watery01 / libyuv

Automatically exported from code.google.com/p/libyuv
0 stars 0 forks source link

NV12ToRGB565_Any is slow #136

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago
sudo LIBYUV_REPEAT=1000 nice --5 ./libyuv_unittest --gtest_filter=*NV* | grep ms

NV12ToRGB565_Any (31865 ms)
NV12ToRGB565_Unaligned (6271 ms)
NV12ToRGB565_Invert (6280 ms)
NV12ToRGB565_Opt (6266 ms)

Original issue reported on code.google.com by fbarch...@chromium.org on 27 Oct 2012 at 7:12

GoogleCodeExporter commented 9 years ago
Improved in r456
NV12ToRGB565_Any (7766 ms)
NV12ToRGB565_Invert (7633 ms)
NV12ToRGB565_Unaligned (7624 ms)
NV12ToRGB565_Opt (7601 ms)

But I420 is still faster
I420ToRGB565_Any (4140 ms)

Suggest doing 1 step conversion.

Original comment by fbarch...@chromium.org on 30 Oct 2012 at 4:24

GoogleCodeExporter commented 9 years ago
Fixed in r465

Neon
NV12ToRGB565_Unaligned (4472 ms)
NV12ToRGB565_Any (4471 ms)
NV21ToRGB565_Any (4452 ms)
NV21ToRGB565_Unaligned (4450 ms)
NV21ToRGB565_Opt (4420 ms)
NV12ToRGB565_Opt (4417 ms)
NV21ToRGB565_Invert (4382 ms)
NV12ToRGB565_Invert (4361 ms)

SSSE3
NV12ToRGB565_Any (1626 ms)
NV21ToRGB565_Any (1597 ms)
NV21ToRGB565_Unaligned (1512 ms)
NV12ToRGB565_Unaligned (1509 ms)
NV21ToRGB565_Opt (1496 ms)
NV12ToRGB565_Opt (1488 ms)
NV21ToRGB565_Invert (1483 ms)
NV12ToRGB565_Invert (1477 ms)

Original comment by fbarch...@chromium.org on 2 Nov 2012 at 6:45