Cuda_fermi SETI@home Enhanced v6.10 work units all fail with computing error.

Message boards : Number crunching : Cuda_fermi SETI@home Enhanced v6.10 work units all fail with computing error.
Message board moderation

To post messages, you must log in.

AuthorMessage
Bill Butler
Avatar

Send message
Joined: 26 Aug 03
Posts: 101
Credit: 4,270,697
RAC: 0
United States
Message 1365930 - Posted: 9 May 2013, 3:28:08 UTC

SETI@home and Astropulse calculations using GPU seem to work fine on my machine with 1 exception. Cuda_fermi SETI@home Enhanced v6.10 work units all fail with computing error.

Such tasks start, run for 5 – 15 Sec and then fail.

Here is an example from my SETI log:

Task: 2984544487
Work Unit: 1239482646 (16ap11ae.32530.201978.13.11.18)
Computer: 6844172 (mine)
Sent: 7 May 2013, 23:40:26 UTC
Time reported: 8 May 2013, 0:22:40 UTC

Status: Error while computing
Run time (sec): 12.25
CPU time (sec): 11.92
Application: SETI@home Enhanced v6.10 (cuda_fermi)

Following reference shows that cuda fermi supports my NVIDIA GeForce GT 640M (2048 MB) as compute capability version 3.0.

http://en.wikipedia.org/wiki/CUDA

The only peculiarity I notice is that this GPU is supposed to be a mobile package (per the reference). Instead, it is installed in the Dell XPS desktop. But that doesn’t seem to be the showstopper.

So, I am not sure if the SETI application software is the problem or my PC.

I kinda’ suspect the application software is the problem because other GPU applications work awesomely. For example, AstroPulse v6 v6.04 (cuda_opencl_100) just burns through a work unit.

I hate to see cuda fermi work units downloaded, fail, and get promptly returned with an error message.
If I can’t run with the big dogs, I have to stay on the porch.

Anyone have a suggestion to try?????

Thanx!!!!

"It is often darkest just before it turns completely black."
ID: 1365930 · Report as offensive
Profile Gatekeeper
Avatar

Send message
Joined: 14 Jul 04
Posts: 887
Credit: 176,479,616
RAC: 0
United States
Message 1365934 - Posted: 9 May 2013, 3:40:39 UTC - in response to Message 1365930.  

ID: 1365934 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 34744
Credit: 261,360,520
RAC: 489
Australia
Message 1365938 - Posted: 9 May 2013, 4:03:26 UTC - in response to Message 1365934.  

See this post

The 2nd part, Kepler unsupported, is what you want.

Also many people have PM'd about the effects of whether doing this will affect anything else and for me it has had no ill effects on any other program that I use, and I use a lot more than most, but the only effect that I've found is all for the best with Kepler based cards.

Nor have I heard of any ill effects.

Cheers.
ID: 1365938 · Report as offensive
Darth Beaver Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Avatar

Send message
Joined: 20 Aug 99
Posts: 6728
Credit: 21,443,075
RAC: 3
Australia
Message 1365944 - Posted: 9 May 2013, 4:21:29 UTC - in response to Message 1365938.  

See this post

The 2nd part, Kepler unsupported, is what you want.

Also many people have PM'd about the effects of whether doing this will affect anything else and for me it has had no ill effects on any other program that I use, and I use a lot more than most, but the only effect that I've found is all for the best with Kepler based cards.

Nor have I heard of any ill effects.

Cheers.


yep here to all good
ID: 1365944 · Report as offensive
Bill Butler
Avatar

Send message
Joined: 26 Aug 03
Posts: 101
Credit: 4,270,697
RAC: 0
United States
Message 1366181 - Posted: 10 May 2013, 2:02:50 UTC

Yup,OK.

I got CUDA_GRID_SIZE_COMPAT system variable set to 1.

Note per WIN8 sysinfo:

Adapter Type GeForce GT 640M, NVIDIA compatible
Driver Version: 9.18.13.721

Installed drivers are:
nvd3dumx.dll,
nvwgf2umx.dll,
nvwgf2umx.dll,
nvd3dum,
nvwgf2um,
nvwgf2um

IRQ Channel: IRQ 16

There are a bunch of cuda_fermi's in the queue. So, will start computing and should see what happens soon.

Thank you for the response with the relevant post.
"It is often darkest just before it turns completely black."
ID: 1366181 · Report as offensive
Bill Butler
Avatar

Send message
Joined: 26 Aug 03
Posts: 101
Credit: 4,270,697
RAC: 0
United States
Message 1366830 - Posted: 11 May 2013, 18:10:28 UTC

YESH!! -- I currently have 3 SETI@home Enhanced v6.10 (cuda_fermi) 's Completed & Awaiting Validation.

According to the given posts, there are no known unintended consequences.

So, all I have to do is just wait for another AstroPulse v6 v6.04 (opencl_nvidia_100) to be downloaded and verify that it completes OK, as usual.

Thanx!!

Carry On.

"It is often darkest just before it turns completely black."
ID: 1366830 · Report as offensive
Bill Butler
Avatar

Send message
Joined: 26 Aug 03
Posts: 101
Credit: 4,270,697
RAC: 0
United States
Message 1366889 - Posted: 11 May 2013, 22:17:51 UTC

...... 10 SETI@home Enhanced v6.10 (cuda_fermi) 's Completed & Validated.
"It is often darkest just before it turns completely black."
ID: 1366889 · Report as offensive
Bill Butler
Avatar

Send message
Joined: 26 Aug 03
Posts: 101
Credit: 4,270,697
RAC: 0
United States
Message 1368280 - Posted: 15 May 2013, 20:37:50 UTC

,,,,, AND Astropulse is processing OK too. All GPU tasks seem to be working OK. That is, I also can confirm no side effects from setting the System Environmental variable.
"It is often darkest just before it turns completely black."
ID: 1368280 · Report as offensive

Message boards : Number crunching : Cuda_fermi SETI@home Enhanced v6.10 work units all fail with computing error.


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.