Vray Beta3 GPU, weird error when rendering with 5 cards - CUDA error 719

[ramon_perez]

Hello All,

I've installed a 5th graphics card to my rig (5x gtx1080ti - watercooled) and I have used a scene to test it on, which I have rendered just fine previously using Vray GPU 3.6 / Windows 7.

I now have installed Windows 10 / Max 2017/ Forest Pack 6 Beta / Vray Next Beta 3 and I'm getting this weird messages (attached)

NOTE that the errors occur half way through the rendering process at any given time. Either with LC or BF methods. AND when I tell Vray to use 3, 4 cards, or only 1 card to render, there are no errors at all. what could be going wrong here?

Please help! Thanks!

[savage309]

Are you sure the error happens only when using more than 4 cards?
If so - can you please make sure that neither of them is overheating?

Does this happen with a particular scene or with any scene? If it is with practiluar one, can you please send it over so we can investigate?

Best,
Blago

[ramon_perez]

Thanks Blago,

It has done it now with a different scene, and also with 4 cards only, This time giving me a Cuda error 700..

The cards hit ~60c with no OC, and I notice theres a lot of throttling in percentage when rendering, before I installed 5th card, they were always fixed using 100% power, no issues..

Im going to check for possible hotspots / or air in the water blocks and retry. Thanks again.

[savage309]

What?s your PSU? Sounds like possible power problem?
It might be a good idea to send the scene over, so we could try to reproduce the crash here. This way we can confirm or deny any potential software problem.

Bets,
Blago.

[ramon_perez]

I have 2 power supplies installed, originally had 1x evga G2 1600w, now added an evga 850w for mobo, cpus, etc. Via add2psu adapter,

I ran a ?stress test? with vray by using a third scene, completely different, trying one card at a time until they hit about 70 c, all 4 older cards worked fine, except for the new card, which now its shut down completely, no video. I think it has finally given up... if it wakes up I will try everything again using only one psu, but Im pretty sure now that I got a bad card.

thx!

[francomanko]

anychance you could share your rig setup? Just interested in the 5 GPU setup...and envious

[ramon_perez]

Hello francomanko I have the cards on EKWB waterblocks connected to the cooling sys using the EKWB hepta terminal. Asus z9pe-D8ws board with two Xeon V2s. One pump/ smaller Rad for CPUs, one pump 2 rads for VGA. Still has room for 2 more cards! Thx

[matanov]

It might be a good idea to send the scene over, so we could try to reproduce the crash here. This way we can confirm or deny any potential software problem.

As Blago is asking, it will be best to run your scene here on a multi-gpu system.

[AlexP]

Hello
We're also getting this error trying the Vray NEXT trial.
Rendering the same scene multiple times, it doesnt always happen. But once it does happen, the GPUs will no longer render, instead giving a 719 error at the start of the render "Device will not be used for rendering"

Additionally, when it was working, the AI Denoiser just kept showing the same denoised image from another camera when rendering from other views.
Clicking update in the denoiser element also just displayed an old denoised image.

The normal denoiser seemed to be fine, just reallllllly slow.

Any ideas?

[AlexP]

Switching to progressive will also throw up the error when it comes to denoising, and then Max will hang


An exception occurred while rendering with 3ds Max.

This render will abort. If you render again without resolving the exception, you may experience instability.

Unknown error (Details: Function "_rtBufferCreate" caught exception: Assertion failed: "rtcRes == RTC_SUCCESS", file: C:\u\workspace\rel5.0-win64-cuda90-VS2015-build-Release\sw\wsapps\raytracing\rtsdk\rel5.0\src\Devi ce\CUDADevice.cpp, line: 180)

[savage309]

Sometimes Windows update messes up the drivers.
Try reinstalling latest driver from nvidia website and restart the PC after that to see if it helps.
Also please provide full log next time if possible.

Best,
Blago.

[AlexP]

Sorry will grab the whole log.

This is happening after latest drivers and restarts, was the first thing I tried.