CUDA error GTX680

chiendarret · 12-12-2023, 02:12 AM

Is any deficiency detectable, from below outputs, of my GTX680 graphic card on debian amd64, main board GA-X79-UD3? To justify the error shown below when trying to run a simulation with namd3 (https://www.ks.uiuc.edu/Research/namd/alpha/3.0alpha/)

Quote:

root@gig64:/home/francesco# lspci | grep -i --color 'vga\|3d\|2d'
02:00.0 VGA compatible controller: NVIDIA Corporation GK104 [GeForce GTX 680] (rev a1)
root@gig64:/home/francesco# lspci -v -s 02:0.0
02:00.0 VGA compatible controller: NVIDIA Corporation GK104 [GeForce GTX 680] (rev a1) (prog-if 00 [VGA controller])
Subsystem: Micro-Star International Co., Ltd. [MSI] GK104 [GeForce GTX 680]
Flags: bus master, fast devsel, latency 0, IRQ 50
Memory at fa000000 (32-bit, non-prefetchable) [size=16M]
Memory at f0000000 (64-bit, prefetchable) [size=128M]
Memory at f8000000 (64-bit, prefetchable) [size=32M]
I/O ports at e000 [size=128]
Expansion ROM at 000c0000 [virtual] [disabled] [size=128K]
Capabilities: [60] Power Management version 3
Capabilities: [68] MSI: Enable+ Count=1/1 Maskable- 64bit+
Capabilities: [78] Express Endpoint, MSI 00
Capabilities: [b4] Vendor Specific Information: Len=14 <?>
Capabilities: [100] Virtual Channel
Capabilities: [128] Power Budgeting <?>
Capabilities: [600] Vendor Specific Information: ID=0001 Rev=1 Len=024 <?>
Kernel driver in use: nvidia
Kernel modules: nvidia

Quote:

NVIDIA-SMI 470.199.02 Driver Version: 470.199.02 CUDA Version: 11.4

Quote:

namd3 heat-01.conf +p6 +idlepoll 2>&1 | tee trial_12Dec2023_heat-01.log

FATAL ERROR: CUDA error cufftPlan3d(&forwardPlan, n[2], n[1], n[0], CUFFT_R2C) in file src/CudaPmeSolverUtil.C, function plan3D
on Pe 0 (gig64 device 0 pci 0:2:0)

Further question: is any free software to check independently the performance of this graphic card?

Thanks for advice

obobskivich · 12-18-2023, 12:10 PM

Wild shot in the dark: does that application require a newer version of CUDA than the GTX 680 (which is quite old, and in your case also running a relatively quite old driver) supports? There are different features added in newer versions of CUDA so if it needs something that the GTX 680 cannot do, it may not handle that gracefully. I poked around on the linked site for a minute or two and couldn't find clearly outlined system requirements, but NVIDIA has a page about NAMD which lists Pascal (GTX 1000 series) as the minimum requirement for an x86_64 instance (see here: https://catalog.ngc.nvidia.com/orgs/hpc/containers/namd).