Continual Guarded Runs

sir-les-mp

sir-les-mp

sir-les-mp

Associate
Joined
23 Aug 2007
Posts
1,699
Location
Rothesay
im Getting error after error with 2 of my gpu clients after installing my GTX280 which arrived this morning.
The GTX280 is working away fine but my 8800GT and 8800GTS refuse to Fold.
Ive posted the log from one of the clients below.
I have even tried the -forcegpu nvidia_g92 flag including on the GTX200 but to no avail
--- Opening Log file [March 16 10:05:41 UTC]


# Windows GPU Console Edition #################################################
###############################################################################

Folding@Home Client Version 6.23

http://folding.stanford.edu

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

Launch directory: C:\Users\Sir-Les-MP\Documents\F@H\F@HGPU2
Executable: C:\Users\Sir-Les-MP\Documents\F@H\F@HGPU2\[email protected]
Arguments: -verbosity 9 -gpu 1 -forcegpu nvidia_g80

[10:05:41] - Ask before connecting: No
[10:05:41] - User name: Sir-Les-MP (Team 10)
[10:05:41] - User ID: A4DE5AD7FDF3B3E
[10:05:41] - Machine ID: 2
[10:05:41]
[10:05:41] Loaded queue successfully.
[10:05:41]
[10:05:41] + Processing work unit
[10:05:41] Core required: FahCore_11.exe
[10:05:41] Core found.
[10:05:41] - Autosending finished units... [March 16 10:05:41 UTC]
[10:05:41] Working on queue slot 05 [March 16 10:05:41 UTC]
[10:05:41] Trying to send all finished work units
[10:05:41] + Working ...
[10:05:41] - Calling '.\FahCore_11.exe -dir work/ -suffix 05 -checkpoint 15 -verbose -lifeline 4204 -version 623'

[10:05:41] + No unsent completed units remaining.
[10:05:41] - Autosend completed
[10:05:41]
[10:05:41] *------------------------------*
[10:05:41] Folding@Home GPU Core - Beta
[10:05:41] Version 1.19 (Mon Nov 3 09:34:13 PST 2008)
[10:05:41]
[10:05:41] Compiler : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86
[10:05:41] Build host: amoeba
[10:05:41] Board Type: Nvidia
[10:05:41] Core :
[10:05:41] Preparing to commence simulation
[10:05:41] - Looking at optimizations...
[10:05:42] - Created dyn
[10:05:42] - Files status OK
[10:05:42] Error: Missing work file=<>
[10:05:42]
[10:05:42] Folding@home Core Shutdown: MISSING_WORK_FILES
[10:05:45] CoreStatus = 74 (116)
[10:05:45] The core could not find the work files specified. Removing from queue
[10:05:45] Deleting current work unit & continuing...
[10:05:49] Trying to send all finished work units
[10:05:49] + No unsent completed units remaining.
[10:05:49] - Preparing to get new work unit...
[10:05:49] + Attempting to get work packet
[10:05:49] - Will indicate memory of 4095 MB
[10:05:49] - Detect CPU. Vendor: AuthenticAMD, Family: 15, Model: 4, Stepping: 2
[10:05:49] - Connecting to assignment server
[10:05:49] Connecting to http://assign-GPU.stanford.edu:8080/
[10:05:50] Posted data.
[10:05:50] Initial: 43AB; - Successful: assigned to (171.67.108.11).
[10:05:50] + News From Folding@Home: Welcome to Folding@Home
[10:05:51] Loaded queue successfully.
[10:05:51] Connecting to http://171.67.108.11:8080/
[10:05:51] Posted data.
[10:05:51] Initial: 0000; - Receiving payload (expected size: 47210)
[10:05:52] - Downloaded at ~46 kB/s
[10:05:52] - Averaged speed for that direction ~45 kB/s
[10:05:52] + Received work.
[10:05:52] + Closed connections
[10:05:57]
[10:05:57] + Processing work unit
[10:05:57] Core required: FahCore_11.exe
[10:05:57] Core found.
[10:05:57] Working on queue slot 06 [March 16 10:05:57 UTC]
[10:05:57] + Working ...
[10:05:57] - Calling '.\FahCore_11.exe -dir work/ -suffix 06 -checkpoint 15 -verbose -lifeline 4204 -version 623'

