• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Problem with the SMP Windows client, can anyone help?

B

Bill Clo

Gawd
Joined
Jan 24, 2002
Messages
959
Normally I'd post this on the official Folding forums, but like many folk, am unable to log in so I can post.

Just noticed that my SMP Windows client was not doing anything...first noticed because my temp program for the system (Intel TAT) showed idle temps...

Checked Task Manager, and the system was not running the usual 4 Core A1(?) processes. Yet the log didn't show anything unusual...log exerpt below:


[17:23:37] - Preparing to get new work unit...
[17:23:37] + Attempting to get work packet
[17:23:37] - Connecting to assignment server
[17:23:37] - Successful: assigned to (171.64.65.64).
[17:23:37] + News From Folding@Home: Welcome to Folding@Home
[17:23:38] Loaded queue successfully.
[17:23:52] + Closed connections
[17:23:52]
[17:23:52] + Processing work unit
[17:23:52] Core required: FahCore_a1.exe
[17:23:52] Core found.
[17:23:52] Working on Unit 06 [August 22 17:23:52]
[17:23:52] + Working ...
[17:23:52]
[17:23:52] *------------------------------*
[17:23:52] Folding@Home Gromacs SMP Core
[17:23:52] Version 1.74 (March 10, 2007)
[17:23:52]
[17:23:52] Preparing to commence simulation
[17:23:52] - Ensuring status. Please wait.
[17:24:09] - Assembly optimizations manually forced on.
[17:24:09] - Not checking prior termination.
[17:24:16] - Expanded 2955470 -> 15225228 (decompressed 515.1 percent)
[17:24:16] - Starting from initial work packet
[17:24:16]
[17:24:16] Project: 2653 (Run 2, Clone 40, Gen 11)
[17:24:16]
[17:24:19] Assembly optimizations on if available.
[17:24:19] Entering M.D.
[17:24:26] Rejecting checkpoint
[17:24:27] Protein: Protein in POPC
[17:24:27] Writing local files
[17:24:28] Extra SSE boost OK.
[17:24:29] Writing local files
[17:24:29] Completed 0 out of 500000 steps (0 percent)
[17:35:11] Writing local files
[17:35:11] Completed 5000 out of 500000 steps (1 percent)
[17:45:53] Writing local files
[17:45:54] Completed 10000 out of 500000 steps (2 percent)
[17:56:35] Writing local files
[17:56:35] Completed 15000 out of 500000 steps (3 percent)
[18:07:18] Writing local files
[18:07:18] Completed 20000 out of 500000 steps (4 percent)
[18:18:02] Writing local files
[18:18:02] Completed 25000 out of 500000 steps (5 percent)
[18:28:46] Writing local files
[18:28:46] Completed 30000 out of 500000 steps (6 percent)
[18:39:30] Writing local files
[18:39:30] Completed 35000 out of 500000 steps (7 percent)
[18:50:14] Writing local files
[18:50:14] Completed 40000 out of 500000 steps (8 percent)
[19:00:58] Writing local files
[19:00:58] Completed 45000 out of 500000 steps (9 percent)
[19:11:40] Writing local files
[19:11:41] Completed 50000 out of 500000 steps (10 percent)
[19:22:24] Writing local files
[19:22:24] Completed 55000 out of 500000 steps (11 percent)
[19:33:07] Writing local files
[19:33:07] Completed 60000 out of 500000 steps (12 percent)
[19:43:51] Writing local files
[19:43:51] Completed 65000 out of 500000 steps (13 percent)
[19:54:35] Writing local files
[19:54:35] Completed 70000 out of 500000 steps (14 percent)
[20:05:20] Writing local files
[20:05:20] Completed 75000 out of 500000 steps (15 percent)
[20:16:03] Writing local files
[20:16:03] Completed 80000 out of 500000 steps (16 percent)
[20:26:47] Writing local files
[20:26:47] Completed 85000 out of 500000 steps (17 percent)
[20:37:31] Writing local files
[20:37:31] Completed 90000 out of 500000 steps (18 percent)
[20:48:15] Writing local files
[20:48:15] Completed 95000 out of 500000 steps (19 percent)
[20:59:00] Writing local files
[20:59:00] Completed 100000 out of 500000 steps (20 percent)
[21:09:43] Writing local files
[21:09:43] Completed 105000 out of 500000 steps (21 percent)
[21:20:26] Writing local files
[21:20:26] Completed 110000 out of 500000 steps (22 percent)
[21:31:10] Writing local files
[21:31:10] Completed 115000 out of 500000 steps (23 percent)
[21:41:54] Writing local files
[21:41:54] Completed 120000 out of 500000 steps (24 percent)
[21:52:38] Writing local files
[21:52:38] Completed 125000 out of 500000 steps (25 percent)
[22:03:22] Writing local files
[22:03:22] Completed 130000 out of 500000 steps (26 percent)
[22:14:05] Writing local files
[22:14:05] Completed 135000 out of 500000 steps (27 percent)
[22:24:49] Writing local files
[22:24:49] Completed 140000 out of 500000 steps (28 percent)
[22:35:32] Writing local files
[22:35:32] Completed 145000 out of 500000 steps (29 percent)
[22:46:15] Writing local files
[22:46:16] Completed 150000 out of 500000 steps (30 percent)
[22:56:59] Writing local files
[22:56:59] Completed 155000 out of 500000 steps (31 percent)
[23:07:44] Writing local files
[23:07:44] Completed 160000 out of 500000 steps (32 percent)
[23:18:29] Writing local files
[23:18:29] Completed 165000 out of 500000 steps (33 percent)
[23:29:13] Writing local files
[23:29:13] Completed 170000 out of 500000 steps (34 percent)
[23:39:57] Writing local files
[23:39:57] Completed 175000 out of 500000 steps (35 percent)
[23:50:42] Writing local files
[23:50:42] Completed 180000 out of 500000 steps (36 percent)
[00:01:26] Writing local files
[00:01:26] Completed 185000 out of 500000 steps (37 percent)
[00:12:11] Writing local files
[00:12:11] Completed 190000 out of 500000 steps (38 percent)
[00:22:55] Writing local files
[00:22:55] Completed 195000 out of 500000 steps (39 percent)
[00:33:40] Writing local files
[00:33:40] Completed 200000 out of 500000 steps (40 percent)
[00:44:25] Writing local files
[00:44:25] Completed 205000 out of 500000 steps (41 percent)
[00:55:10] Writing local files
[00:55:10] Completed 210000 out of 500000 steps (42 percent)
[01:05:55] Writing local files
[01:05:55] Completed 215000 out of 500000 steps (43 percent)
[01:16:40] Writing local files
[01:16:40] Completed 220000 out of 500000 steps (44 percent)


