Re: CUDA error cudaMalloc during running namd

From: Yoshio Alarcon Lopez (yoshalar_at_gmail.com)
Date: Sun Aug 28 2022 - 13:01:36 CDT

Dear Shivam.
Thank you for your email.
If you are using the cuda toolkit from the ubuntu repository, sometimes it
does not work fine.
I kindly suggest to install cuda toolkit available in the page of Nvidia.
https://urldefense.com/v3/__https://developer.nvidia.com/cuda-zone__;!!DZ3fjg!4eA30L13pTA2zoyUwzWOhLjgdPtnQFqVOJh96sjU0Bt6ZF_UK5vQEXwbQPsnULusTov42bf5P2mHWXFVsw$
You need to select which version of CUDA is compatible with your graphic
card, and I suggest you to install it via the .run file
It seems that your Quadro K620 has a cuda compute capability of 5.0, I
think installing CUDA via the runfile will solve your issue.
Before installing cuda, disabling nouveau is mandatory.
Please read the documentation:
https://urldefense.com/v3/__https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html__;!!DZ3fjg!4eA30L13pTA2zoyUwzWOhLjgdPtnQFqVOJh96sjU0Bt6ZF_UK5vQEXwbQPsnULusTov42bf5P2lbZL_2zQ$

Have a nice day, let me know if this works for you.
Best regards.

El jue, 25 ago 2022 a las 7:34, SHIVAM TIWARI (<t.shivam_at_iitg.ac.in>)
escribió:

> Dear all,
>
> I am using a workstation with ubuntu 20.04 and with two k620 Quadro Nvidia
> graphic cards installed on it, and I had installed driver for the cards
> previously from the Ubuntu repository. Recently I installed the Cuda tool
> kit (available on the ubuntu repository). And I downloaded Cuda enabled
> namd 2.14 binary (Linux-x86_64-multicore-CUDA
> <https://www.ks.uiuc.edu/Development/Download/download.cgi?UserID=&AccessCode=&ArchiveID=1625>
> (NVIDIA CUDA acceleration). But when I am trying to run the simulation, I
> am getting the following error:
>
> FATAL ERROR: CUDA error cudaMalloc(pp, sizeofT*(*curlen)) in file
> src/CudaUtils.C, function reallocate_device_T, line 130
> on Pe 0 (param-HP-Z440-Workstation device 0 pci 0:2:0): out of memory
> [Partition 0][Node 0] End of program
>
> I am using the following command to run the simulation.
>
> namd2 +idlepoll benchmark_gpu.conf > benchmark_gpu.log
>
> Please help me to sort out this error.
>
> thanks in advance
>
> regards
> shivam
>
>

-- 
*M. en C. Q. Aldo Yoshio Alarcon López*
*Laboratorio de Química Medicinal*
*FESC-UNAM*

This archive was generated by hypermail 2.1.6 : Tue Dec 13 2022 - 14:32:44 CST