[10:05:58]
[10:05:58] *------------------------------*
[10:05:58] Folding@Home GPU Core - Beta
[10:05:58] Version 1.19 (Mon Nov 3 09:34:13 PST 2008)
[10:05:58]
[10:05:58] Compiler : Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 14.00.50727.762 for 80x86
[10:05:58] Build host: amoeba
[10:05:58] Board Type: Nvidia
[10:05:58] Core :
[10:05:58] Preparing to commence simulation
[10:05:58] - Looking at optimizations...
[10:05:58] - Created dyn
[10:05:58] - Files status OK
[10:05:58] - Expanded 46698 -> 252912 (decompressed 541.5 percent)
[10:05:58] Called DecompressByteArray: compressed_data_size=46698 data_size=252912, decompressed_data_size=252912 diff=0
[10:05:58] - Digital signature verified
[10:05:58]
[10:05:58] Project: 5766 (Run 11, Clone 256, Gen 1897)
[10:05:58]
[10:05:58] Assembly optimizations on if available.
[10:05:58] Entering M.D.
[10:06:05] Working on Protein
[10:06:07] Run: exception thrown during GuardedRun
[10:06:07] Run: exception thrown in GuardedRun -- Gromacs cannot continue further.
[10:06:07] Going to send back what have done -- stepsTotalG=0
[10:06:07] Work fraction=0.0000 steps=0.
[10:06:11] logfile size=0 infoLength=0 edr=0 trr=23
[10:06:11] - Writing 635 bytes of core data to disk...
[10:06:11] Done: 123 -> 124 (compressed to 100.8 percent)
[10:06:11] ... Done.
[10:06:12]
[10:06:12] Folding@home Core Shutdown: UNSTABLE_MACHINE
[10:06:16] CoreStatus = 7A (122)
[10:06:16] Sending work to server
[10:06:16] Project: 5766 (Run 11, Clone 256, Gen 1897)
[10:06:16] - Read packet limit of 540015616... Set to 524286976.


[10:06:16] + Attempting to send results [March 16 10:06:16 UTC]
[10:06:16] - Reading file work/wuresults_06.dat from core
[10:06:16] (Read 636 bytes from disk)
[10:06:16] Connecting to http://171.67.108.11:8080/
[10:06:16] ***** Got a SIGTERM signal (2)
[10:06:16] Killing all core threads

Folding@Home Client Shutdown.
Click to expand...
 
1. On each vga device in device manager, right click and reinstall the drivers.
2. Check the permissions on the folders.
3. I am not 100% but there may be some issues still with disparate vga cores on the one machine still.
4. Try running memtestg80
 
Last edited:
I get these on and off on my GTX295. I'll get a few in quick succession (sometimes 3 in one day) and then everything is fine for a few days and then it happens again. Nothing I've tried seems to make any difference. I've re-installed drivers, tried different drivers and tried reducing the overclock on the card (even tried underclocking it) - to no avail. I've searched the internet for solutions but haven't found anything of any use at all.

Weird thing is, when it happens on my GPU1 client, it just dumps the WU and starts again just fine. If it happens on my GPU2 client, it dumps the WU but when it starts again, the ppd drops to less than half what it should be and restarting the client doesn't help - I have to reboot the system to get it crunching properly :confused:

If anyone can find a solution to this, I'd be delighted to try it.
 
Anther thing is PSU load, a 295 can draw up to 289w, not sure what a GTX280 would add to the system but what sort of PSU sir-les?
 
I ve solved it i hope thanks to help from the foldong forums run at stanton
I had to remove the old cores to force download of new versions and now all clients are running.
Although the GTX 280 appears to be slower at crunching compared to my 8800GT
GTX running at 640 core, 1700 Shaders and 1107 memory, GT running 600 core, 1750 Shaders and 900 memory
 
On units originally processed on the PS3 and now being checked using the GPU client not many shader processors are used (less than 96) so the main thing that affects the ppd for these WUs is shader clock. Therefore all 8800GS, 8800GT, 9800GT, 9800GTX, 9800GTX+ and GTS240 and GTS250 cards (all the cards using G80, and G92 core variants) will get a higher ppd for these WUs compared with later generation cards (GTX 260 upwards) as the shader clocks are higher on the older technology cards.

