GBGromac frame times

T

torgojr

n00b
Joined
Dec 23, 2004
Messages
27
Three of my machines have the same protein, p2082_p53_in_AGBNP, all 255 point.

XP64 - A64 2.4 2GB PC3200 - 24min
Suse 9.2 - A64 1.8 1GB PC3200 - 1hr 7min
W2000 - Cel 2.0 768MB PC2700 - 40min

I've verified all the machines are running maxed out, so WTF? How is a crummy Celeron beating a 2800+? And by that much?

 
I just looked in the log, and it is running SSE boost

[05:41:26] Writing local files
[05:41:26] GB activated
[05:41:26] Extra SSE boost OK.
[05:41:27] Writing local files
[05:41:27] Completed 0 out of 2000000 steps (0%)
 
could you post the entire log for that WU?


Keep on Folding!! For the [H]orde!!

 
Have all of them made it through enough frames that those time are correct and stabilized in whatever monitoring program you are using? The logs would help, if you could get them.

 
I would also check the times of your previous protiens on that box.
Are they running consistantly slow or not ?

Luck........ :D
 
They do seem to run knida slox on that linux box, but I have no idea why. It always runs the core ~ 98%, and I've verified the that everything is set right bios wise. Here's some logs:

A64 1.8 on Suse 9.2
[05:41:08] + Attempting to get work packet
[05:41:08] - Connecting to assignment server
[05:41:08] - Successful: assigned to (171.65.103.157).
[05:41:08] + News From Folding@Home: Welcome to Folding@Home
[05:41:08] Loaded queue successfully.
[05:41:09] + Closed connections
[05:41:09]
[05:41:09] + Processing work unit
[05:41:09] Core required: FahCore_7a.exe
[05:41:09] Core not found.
[05:41:09] - Core is not present or corrupted.
[05:41:09] - Attempting to download new core...
[05:41:09] + Downloading new core: FahCore_7a.exe
[05:41:09] + 10240 bytes downloaded
[05:41:09] + 20480 bytes downloaded
[05:41:09] + 30720 bytes downloaded
[05:41:09] + 40960 bytes downloaded
...
[05:41:11] + 1218560 bytes downloaded
[05:41:11] + 1219181 bytes downloaded
[05:41:11] Verifying core Core_7a.fah...
[05:41:11] Signature is VALID
[05:41:11]
[05:41:11] Trying to unzip core FahCore_7a.exe
[05:41:12] Decompressed FahCore_7a.exe (3102232 bytes) successfully
[05:41:12] + Core successfully engaged
[05:41:19]
[05:41:19] + Processing work unit
[05:41:19] Core required: FahCore_7a.exe
[05:41:19] Core found.
[05:41:19] Working on Unit 03 [March 26 05:41:19]
[05:41:19] + Working ...
[05:41:20]
[05:41:20] *------------------------------*
[05:41:20] Folding@Home GB Gromacs Core
[05:41:20] Version 1.90 (March 8, 2006)
[05:41:20]
[05:41:20] Preparing to commence simulation
[05:41:20] - Assembly optimizations manually forced on.
[05:41:20] - Not checking prior termination.
[05:41:20] - Expanded 72393 -> 377333 (decompressed 521.2 percent)
[05:41:20] - Starting from initial work packet
[05:41:20]
[05:41:20] Project: 2082 (Run 29, Clone 11, Gen 2)
[05:41:20]
[05:41:20] Assembly optimizations on if available.
[05:41:20] Entering M.D.
[05:41:26] Protein: p2082_p53_in_AGBNP
[05:41:26]
[05:41:26] Writing local files
[05:41:26] GB activated
[05:41:26] Extra SSE boost OK.
[05:41:27] Writing local files
[05:41:27] Completed 0 out of 2000000 steps (0%)
[06:48:35] Writing local files
[06:48:35] Completed 20000 out of 2000000 steps (1%)
[07:55:43] Writing local files
[07:55:43] Completed 40000 out of 2000000 steps (2%)
[09:02:50] Writing local files
[09:02:50] Completed 60000 out of 2000000 steps (3%)
[10:09:58] Writing local files
[10:09:58] Completed 80000 out of 2000000 steps (4%)
[11:17:02] Writing local files
[11:17:02] Completed 100000 out of 2000000 steps (5%)
[12:24:57] Writing local files
[12:24:57] Completed 120000 out of 2000000 steps (6%)
[13:32:04] Writing local files
[13:32:04] Completed 140000 out of 2000000 steps (7%)
[14:39:08] Writing local files
[14:39:08] Completed 160000 out of 2000000 steps (8%)
[15:46:12] Writing local files
[15:46:12] Completed 180000 out of 2000000 steps (9%)
[16:53:19] Writing local files
[16:53:19] Completed 200000 out of 2000000 steps (10%)

Now, I know this system is stable, because it used to be my main machine. ASM is forced on from the last time it was rebooted, I hadn't terminated the thread correctly. But that was several WU's ago.

