cpu-used=2 encoder corruption - trek

[GoogleCodeExporter]

What is the expected behavior?
psnr 44

What do you see instead?
psnr 22

What version are you using?
aug 29  VP9 Encoder v1.2.0-4156-ge80bf80

On what operating system?
win64

Can you reproduce using the vpxdec or vpxenc tools?
yes

What command line are you using?
vpxenc --tile-columns=1 --tile-rows=1 -t 1 -w 1280 -h 544 --fps=24000/1001 
--target-bitrate=1200 trek.1280x544_24Hz_P420.yuv -o trek0.tmp.vp9.webm 
--fpf=trek_vp9.fpf -p 2 --pass=2 --codec=vp9 --good --cpu-used=2 
--lag-in-frames=25 --min-q=0 --max-q=63 --end-usage=vbr --auto-alt-ref=1 
--kf-max-dist=9999 --kf-min-dist=0 --drop-frame=0 --static-thresh=0 
--bias-pct=50 --minsection-pct=0 --maxsection-pct=2000 --arnr-maxframes=7 
--arnr-strength=5 --arnr-type=3 --sharpness=0 --undershoot-pct=100 -v --psnr 

Please provide any additional information below.

Stream 0 PSNR (Overall/Avg/Y/U/V) 23.244 45.575 44.555 50.233 50.056

e:\mediatests\trek>psnr -s 1280 544 trek_vp9.webm.1280x544_24Hz_P420.yuv 
trek.1280x544_24Hz_P420.yuv
Frame    PSNR-Y          PSNR-U          PSNR-V          PSNR-All        Frame
Global:  21.752750       47.643511       47.867727       23.508213        730
Avg:     44.496846       50.219434       49.986592       45.518261        730
Min:      0.803274       41.275928       41.295014        2.564185         76

Issue affects 'surround' as well.  psnr is 11 on that one.

Original issue reported on code.google.com by fbarch...@google.com on 30 Aug 2013 at 5:57

[GoogleCodeExporter]

A similar issue was observed in another encoding, which appeared to be that the 
SSE2 version of the fdct32x32_rd code has an out-of-range issue. 

Original comment by ya...@google.com on 30 Aug 2013 at 6:08

• Changed state: Started

[GoogleCodeExporter]

Does not occur with Aug 24 build.  Does occur with Aug 29 build.
Does not occur with --cpu-used=1.  Only --cpu-used=2 

Original comment by fbarch...@google.com on 30 Aug 2013 at 6:26

[GoogleCodeExporter]

proposed fix here: 
https://gerrit.chromium.org/gerrit/#/c/67080/

Original comment by ya...@google.com on 1 Sep 2013 at 8:33

[GoogleCodeExporter]

Confirmed this is fixed in Sept 03 master build: VP9 Encoder 
v1.2.0-4176-g010c0ad

Frame    PSNR-Y          PSNR-U          PSNR-V          PSNR-All        Frame
Global:  43.204356       47.600458       47.833014       44.257632        730
Avg:     45.497396       50.324517       50.073885       46.500709        730
Min:     36.045348       40.889971       41.330013       37.277371        476

Original comment by fbarch...@google.com on 3 Sep 2013 at 11:25

[GoogleCodeExporter]

This issue was fixed:

https://gerrit.chromium.org/gerrit/#/c/67080/

Original comment by jingn...@google.com on 3 Sep 2013 at 11:28

• Changed state: Fixed