[RESOLVED] Video not decoded (on 520)

[jpsdr]

To power OFF i've done :
- Fisrt, power OFF in Windows.
- Then, cut sector off with the switch at the back of power supply. If you've not such a thing, unplug sector cable. Because only power off in Windows may not totaly cut power of on mother board.
- Wait 2 or 3 minutes. To be sure everything has gone totaly off (capacitor discharged, etc...).

As power ON/OFF is something stressfull for hardware, i've not tested it several times.
Nevertheless, i've just now power ON my video PC, and exactly reproduce the issue.
While i've not loading a "nice/correct" file, all the files from all the blu-ray of my box are not decoded, even if i try several times, starting and stoping DGIndex, etc...

So, i can test your version.
Don't expect an immediate result, i'll start a video process on my PC video, wich will take probably between 24h and 48h. When finished, i'll be able to test your version.

[admin]

OK, thanks, that will give me time to prepare the test build.

[admin]

The test version is here:

http://rationalqm.us/misc/DGIndexNV_test.rar

When you get into a fail state with the release version, run this one and see if it succeeds or also fails.

[jpsdr]

My video processing was shorter than i thought.
I've switched on my PC video this morning, issue was again present, i tested your test version, unfortunately, it changed nothing. I still have to load a "correct" file to trig display.

[jpsdr]

I've tested the 306.02 beta driver, issue is still the same. After a power OFF/ON i first have to load a "proper" file to make the file working.

[admin]

Me too. No news from nVidia yet.

[jpsdr]

Ok.

As in one of your previous post you said "I would try it but I have been unable to get the failure state to occur again.", i thought you weren't able to get the faillure again.

Waiting for nVidia, issue is not critical, as there is a workaround, so i'm not in a dead end.
Nevertheless, there is the fact that dxva in mpc-hc plays fine directly on the first attempt.
Is it possible that mpc-hc does something, some kind of initialization that dgindex is not doing ?

[admin]

As in one of your previous post you said "I would try it but I have been unable to get the failure state to occur again.", i thought you weren't able to get the faillure again.

Since that post, I've encountered it a few times.

Waiting for nVidia, issue is not critical, as there is a workaround, so i'm not in a dead end.
Nevertheless, there is the fact that dxva in mpc-hc plays fine directly on the first attempt.
Is it possible that mpc-hc does something, some kind of initialization that dgindex is not doing ?

I think it's more likely to be a race condition. Obviously we have different code paths. My CUDA/CUVID initialization was developed in consultation with nVidia.

[jpsdr]

I've tested the new official 306.23 driver version, and still the same issue.

[jpsdr]

I've tested the new official 306.81 driver version, and still the same issue.

[jpsdr]

I've tested the new beta 310.54 driver version, and still the same issue.

[jpsdr]

I've tested the 327.23 driver version, issue seems to be fixed.

[admin]

Oh, finally! Thanks for letting us know about it!