P.S. I've looked at the logs from the other machines, and they are running correctly, and the frame times have stabilized.


A64 2.4 on XP64:
[02:04:49] Folding@Home GB Gromacs Core
[02:04:49] Version 1.87 (February 26, 2006)
[02:04:49]
[02:04:49] Preparing to commence simulation
[02:04:49] - Looking at optimizations...
[02:04:49] - Created dyn
[02:04:49] - Files status OK
[02:04:49] - Expanded 72593 -> 377333 (decompressed 519.7 percent)
[02:04:49] - Starting from initial work packet
[02:04:49]
[02:04:49] Project: 2082 (Run 74, Clone 40, Gen 1)
[02:04:49]
[02:04:49] Assembly optimizations on if available.
[02:04:49] Entering M.D.
[02:04:55] Protein: p2082_p53_in_AGBNP
[02:04:55]
[02:04:55] Writing local files
[02:04:55] GB activated
[02:04:55] Extra SSE boost OK.
[02:04:55] Writing local files
[02:04:55] Completed 0 out of 2000000 steps (0)
[02:28:07] Writing local files
[02:28:07] Completed 20000 out of 2000000 steps (1)
[02:51:11] Writing local files
[02:51:11] Completed 40000 out of 2000000 steps (2)
[03:14:14] Writing local files
[03:14:14] Completed 60000 out of 2000000 steps (3)
[03:37:18] Writing local files
[03:37:18] Completed 80000 out of 2000000 steps (4)
[04:00:23] Writing local files
[04:00:23] Completed 100000 out of 2000000 steps (5)
[04:24:14] Writing local files
[04:24:14] Completed 120000 out of 2000000 steps (6)
[04:50:20] Writing local files
[04:50:20] Completed 140000 out of 2000000 steps (7)
[05:16:19] Writing local files
[05:16:19] Completed 160000 out of 2000000 steps (8)
[05:41:21] Writing local files
etc...

Celeron 2.0 on W2K:
[10:57:53] Folding@Home GB Gromacs Core
[10:57:53] Version 1.90 (March 8, 2006)
[10:57:53]
[10:57:53] Preparing to commence simulation
[10:57:53] - Looking at optimizations...
[10:57:53] - Created dyn
[10:57:53] - Files status OK
[10:57:53] - Expanded 72416 -> 377333 (decompressed 521.0 percent)
[10:57:53] - Starting from initial work packet
[10:57:53]
[10:57:53] Project: 2082 (Run 49, Clone 80, Gen 1)
[10:57:53]
[10:57:53] Assembly optimizations on if available.
[10:57:53] Entering M.D.
[10:57:59] Protein: p2082_p53_in_AGBNP
[10:57:59]
[10:57:59] Writing local files
[10:57:59] GB activated
[10:57:59] Extra SSE boost OK.
[10:58:00] Writing local files
[10:58:00] Completed 0 out of 2000000 steps (0)
[11:38:29] Writing local files
[11:38:29] Completed 20000 out of 2000000 steps (1)
[12:18:47] Writing local files
[12:18:47] Completed 40000 out of 2000000 steps (2)
[12:59:13] Writing local files
[12:59:13] Completed 60000 out of 2000000 steps (3)
[13:39:42] Writing local files
[13:39:42] Completed 80000 out of 2000000 steps (4)
[14:20:20] Writing local files
[14:20:20] Completed 100000 out of 2000000 steps (5)
etc...


 
Well, I sure feel stupid. Turns out that Suse's power management had turned down the clock speed to 1GHz as opposed to the 1.8 it should have been. Now I'm getting 33min frame times like it should be. Thank you everyone for the help and responses, though.



 
torgojr said:
Well, I sure feel stupid. Turns out that Suse's power management had turned down the clock speed to 1GHz as opposed to the 1.8 it should have been. Now I'm getting 33min frame times like it should be. Thank you everyone for the help and responses, though.



Click to expand...

LOL, glad you got it running up to speed. I was gonna mention that maybe linux was slowing it down, but to be slower than that 2.0ghz celery is WAY more overhead that i would have expected. So the next question is, why is it not overclocked past the 1.8? HUH? WELL?! HUH?!!! GIVE US A REASON?!!!!!!!!!!!!!111!!!!!!!!11!!!1!1! :D
 
torgojr said:
Well, I sure feel stupid. Turns out that Suse's power management had turned down the clock speed to 1GHz as opposed to the 1.8 it should have been. Now I'm getting 33min frame times like it should be. Thank you everyone for the help and responses, though.
Click to expand...
Thanks for sharing. That gives us something else to suggest if we have this question pop up again.
 
You must log in or register to reply here.
Back
Top