GPU Hangs

If you think it might be a driver problem, see viewforum.php?f=79

Moderators: slegrand, Site Moderators, PandeGroup

GPU Hangs

Postby UncleWilley » Tue May 08, 2012 10:45 pm

Trying to run GPU3 on new Dell XPS Laptop with GTX 555M GPU. I have tried multiple times and the GPU client starts but stops right away showing 3 steps completed. Is the 555M GPU supported?

Code: Select all
                       Folding@Home Client Version 6.41r2

                          http://folding.stanford.edu

###############################################################################
###############################################################################

Launch directory: C:\Users\Terry\AppData\Roaming\Folding@home-gpu


[22:34:55] - Ask before connecting: No
[22:34:55] - User name: Unclewilley (Team 80856)
[22:34:55] - User ID: 74805A640EA48969
[22:34:55] - Machine ID: 2
[22:34:55]
[22:34:55] Gpu type=3 species=21.
[22:34:55] Loaded queue successfully.
[22:34:55] Initialization complete
[22:34:55]
[22:34:55] + Processing work unit
[22:34:55] Core required: FahCore_15.exe
[22:34:55] Core found.
[22:34:55] Working on queue slot 01 [May 8 22:34:55 UTC]
[22:34:55] + Working ...
[22:34:55]
[22:34:55] *------------------------------*
[22:34:55] Folding@Home GPU Core
[22:34:55] Version                2.22 (Thu Dec 8 17:08:05 PST 2011)
[22:34:55] Build host             SimbiosNvdWin7
[22:34:55] Board Type             NVIDIA/CUDA
[22:34:55] Core                   15
[22:34:55]
[22:34:55] Window's signal control handler registered.
[22:34:55] Preparing to commence simulation
[22:34:55] - Looking at optimizations...
[22:34:55] - Files status OK
[22:34:55] sizeof(CORE_PACKET_HDR) = 512 file=<>
[22:34:56] - Expanded 550559 -> 983664 (decompressed 178.6 percent)
[22:34:56] Called DecompressByteArray: compressed_data_size=550559 data_size=983664, decompressed_data_size=983664 diff=0
[22:34:56] - Digital signature verified
[22:34:56]
[22:34:56] Project: 7644 (Run 122, Clone 0, Gen 13)
[22:34:56]
[22:34:56] Assembly optimizations on if available.
[22:34:56] Entering M.D.
[22:34:58] Tpr hash work/wudata_01.tpr:  1623228847 278976962 1326092051 493591902 889669470
[22:34:58] GPU device info: vendor=0 device=0 name=<NA> match=0
[22:34:58] Working on Protein in water
[22:34:58] Client config found, loading data.
[22:34:58] Starting GUI Server
[22:36:54] Setting checkpoint frequency: 25000
[22:36:54] Completed         3 out of 2500000 steps (0%).


Mod Edit: Changed Position Of Code Tags - PantherX
UncleWilley
 
Posts: 13
Joined: Fri Feb 20, 2009 10:08 pm

Re: GPU Hangs

Postby 7im » Tue May 08, 2012 10:49 pm

Nothing is blocked on the v6 client, so all GPUs are technically supported.

Read the v6 GPU install guide and add the force gpu command line switch for your hardware...
How to provide enough information to get helpful support
Tell me and I forget. Teach me and I remember. Involve me and I learn.
User avatar
7im
 
Posts: 14648
Joined: Thu Nov 29, 2007 4:30 pm
Location: Arizona

Re: GPU Hangs

Postby UncleWilley » Fri May 11, 2012 3:36 pm

GPU on laptop is folding BUT EXTREMELY SLOW! Current WU is P7641(R406, C0, G24), Core: OPENMMGPU, Core ID: 15; TPF 32:42

FLAGS: -forcegpu nvidia_fermi - verbosity 9

What is Core: OPENMMGPU??
Any suggestions on how to speed up the WUs??

Code: Select all
[07:33:19] + Attempting to get work packet
[07:33:19] Passkey found
[07:33:19] - Will indicate memory of 6038 MB
[07:33:19] Gpu type=3 species=20.
[07:33:19] - Detect CPU. Vendor: GenuineIntel, Family: 6, Model: 10, Stepping: 7
[07:33:19] - Connecting to assignment server
[07:33:19] Connecting to http://assign-GPU.stanford.edu:8080/
[07:33:20] Posted data.
[07:33:20] Initial: 40AB; - Successful: assigned to (171.64.65.93).
[07:33:20] + News From Folding@Home: Welcome to Folding@Home
[07:33:20] Loaded queue successfully.
[07:33:20] Gpu type=3 species=20.
[07:33:20] Sent data
[07:33:20] Connecting to http://171.64.65.93:8080/
[07:33:21] Posted data.
[07:33:21] Initial: 0000; - Receiving payload (expected size: 550402)
[07:33:23] - Downloaded at ~268 kB/s
[07:33:23] - Averaged speed for that direction ~268 kB/s
[07:33:23] + Received work.
[07:33:23] Trying to send all finished work units
[07:33:23] + No unsent completed units remaining.
[07:33:23] + Closed connections
[07:33:23]
[07:33:23] + Processing work unit
[07:33:23] Core required: FahCore_15.exe
[07:33:23] Core found.
[07:33:23] Working on queue slot 02 [May 11 07:33:23 UTC]
[07:33:23] + Working ...
[07:33:23] - Calling '.\FahCore_15.exe -dir work/ -suffix 02 -nice 19 -checkpoint 15 -verbose -lifeline 4628 -version 641'

