SMP client not working properly

Soldato
Joined
3 Jan 2006
Posts
5,009
Hi there:

So I've recently started to use the multi-core client for the quad core but it always stop working after some time, I know because it stucks at a certain percentage and not utilising any CPU. I ran 4 single clients before and they worked flawlessly so don't know what's stopping the SMP?

Any idea appreciated :)
 
OK this is the log from last night:


--- Opening Log file [February 26 00:29:01]


# SMP Client ##################################################################
###############################################################################

Folding@Home Client Version 5.91beta6

http://folding.stanford.edu

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

Launch directory: C:\Program Files\Folding@Home Windows SMP Client V1.01
Executable: C:\Program Files\Folding@Home Windows SMP Client V1.01\fah.exe


[00:29:01] - Ask before connecting: No
[00:29:01] - User name: steve258 (Team 10)
[00:29:01] - User ID: 217C6DD05BF947E5
[00:29:01] - Machine ID: 1
[00:29:01]
[00:29:01] Loaded queue successfully.
[00:29:01]
[00:29:01] + Processing work unit
[00:29:01] Core required: FahCore_a1.exe
[00:29:01] Core found.
[00:29:01] Working on Unit 01 [February 26 00:29:01]
[00:29:01] + Working ...
[00:29:01]
[00:29:01] *------------------------------*
[00:29:01] Folding@Home Gromacs SMP Core
[00:29:01] Version 1.74 (March 10, 2007)
[00:29:01]
[00:29:01] Preparing to commence simulation
[00:29:01] - Ensuring status. Please wait.
[00:29:18] - Looking at optimizations...
[00:29:18] - Working with standard loops on this execution.
[00:29:18] Examination of work files indicates 8 consecutive improper terminations of core.
[00:29:22] - Expanded 2948770 -> 15225228 (decompressed 516.3 percent)
[00:29:22]
[00:29:22] Project: 2653 (Run 2, Clone 158, Gen 65)
[00:29:22]
[00:29:23] Entering M.D.
[00:29:29] Calling FAH init
[00:29:30] in POPC
[00:29:30] Writing local files
[00:29:30] Completed 350000 out of 500000 steps (70 percent)
[00:29:30] PC
[00:29:30] Writing local files
[00:29:30] Completed 350000 out of 500000 steps (70 percent)
[00:29:31] Extra SSE boost OK.
[00:38:04] Writing local files
[00:38:04] Completed 355000 out of 500000 steps (71 percent)
[00:46:37] Writing local files
[00:46:38] Completed 360000 out of 500000 steps (72 percent)
[00:55:10] Writing local files
[00:55:10] Completed 365000 out of 500000 steps (73 percent)
[01:03:43] Writing local files
[01:03:43] Completed 370000 out of 500000 steps (74 percent)
[01:12:16] Writing local files
[01:12:17] Completed 375000 out of 500000 steps (75 percent)
[01:20:49] Writing local files
[01:20:50] Completed 380000 out of 500000 steps (76 percent)
[01:29:22] Writing local files
[01:29:22] Completed 385000 out of 500000 steps (77 percent)
[01:37:54] Writing local files
[01:37:54] Completed 390000 out of 500000 steps (78 percent)
[01:46:26] Writing local files
[01:46:26] Completed 395000 out of 500000 steps (79 percent)
[01:54:59] Writing local files
[01:54:59] Completed 400000 out of 500000 steps (80 percent)
[02:03:31] Writing local files
[02:03:31] Completed 405000 out of 500000 steps (81 percent)
[02:12:03] Writing local files
[02:12:03] Completed 410000 out of 500000 steps (82 percent)
[02:20:34] Writing local files
[02:20:34] Completed 415000 out of 500000 steps (83 percent)
[02:29:05] Writing local files
[02:29:05] Completed 420000 out of 500000 steps (84 percent)
[02:37:36] Writing local files
[02:37:36] Completed 425000 out of 500000 steps (85 percent)
[02:45:57] Writing local files
[02:45:57] Completed 430000 out of 500000 steps (86 percent)

Folding@Home Client Shutdown at user request.

