[Libva] [PATCH intel-driver 0/2] Speed up jpeg encode tests

Sean V Kelley sean.v.kelley at intel.com
Mon Oct 17 15:51:34 UTC 2016


On Mon, Oct 17, 2016 at 7:04 AM, Scott D Phillips
<scott.d.phillips at intel.com> wrote:
> On Sun, Oct 16, 2016 at 09:20:49PM -0700, Xiang, Haihao wrote:
>>
>> Hi Scott,
>>
>> Does your patches change the pass criteria? I don't see any failure
>> after applying your patches, but Common/JPEGEncodeInputTest.Full/95
>> should be failed if no change to pass criteria.
>
> It should not alter the pass-criteria for any test. Looking closer at the
> change, because the comparison is done with int8_t, it is possible that a
> surface of all 0 will compare as 'close enough' to a surface of all 255. Let me see if I
> can fix that.

Yes, I was seeing the same thing as Haihao, and was expecting the failure.

Sean

>
>>
>> Thanks
>> Haihao
>>
>>
>> > For me these two patches take the Big encode test from 44sec to
>> > ~7sec.
>> >
>> > Scott D Phillips (2):
>> >   test: use valarray for raw image comparison
>> >   test: read jpeg test data from /dev/urandom
>> >
>> >  test/i965_jpeg_encode_test.cpp | 29 +++++++++++++----------------
>> >  test/i965_jpeg_test_data.cpp   |  6 +++---
>> >  2 files changed, 16 insertions(+), 19 deletions(-)
>> >
> _______________________________________________
> Libva mailing list
> Libva at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/libva



-- 
Sean V. Kelley <sean.v.kelley at intel.com>
Open Source Technology Center / SSG
Intel Corp.


More information about the Libva mailing list