Issues folding with nVidia gpu

It seems that a lot of GPU problems revolve around specific versions of drivers. Though NVidia has their own support structure, you can often learn from information reported by others who fold.

Moderators: Site Moderators, FAHC Science Team

Post Reply
DoctorBob
Posts: 10
Joined: Sun Sep 12, 2021 9:28 pm

Issues folding with nVidia gpu

Post by DoctorBob »

I have installed FAHClient (version 7.6.21) on a headless laptop running Ubuntu Server.
I access the client from my Windows PC via FAHContol (version 7.6.21).
The Intel Duo Core CPUs are folding just fine.

xxxxxx@folding-laptop:~$ top
%Cpu(s): 0.0 us, 0.7 sy, 45.9 ni, 53.2 id, 0.1 wa, 0.0 hi, 0.0 si, 0.0 st
MiB Mem : 3924.2 total, 2652.7 free, 383.9 used, 887.6 buff/cache
MiB Swap: 3924.0 total, 3924.0 free, 0.0 used. 3300.4 avail Mem

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
1007 root 39 19 235772 64860 12788 S 48.7 1.6 480:02.54 FahCore_a8
1030 root 39 19 235972 64508 12584 S 43.9 1.6 450:06.76 FahCore_a8

However, when it comes to getting the gpu to fold, I am stuck.

The laptop is an oldie but comes with a dedicated nvidia gpu.
Here is some hardware info:

xxxxxx@folding-laptop:~$ sudo lshw
folding-laptop
description: Portable Computer
product: Inspiron 1520
serial: 11RMQ1S
width: 64 bits
capabilities: smbios-2.4 dmi-2.4 smp vsyscall32
configuration: boot=normal chassis=portable uuid=4c4c4544-0031-5210-804d-b1c04f513153

xxxxxx@folding-laptop:~$ lspci | grep -i nvidia
01:00.0 VGA compatible controller: NVIDIA Corporation G84M [GeForce 8600M GT] (rev a1)

xxxxxx@folding-laptop:~$ nvidia-smi
Fri Aug 5 08:16:21 2022
+------------------------------------------------------+
| NVIDIA-SMI 340.108 Driver Version: 340.108 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce 8600M GT Off | 0000:01:00.0 N/A | N/A |
|100% 52C P0 N/A / N/A | 3MiB / 255MiB | N/A Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Compute processes: GPU Memory |
| GPU PID Process name Usage |
|=============================================================================|
| 0 Not Supported |
+-----------------------------------------------------------------------------+


I am running Ubuntu Server 22.04 on the laptop, access it via ssh:

xxxxxx@folding-laptop:~$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 22.04.1 LTS
Release: 22.04
Codename: jammy


I installed the nvidia drivers for the GeForce 8600M GT device following instructions from this url:
https://pupli.net/2021/12/install-nvidi ... on-ubuntu/

I also installed OpenCL dev files that is mentioned in this url:
viewtopic.php?p=356818

xxxxxx@folding-laptop:~$ sudo apt list | grep ocl-icd-opencl-dev
ocl-icd-opencl-dev/jammy,now 2.2.14-3 amd64 [installed]


I added gpu to FAHControl --> Configure --> Slot
and
FAHControl --> Configure --> Expert --> Extra client options (gpu true)
The Expert entry disappears from the list after a while.

What I see in the FAHContol Folding Slots is:
ID Status Description
00 Running cpu:0
01 Running cpu:0
03 Disabled gpu:-1:-1

Here is the FAHClient config.xml file:
xxxxxx@folding-laptop:~$ cat /etc/fahclient/config.xml
<config>
<!-- Client Control -->
<fold-anon v='true'/>

<!-- HTTP Server -->
<allow v='127.0.0.1 192.168.0.0/24'/>

<!-- Network -->
<proxy v=':8080'/>

<!-- Remote Command Server -->
<command-allow-no-pass v='127.0.0.1 192.168.0.0/24'/>

<!-- Slot Control -->
<power v='full'/>

<!-- User Information -->
<passkey v='********************************'/>
<user v='DoctorBob'/>

<!-- Folding Slots -->
<slot id='0' type='CPU'>
<cpu-usage v='48'/>
</slot>
<slot id='1' type='CPU'>
<cpu-usage v='45'/>
</slot>
<slot id='3' type='GPU'/>


So I am at a loss as to why the gpu is not folding.
Do I need CUDA installed and if so where do I get it for Driver Version: 340.108?
Or do I need to install a newer driver version and if so what is the latest driver version supported by a GeForce 8600M GT video device?

Any help would be most appreciated.
JimboPalmer
Posts: 2473
Joined: Mon Feb 16, 2009 4:12 am
Location: Greenwood MS USA

Re: Issues folding with nVidia gpu

Post by JimboPalmer »

https://www.techpowerup.com/gpu-specs/g ... 00-gt.c198

It won't fold. The 8600 GT is neither OpenCL 1.2 compliant nor is can it perform 64 bit floating point math. Both are prerequisites for folding. The GPU is 15 years out of date, think of dog years.

Sorry.
Tsar of all the Rushers
I tried to remain childlike, all I achieved was childish.
A friend to those who want no friends
DoctorBob
Posts: 10
Joined: Sun Sep 12, 2021 9:28 pm

Re: Issues folding with nVidia gpu

Post by DoctorBob »

Hello Jimbo,

Thank you for such a quick reply.
Very good information. You saved me from a lot of hair pulling!
Yes, old but still going, like me!


Cheers,

Robert
JimboPalmer
Posts: 2473
Joined: Mon Feb 16, 2009 4:12 am
Location: Greenwood MS USA

Re: Issues folding with nVidia gpu

Post by JimboPalmer »

Your GPU uses an architecture called Tesla.

https://en.m.wikipedia.org/wiki/Tesla_( ... hitecture)

The much later Maxwell architecture is the oldest truly supported Nvidia GPUs, Kepler kinda works.

https://en.m.wikipedia.org/wiki/Maxwell ... hitecture)
Tsar of all the Rushers
I tried to remain childlike, all I achieved was childish.
A friend to those who want no friends
Joe_H
Site Admin
Posts: 7339
Joined: Tue Apr 21, 2009 4:41 pm
Hardware configuration: Mac Pro 2.8 quad 12 GB smp4
MacBook Pro 2.9 i7 8 GB smp3
Location: W. MA

Re: Issues folding with nVidia gpu

Post by Joe_H »

Kepler mostly works for now, but Nvidia has ended support and driver upgrades for that architecture. So anything that requires features in the latest drivers will not work well.

The architecture between Tesla and Kepler was Fermi. GPUs based on that "worked" up until about 1 1/2 years ago. They were a mess support-wise. Nvidia had added 64-bit operations to Fermi, but only partially supported them. 64-bit was supposed to require OpenCL 1.2, Nvidia implemented it by adding support for some OpenCL commands from 1.2 to their OpenCL 1.1 drivers. If certain commands were not implemented, WU processing failed. But not all projects would trigger calls to unimplemented instructions. So there were intermittent errors with Tesla cards that were finally tracked back to that partial support.

The GPU folding core has the option of running OpenCL or CUDA commands to a Nvidia GPU. CUDA requires a minimum version level, Kepler no longer meets that and should fall back to OpenCL processing.
Image

iMac 2.8 i7 12 GB smp8, Mac Pro 2.8 quad 12 GB smp6
MacBook Pro 2.9 i7 8 GB smp3
Post Reply