[07:33:23]
[07:33:23] *------------------------------*
[07:33:23] Folding@Home GPU Core
[07:33:23] Version                2.22 (Thu Dec 8 17:08:05 PST 2011)
[07:33:23] Build host             SimbiosNvdWin7
[07:33:23] Board Type             NVIDIA/CUDA
[07:33:23] Core                   15
[07:33:23]
[07:33:23] Window's signal control handler registered.
[07:33:23] Preparing to commence simulation
[07:33:23] - Looking at optimizations...
[07:33:23] DeleteFrameFiles: successfully deleted file=work/wudata_02.ckp
[07:33:23] - Created dyn
[07:33:23] - Files status OK
[07:33:23] sizeof(CORE_PACKET_HDR) = 512 file=<>
[07:33:24] - Expanded 549890 -> 974664 (decompressed 177.2 percent)
[07:33:24] Called DecompressByteArray: compressed_data_size=549890 data_size=974664, decompressed_data_size=974664 diff=0
[07:33:24] - Digital signature verified
[07:33:24]
[07:33:24] Project: 7641 (Run 406, Clone 0, Gen 24)
[07:33:24]
[07:33:24] Assembly optimizations on if available.
[07:33:24] Entering M.D.
[07:33:26] Tpr hash work/wudata_02.tpr:  3520509978 770642441 1924129824 1235436324 4015850054
[07:33:26] GPU device info: vendor=0 device=0 name=<NA> match=0
[07:33:26] Working on Protein in water
[07:33:26] Client config found, loading data.
[07:33:26] Starting GUI Server
[07:35:40] Setting checkpoint frequency: 25000
[07:35:40] Completed         3 out of 2500000 steps (0%).
[08:08:21] Completed     25000 out of 2500000 steps (1%).
[08:41:04] Completed     50000 out of 2500000 steps (2%).
[09:13:47] Completed     75000 out of 2500000 steps (3%).
[09:46:30] Completed    100000 out of 2500000 steps (4%).
[10:19:09] Completed    125000 out of 2500000 steps (5%).
[10:34:11] - Autosending finished units... [May 11 10:34:11 UTC]
[10:34:11] Trying to send all finished work units
[10:34:11] + No unsent completed units remaining.
[10:34:11] - Autosend completed
[10:34:11] + Working...
[10:51:49] Completed    150000 out of 2500000 steps (6%).
[11:24:31] Completed    175000 out of 2500000 steps (7%).
UncleWilley
 
Posts: 13
Joined: Fri Feb 20, 2009 10:08 pm

Re: GPU Hangs

Postby bruce » Fri May 11, 2012 7:41 pm

Project 7644 is one of the biggest proteins FAH has analyzed on GPUs. (It also earns more points.) At TPF of 32:42 you should be able to complete this WU in 2 days 7 hrs. That is well under the 13 day timeout or the 18 day deadline. I don't see a problem.
bruce
 
Posts: 21546
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.

Re: GPU Hangs

Postby UncleWilley » Fri May 11, 2012 10:57 pm

bruce wrote:Project 7644 is one of the biggest proteins FAH has analyzed on GPUs. (It also earns more points.) At TPF of 32:42 you should be able to complete this WU in 2 days 7 hrs. That is well under the 13 day timeout or the 18 day deadline. I don't see a problem.

Bruce

Thanks for the reply. Apparently the GPU in my new laptop is getting very different projects vs. my 2 other desktop GPUs. All 3 of my desktop GPUs right now have projects 10503 or 10502 with Core ID 11 & Core GROGPU2. These projects WUs complete each % in app. 2 minutes vs. the 32 min for the new GPU/Projects.

What does the CORE: OPENMMGPU mean vs. CORE: GROGPU2?

Also, is it correct to use the FLAG -forcegpu nvidia_fermi for my GT 555M GPU?

Thanks for the info.
UncleWilley
 
Posts: 13
Joined: Fri Feb 20, 2009 10:08 pm

Re: GPU Hangs

Postby bruce » Fri May 11, 2012 11:45 pm

Each project uses a specific FahCore which is a particular analysis method designed to work with a certain class of hardware. Each one is maintained separately, so if a bug needs to be fixed, a different revision number may need to be released for more than one FahCore, (such as when FahCores are released using the same analysis method but optimized for different hardware.

http://fahwiki.net/index.php/Cores has a nice table of all the FahCores.

Yes, the GT 555M is a Fermi-architecture GPU from nvidia which can be run on V6 with that flag.
bruce
 
Posts: 21546
Joined: Thu Nov 29, 2007 10:13 pm
Location: So. Cal.


Return to V6 GPU3 beta (including Fermi) OpenMM

Who is online

Users browsing this forum: No registered users and 1 guest

cron