Hey.
I've just installed two new GPU:s, one 680 and one 670. So now I have:
2 GTX 670
1 GTX 680
The problem is with the new 670 that fails my renders when I try to use it. I can render the scene with the other two GPU:s, but when I enable that one I get "CUDA error 719" when I try to render it in both Cinemas picture viewer and in octanes live viewer window.
If I just make a very simple scen with one cube with a material with a texture I don't get the CUDA error, but I get like small blocks of pixels that look a little off. And they dissapear when I turn that gtx 670 off, so there seems to be something wrong with that one.
Here's the log (but without the last 268 lines that repeat the last one):
Started logging on 13.09.14 12:59:32
OctaneRender version 2.03 (2030000)
Can't Read Preset File: C:\Program Files\MAXON\CINEMA 4D R15\plugins\c4dOctanepresets.dat
Can't Read Preset File: C:\Program Files\MAXON\CINEMA 4D R15\plugins\c4dOctanepresets.dat
FileName::toJuceFile() can be called only on absolute file names
File Load Error: Could not load 'G_Nougatsas_CM.tif:rgba'.
FileName::toJuceFile() can be called only on absolute file names
File Load Error: Could not load 'Glasssaser_Alpha.tif:rgba'.
FileName::toJuceFile() can be called only on absolute file names
File Load Error: Could not load 'G_Saltlakrits_KORK_CM.tif:rgba'.
--------------------------- EXPORT LOG ---------------------------
Export materials time= 4463.669 ms
Collect objects time= 19.089 ms
Meshes creation time = 403.587 ms
Octane voxelization time = 0 ms
Sending to Octane engine time = 45.77 ms
Total export Time = 12694.23 ms
CUDA error 719 on device 1: An exception occurred on the device while executing a kernel. The context cannot be used anymore and must be destroyed. All existing device memory allocations from this context are invalid and must be reconstructed.
-> kernel execution failed (dl)
device 1: direct lighting failed
<<< Render Failure >>>
CUDA error 719 on device 1: An exception occurred on the device while executing a kernel. The context cannot be used anymore and must be destroyed. All existing device memory allocations from this context are invalid and must be reconstructed.
-> failed to deallocate device memory
CUDA error 719 on device 1: An exception occurred on the device while executing a kernel. The context cannot be used anymore and must be destroyed. All existing device memory allocations from this context are invalid and must be reconstructed.
-> could not get memory info
Cuda error 719 on one gtx 670
Moderators: ChrisHekman, aoktar
ok, check you free memory of gpus on Live Viewer. What's seeing when you activated 670? If you see low numbers, so that maybe reason for error.
Octane For Cinema 4D developer / 3d generalist
3930k / 16gb / 780ti + 1070/1080 / psu 1600w / numerous hw
3930k / 16gb / 780ti + 1070/1080 / psu 1600w / numerous hw
- HandsomeRibb
- Posts: 6
- Joined: Fri Aug 29, 2014 1:50 pm
I don't see any low numbers. The numbers look about the same as my other gtx 670 card, and they have the same amount of Vmemory so that should'nt be a problem.
I'm afraid there is something wrong with the installation of the card, because I get wierd flickering inside photoshop now as well. I have put the card in myself, and I'm a newbie at these kind of things. The faulty gtx 670 had two 6-pin ports in to the card, so I plugged it in to two SATA-ports in my PSU. My other two cards have one 6+2 connector, do you think this card needs that as well? Could this be a power-supply thing?
Thanks,
Roger
I'm afraid there is something wrong with the installation of the card, because I get wierd flickering inside photoshop now as well. I have put the card in myself, and I'm a newbie at these kind of things. The faulty gtx 670 had two 6-pin ports in to the card, so I plugged it in to two SATA-ports in my PSU. My other two cards have one 6+2 connector, do you think this card needs that as well? Could this be a power-supply thing?
Thanks,
Roger
- HandsomeRibb
- Posts: 6
- Joined: Fri Aug 29, 2014 1:50 pm
I have a 1200 watt PSU, so I guess that should be enough for everything.
I realized that the card was put into a PCI 4.0 port on my motherboard, so I tried to switch it around and removed the gtx 680 and put the faulty card into a PCI 16/8 but I still got this cuda error.
But it's a little bit wierd. I tried another rather heavy scene without a problem...until I downloaded the "fake glas" shader, then I got the same cuda error again on that card.
It's like the card works most of the time, but sometimes it doesn't really want to play.
Strange.
I realized that the card was put into a PCI 4.0 port on my motherboard, so I tried to switch it around and removed the gtx 680 and put the faulty card into a PCI 16/8 but I still got this cuda error.
But it's a little bit wierd. I tried another rather heavy scene without a problem...until I downloaded the "fake glas" shader, then I got the same cuda error again on that card.
It's like the card works most of the time, but sometimes it doesn't really want to play.
Strange.
sounds like a faulty card, try to run benchmark test, probably it can't compute somethings which is needed for octane
try different cuda and 3D benchmarks
I had similar card which wasn't able to work with Dx9 but Dx11 and 10 was ok and it was working slow on rendering
try different cuda and 3D benchmarks
I had similar card which wasn't able to work with Dx9 but Dx11 and 10 was ok and it was working slow on rendering
- HandsomeRibb
- Posts: 6
- Joined: Fri Aug 29, 2014 1:50 pm
Thanks.
There seems to be alot of different benchmarking software out there. What would you recommend using?
Thanks again.
There seems to be alot of different benchmarking software out there. What would you recommend using?
Thanks again.
3D mark would be enough
and if no problem then run arionbench and Cycles in blender
and if no problem then run arionbench and Cycles in blender