That's it, just hung there for nearly an hour before I discovered it. I restarted it manually, and now the log shows:


[02:04:48] + Processing work unit
[02:04:48] Core required: FahCore_a1.exe
[02:04:48] Core found.
[02:04:48] Working on Unit 06 [August 23 02:04:48]
[02:04:48] + Working ...
[02:04:48]
[02:04:48] *------------------------------*
[02:04:48] Folding@Home Gromacs SMP Core
[02:04:48] Version 1.74 (March 10, 2007)
[02:04:48]
[02:04:48] Preparing to commence simulation
[02:04:48] - Ensuring status. Please wait.
[02:05:05] - Assembly optimizations manually forced on.
[02:05:05] - Not checking prior termination.
[02:05:13] - Expanded 2955470 -> 15225228 (decompressed 515.1 percent)
[02:05:13]
[02:05:13] Project: 2653 (Run 2, Clone 40, Gen 11)
[02:05:13]
[02:05:17] Assembly optimizations on if available.
[02:05:17] Entering M.D.
[02:05:23] Calling FAH init
[02:05:24] in POPC
[02:05:24] Writing local files
[02:05:24] checkpoint)
[02:05:24] Read checkpoint
[02:05:24] 0 steps (44 percent)
[02:05:24] Extra SSE boost OK.
[02:05:24] es
[02:05:24] Completed 220000 out of 500000 steps (44 percent)
[02:05:25] Extra SSE boost OK.
[02:16:12] Writing local files
[02:16:12] Completed 225000 out of 500000 steps (45 percent)
[02:26:57] Writing local files
[02:26:57] Completed 230000 out of 500000 steps (46 percent)

Any ideas?
 
Did you by chance lose your network connection? That will stop it dead in it's tracks till you restart the client.

Disable power management for your NIC or wireless card.
 