Folding@Home Client Shutdown.

----------------------------------------------------------------------------------

I shut it down this morning at around 10am.
 
If you restart it does it then continue and complete the WU? Seems to have restarted again after being previously shutdown so I guess that's the case.

I would stop the client, delete FahCore_a1.exe from the folding directory, then restart it to download a fresh core. Might not do the trick but can't hurt at least. If you can hobble to the end of this WU it might be best to do it then just incase the client throws a fit :o


edit: Usually when the client stops responding it is due to a change to the network it is connected through (Win SMP uses the localhost address on your PC to bounce work back to itself - that's how multiple cores work together under windows) so if for instance you have slightly flaky wireless that could easily be causing it, or if you restart a wired router/unplug a wired connection that is enough to stall the client.

Hope this helps :)
 
Last edited:
Hi there:

Yes if I shut it down then restart it it will continue to finish current work unit. The current WU has just finished as I type this so will try deleting FahCore_a1 like you suggested.

You're right I am on wireless connection and it's not the best as I have cheap belkin router and ASUS PCI wireless card. I've fixed my IP address of my wireless card and put it on a DMZ if that makes any difference?

Does that mean there's not a fix and I'll just have to return to running 4 instances manually?
 
Most likely it's the wireless dropping out then, it only takes a moment to upset the client and you will see the exact behaviour you have there.

If you have a separate antenna try moving it around the place to see if you get more reliable reception, you could even try changing the wireless channel the router and the card use in case you are getting interference from other networks or other devices using 2.4GHz radio (these can include wireless keyboard/mouse, bluetooth devices, cordless house phones) - by default they'll probably be set to channel 6 so try something either end of the range.

My brother has an Asus on-board WiFi with a Belkin router and his used to be very flaky until we changed the channel off the channel all his neighbours seemed to be using and moved the antenna out from under the desk to a position behind his monitor.

Otherwise if the router isn't too far away get laying cable :p
 
Thanks again :)

Have changed the channel from 1 to 13 and it instantly cut off my SMP client :p. Just found out it's actually a D-link router but a cheap one nonetheless. Will see how it gets on in ch13. Don't think reception is a problem atm it's between 4/5 bars and the transfer speed is full 54mb. I have the cable long enough to connect my pc to the router downstairs but don't think the landlord will appreciate the messiness :p

Thanks very much for your help.
 
i tend to get that same issue every once in a while. When my wireless appears to lose the signal then folding stops. It's just a case of closing down the core and starting it up again. Once in a while it happens but nothing i can do about it.
 
edit: Usually when the client stops responding it is due to a change to the network it is connected through (Win SMP uses the localhost address on your PC to bounce work back to itself - that's how multiple cores work together under windows) so if for instance you have slightly flaky wireless that could easily be causing it, or if you restart a wired router/unplug a wired connection that is enough to stall the client.

that makes a lot of sense thank you. it has happened to me a few time, i just restarted the machine and it carried on fine. my connection isn't brillient so it cuts out every now and then
 
that makes a lot of sense thank you. it has happened to me a few time, i just restarted the machine and it carried on fine. my connection isn't brillient so it cuts out every now and then

no worries - I always forget about wireless and WinSMP as I tried it a while ago and it was just a bit too unreliable for my liking, the only things I use wireless for now are my work laptop when I bring that home and my Wii which I'd have to spend extra on if I wanted a wired connection.

I've since got a job where wireless is a big part of the operation (warehouse management systems and the like) so it's nice to be able to feed it back to the fun bits of my life :)

I've also found that mini-trunking only costs £1.90 for a 3 metre length from my friendly local electrical supplies shop so I'll soon have my wires slightly more tidy :D
 
Back
Top Bottom