4x4_frame_sse2: 260

.../x264-holger$ ./checkasm --bench=zig 2>/dev/null
nop: 80
zigzag_scan_4x4_field_c: 138
zigzag_scan_4x4_field_mmx: 107
zigzag_scan_4x4_frame_c: 224
zigzag_scan_4x4_frame_mmx: 117
zigzag_scan_8x8_frame_c: 750
zigzag_scan_8x8_frame_mmx: 494
zigzag_scan_8x8_frame_sse2: 821

Regards,
Axel


More information about the x264-devel mailing list