Madvr doesn't work anymore ?!

[sepheas]

Hello,

I was using madvr fine since a long time and everything worked fine. After nvidia driver's update I simply cannot load anything with this renderer.

MPC-HC and any other player fail. I load a file and nothing happen. Just a freeze from MPC-HC.

Do I need to wait for newer nvidia drivers ?

Maybe a simple way is to downgrade to 0.9 ?

Yes, with older drivers it worked, but I want to keep update my vidéocard..

I'm using a gt240 on a samsung DTV.

Thx for any explanation.

And sorry for my english

[Peuj]

Hi,

Be sure you output YUY2.

[mark0077]

It's the problem with non open source software like madvr and haali as examples. When support goes. The usefulness diminishes over time. Madvr is gone for my pc as it doesn't work at all with latest nvidia drivers. Open source wins in the end over quality like madvr was.

[cyberbeing]

Hi,

Be sure you output YUY2.
madVR only accepts YV12. If you output YUY2 it definitely won't work.

It's the problem with non open source software like madvr and haali as examples. When support goes. The usefulness diminishes over time. Madvr is gone for my pc as it doesn't work at all with latest nvidia drivers. Open source wins in the end over quality like madvr was.

Since this is very likely a NVIDIA driver bug which is outside of madshi's control, it wouldn't be a bad idea for everybody affected to submit a bug report to NVIDIA if you want to see it fixed:
http://www.nvidia.com/object/driverqualityassurance.html

In the meantime you could try using 191.87, the latest Quadro driver (http://forums.guru3d.com/showpost.php?p=3455848&postcount=1) (which is newer then both 196.21 & 196.34).

[leeperry]

yay, looks like I'm gonna stick to the XP 195.62 for a while :p

[Mangix]

195.94 also has no problems with madvr :)

[edison]

AFAIK, the NVIDIA DX 10.1 part GPUs can not use madVR with FW 196+ driver, the G92,GT200 still work.

[Peuj]

madVR only accepts YV12. If you output YUY2 it definitely won't work.

Yes this what I meant... I thought YV12 and I've written YUY2... sorry

[namaiki]

yay, looks like I'm gonna stick to the XP 195.62 for a while :p

yup, I'll be sticking with 186.81 and 186.88 both for my Windows 7 setup, since that is what is working for me.

[pirlouy]

I use 196.21 drivers, and madVR works when I try it (ok, it's not perfect, but this is known issues in renderer). I use Seven x64 and Geforce 8800GT (512MB Ram version).

[sepheas]

madVR only accepts YV12. If you output YUY2 it definitely won't work.



Since this is very likely a NVIDIA driver bug which is outside of madshi's control, it wouldn't be a bad idea for everybody affected to submit a bug report to NVIDIA if you want to see it fixed:
http://www.nvidia.com/object/driverqualityassurance.html

In the meantime you could try using 191.87, the latest Quadro driver (http://forums.guru3d.com/showpost.php?p=3455848&postcount=1) (which is newer then both 196.21 & 196.34).

To report this bug to nvidia, I think it's a great idea ! We can tell this to nvidia, but I'm not sure that nvidia will look at this problem closer.

yeowww I was using 3dlut with spec. from my hdtv. So, I'm a little disappointed.

And yes... I follow the project of madvr since the beginning. That's why I output in yv12. ^^

[leeperry]

how about rolling back to 195.x

[flanger216]

AFAIK, the NVIDIA DX 10.1 part GPUs can not use madVR with FW 196+ driver, the G92,GT200 still work.

Nyet... my GT200 is also a no-go with the latest drivers.

[cyberbeing]

Anybody who was having issues with madVR try the new 196.75 driver yet?

[Razoola]

Its not working here with the latest 196.75 driver on my gtx295 or gt240. I have not got madvr working at all with any 195.xx or 196.xx drivers.

[sepheas]

And then?

We ACT or ?

Let's report it to nvidia ?

[Razoola]

And then?

We ACT or ?

Let's report it to nvidia ?

I think people would rather have a new version of madVR that fixes the issue than a driver fix..

[cyberbeing]

I think people would rather have a new version of madVR that fixes the issue than a driver fix..

That is the wrong way to look at it.

Best case scenario NVIDIA only changed how a particular function needed by madVR was called.
Slightly worse would be NVIDIA introducing a bug in a function needed by madVR.
Worse case scenario is NVIDIA flat-out removed an important function which was needed by madVR.

Your top priority should be convincing NVIDIA to issue a driver fix.
Even if madshi is able to create a temporary workaround, what's to stop NVIDIA from breaking madVR yet again in their next driver release?

I for one would rather madshi spend his very limited time coding new features, rather then creating workarounds for NVIDIA's driver bugs.

[Keiyakusha]

I for one would rather madshi spend his very limited time coding new features, rather then creating workarounds for NVIDIA's driver bugs.

Basically you saying "I have ATI so I don't care. Give me new features!". IMO thats a wrong way of thinking.

NVIDIA only changed how a particular function - madvr should change its code. unlikely it will be rolled back on nvidia's side
NVIDIA introducing a bug - author of the madvr should make a bug report to nvidia so it can be fixed
NVIDIA flat-out removed an important function - madvr should use another function as a replacement

[cyberbeing]

Basically you saying "I have ATI so I don't care. Give me new features!". IMO thats a wrong way of thinking.

Actually I hate ATI, and have only bought NVIDIA for many years now. You could even classify me as an NVIDIA fanboy to a certain extent. So much for your theory.

I have no problem using older drivers which work fine (currently using 191.87), but I see it as NVIDIA's responsibility to fix what they broke, if they did indeed break something. Knowing how long NVIDIA usually takes to fix bugs introduced in new driver series is why I suggested people submit bug reports to NVIDIA en masse in the first place. As a NVIDIA user myself, I care greatly about seeing this eventually fixed, and I wouldn't mind 'eventually' being sooner rather than later.

And yes, since using older drivers is not really a big deal, I'd rather have the new features which we have been waiting on for nearly a year now. Now if madshi wasn't highly inactive, as he has been for the past year, I may have had a different opinion.

NVIDIA only changed how a particular function - madvr should change its code. unlikely it will be rolled back on nvidia's side
Correct. Being the best case scenario, we can all hope that all it requires is a simple fix by madshi.

NVIDIA introducing a bug - author of the madvr should make a bug report to nvidia so it can be fixed
And users should report the bug as well so NVIDIA feels motivated to fix the problem. Many people complaining has a higher chance of seeing results then only a single developer complaining. madshi still reads these boards, so it's likely he is aware of this issue and has already reported it to NVIDIA, yet we still don't have a fix from either him or NVIDIA.

NVIDIA flat-out removed an important function - madvr should use another function as a replacement
If this ends up being the case, what is to say there even is another function for madVR to use as a replacement?