I don't know if this helps, but my system has issues with 2653 WU's. Mine didn't hang, it actually blue screened me at around 40% and again at 49%. After that, it finished fine and my next one is a 2651 which is running perfectly. I believe the last 2653 blue screened me also, but that was a week or two back and I didn't keep the log. I did some hardware troubleshooting and diags and stress testing and everything ran perfect after I fixed a corrupted tcip stack and winsock. Maybe its a buggy WU? Other than that, check what Killer[MoB] said - SMP client is very touchy about the network connection.
 
If the client is open but not folding then its probably a network outage.

It's why I check the FAHlog file useing FahSpy every so often.
If the time shown by one of the clients is well out in know it stopped due to this reason.
Its the only error FahSpy doesn't pick up.

Luck ............... :D
 
wolff said:
I don't know if this helps, but my system has issues with 2653 WU's. Mine didn't hang, it actually blue screened me at around 40% and again at 49%. After that, it finished fine and my next one is a 2651 which is running perfectly. I believe the last 2653 blue screened me also, but that was a week or two back and I didn't keep the log. I did some hardware troubleshooting and diags and stress testing and everything ran perfect after I fixed a corrupted tcip stack and winsock. Maybe its a buggy WU? Other than that, check what Killer[MoB] said - SMP client is very touchy about the network connection.
Click to expand...


I also had problem with stability while working on 2653. It look like its stressing system more than usual WU because as soon as I lowered my OC from 3.45 to 3.35 proble dissapeared.
 
I don't know of a network outage, but who knows? Why is the SMP client so sensitive to network problems?
 
Bill Clo said:
I don't know of a network outage, but who knows? Why is the SMP client so sensitive to network problems?
Click to expand...

I believe it uses your network connection to communicate between the threads. I'm sure somebody else can explain it better.


 
aldamon said:
I believe it uses your network connection to communicate between the threads. I'm sure somebody else can explain it better.


Click to expand...

Yes, mpiexec will spawn 4 threads (FahCore_a1.exe) and those 4 threads will pass data between them using the networking layer (127.0.0.1 which is the loopback address). The networking stack in most OS is not well written so when there is a IP change or network stall, the 4 cores close without waiting for it to get fixed. mpiexec won't pick this behavior so that's why it stops dead on his tracks. I'm sure a future version of the SMP client will fix this issue.

However, in Vista, it seems to be less sensitive to this and I successfully unplugged my main PC with Vista while the SMP client is running without stopping it. I can even mess with this connection without making the client blink... Vista's networking stack is rewritten and it seems to behave correctly.
 
Xilikon said:
However, in Vista, it seems to be less sensitive to this and I successfully unplugged my main PC with Vista while the SMP client is running without stopping it. I can even mess with this connection without making the client blink... Vista's networking stack is rewritten and it seems to behave correctly.
Click to expand...


Wow somthing in Vista works correctly. There may be hope after all :) Now if they could just make it run in less than 4GB of RAM on a dual core 2Ghz C2D or better with a 8800GTX or 3 worth a damn I would give it another shot :)
 
Tormond said:
Wow somthing in Vista works correctly. There may be hope after all :) Now if they could just make it run in less than 4GB of RAM on a dual core 2Ghz C2D or better with a 8800GTX or 3 worth a damn I would give it another shot :)
Click to expand...

Vista works perfectly with 2 Gb of ram, just check my specs in sig since this is the one running Vista. It has a excellent memory manager ;)
 
Xilikon said:
Yes, mpiexec will spawn 4 threads (FahCore_a1.exe) and those 4 threads will pass data between them using the networking layer (127.0.0.1 which is the loopback address). The networking stack in most OS is not well written so when there is a IP change or network stall, the 4 cores close without waiting for it to get fixed. mpiexec won't pick this behavior so that's why it stops dead on his tracks. I'm sure a future version of the SMP client will fix this issue.

However, in Vista, it seems to be less sensitive to this and I successfully unplugged my main PC with Vista while the SMP client is running without stopping it. I can even mess with this connection without making the client blink... Vista's networking stack is rewritten and it seems to behave correctly.
Click to expand...


Xilikon .....

I have noticed the same good thing on my quad with Vista 64 bit OS. Unplugging the power to my network connection knocks out all of my XP boxen running the SMP but the Vista boxen chugs along fine.
 
Xilikon said:
Vista works perfectly with 2 Gb of ram, just check my specs in sig since this is the one running Vista. It has a excellent memory manager ;)
Click to expand...


Does Vista work fine with 2 GB and a quad core with the SMP?

I plan to do just that... but I need some more money for another Quad.
 
You must log in or register to reply here.
Back
Top