I'd like to find out if a compliant/non-compliant check would help out most users. I'm not saying that's the only problem.
I wonder if ATI disclosed in the driver release notes (it was 10.4-ish) which product lines would benefit from the new driver support for out of spec files.
IIRC Ctrl+J toggles through various OSD stats in mpc-hc. The EVR can display stats on my computer.
Though I wasn't clear earlier, I'm looking for information about playback quality (pixellation/noise) and whether DXVA is used or not.
Note that mpc-hc can use different levels of hardware acceleration, so check the whole string. If it says VLD somewhere, that's full acceleration.
Anything else (iDCT for example) is partial accel and can't be done with xbmc/ffmpeg at the moment.
mpc-hc has an option to override the lock on DXVA for out-of-spec video. Properties of the decoder I think? Maybe that would bring the same type of artifacts.
defrag: the Intel i3/5/7 do not support the standard full acceleration method AFAIK and dxva is never used with those. You can easily check in a debug log that the dxva methods are enumerated and rejected.
This thread is the first time I read about ATI 5xxx with noise/artifacts. The only known problem was for a black screen instead of video for a few persons.
I suggest a separate thread with a sample so that others can confirm the issue.
@
theccap10 thanks for confirmation on the out-of-spec idea and mpc-hc. mpc-hc is open source and it won't be hard to figure out their formula. Can you try the override on out of spec files?
I'd like to see other confirmations before thinking about doing something similar in xbmc.