This applies to some of the 450 point WUs and a few others that I don't know the names/point value of.

phew... ;)
 
When I used to run my old 8800GTX with an 8800GT (GTX in the top slot), the GT would fold at maybe 55-60% of its potential. This was apparently because the number of shader cores didn't divide into each other evenly (128 and 112, it needed to be either 32/64/96 etc). Not sure if this still applies?
 
verbal said:
On units originally processed on the PS3 and now being checked using the GPU client not many shader processors are used (less than 96) so the main thing that affects the ppd for these WUs is shader clock. Therefore all 8800GS, 8800GT, 9800GT, 9800GTX, 9800GTX+ and GTS240 and GTS250 cards (all the cards using G80, and G92 core variants) will get a higher ppd for these WUs compared with later generation cards (GTX 260 upwards) as the shader clocks are higher on the older technology cards.

This applies to some of the 450 point WUs and a few others that I don't know the names/point value of.

phew... ;)
Click to expand...

the GTX280 has its shaders at 1700Mhz and the 8800GT has its shaders as 1750Mhz so it should not be that
and the gtx280 has 240cores against the 112 cores for the 8800gt
 
Is it possible the older 8800GTS 320 card is holding back the 8800gt and GTX280
They are all currently running p6600 WU's the 450 point ones and there ppd are as follows.
8800GT 4860ppd
8800GTS 4002ppd
GTX280 4272ppd

a quick troll on the F@H Forums i found that the 8800gt should be nearer 7200ppd for this wu and the GTX280 should be nearer 1100 ppd with the overclock i currently have on it
 
Not sure how the lower spec cards affect the others but I seem to remember reading somewhere that they can hold other cards back - maybe someone else knows better.

You're right about the GTX280 being well below par. I get 10,500 per core on my GTX295 (which is basically 2 GTX280s glued together) with a slightly lower OC.
 
I've taken the 8800GTS 320 out the system for now to see what happens
its trying to work out whether having three cards running at similar performance is going to be better than running 1 card with the approx performance of all three together.

i Know ive got another GTX 200 sereis card to arrive yet from leadtek but i dont know when thats going to be.
 
Project : 6600
Core : GPUv2 Gromacs
Frames : 100
Credit : 450


-- GTX260 --

Min. Time / Frame : 52s - 7476.92 ppd
Avg. Time / Frame : 52s - 7476.92 ppd

Your 280 is way down as others have said. If you can run two cards at 100% that more points and less watts/noise/heat.

is the 280 getting heated up but the other cards? may be slowing down to protect it's self...
 
Try putting the card with the least shader processors in the highest slot. That's what I had to do when I had 1 GS and 1 GT
 
shadowscotland said:
Project : 6600
Core : GPUv2 Gromacs
Frames : 100
Credit : 450


-- GTX260 --

Min. Time / Frame : 52s - 7476.92 ppd
Avg. Time / Frame : 52s - 7476.92 ppd

Your 280 is way down as others have said. If you can run two cards at 100% that more points and less watts/noise/heat.

is the 280 getting heated up but the other cards? may be slowing down to protect it's self...
Click to expand...

Heat wise ive yet to see it above 60c on stock fan of 40%

verbal said:
Try putting the card with the least shader processors in the highest slot. That's what I had to do when I had 1 GS and 1 GT
Click to expand...

i will give that a try see what happens
 
Ive solved the problem by the looks of things
it was the oc on the shaders that was doing it
because they were too high the card auto went to 2d clocks giving half the performance.
GTX280 now sitting at core 675, shaders 1458 and memory 1107
Now i need to re install windows again to solve the problems i caused trying to get the GTX280 running at full speed.
 
Well ive finally got a card from Leadtek only problem is i missed the delivery and wont get it till monday.

So the question is what card have they sent me instead of the GTX260 i refused.

Poll for 1 of these cards.

GTX275

GTX280

GTX285

GTX295

So whats the consensis here then considering it has teken them since the 6th Jan to replace my 9800GX2.
 
You must log in or register to reply here.
Back
Top Bottom