Really Poor Folding WU..

BigDom

BigDom

BigDom

Associate
Joined
30 Nov 2002
Posts
1,851
Location
Birmingham
I seem to have a really bad Folding WU and wondered if thee is much one can do abou tit?

GBGromacs
Project 2096
178 points...

ETA to finish... 6d10h at 4%
By comparision on the same PC but other core
Project 2106 15h26m at 74% (less than 3 days total).
 
something has to be wrong - my worst rig gets the following...

Project : 2096
Core : Unknown
Frames : 100
Credit : 178


-- List of benchmarks --

David ->
Min. Time / Frame : 19mn 11s - 133.61 ppd
Avg. Time / Frame : 19mn 11s - 133.61 ppd

that's an XP2100+ at stock with PC133 SDRAM, second slowest machine I have
In fact this WU is one of the better ones I've seen on that machine :o
sounds more like the SSE hasn't kicked in for some reason on yours - check the log for this:

[22:10:59] Project: 2096 (Run 12, Clone 23, Gen 6)
[22:10:59]
[22:10:59] Assembly optimizations on if available.
[22:10:59] Entering M.D.
[22:11:05] Protein: p2096_A21_agbnp_amber99
[22:11:05]
[22:11:05] Writing local files
[22:11:05] GB activated
[22:11:05] Extra SSE boost OK.
[22:11:05] Writing local files
[22:11:05] Completed 0 out of 4000000 steps (0)
[22:21:59] Timered checkpoint triggered.
[22:32:01] Timered checkpoint triggered.
[22:33:20] Writing local files
[22:33:20] Completed 40000 out of 4000000 steps (1)

If you've not already switched machines then try restarting that core to see if that does the trick
 
Yep SSE boost appears to be working. I'm going to drop the WU on an old XP processor and just let it munch it.

Project : 2096
Core : GBGromacs
Frames : 100
Credit : 178


-- Dom - C2 --

Min. Time / Frame : 32mn 09s - 79.73 ppd
Avg. Time / Frame : 1h 38mn 04s - 26.14 ppd

[14:12:00] Folding@Home GB Gromacs Core
[14:12:00] Version 1.90 (March 8, 2006)
[14:12:00]
[14:12:00] Preparing to commence simulation
[14:12:00] - Assembly optimizations manually forced on.
[14:12:00] - Not checking prior termination.
[14:12:01] - Expanded 16725 -> 142191 (decompressed 850.1 percent)
[14:12:01]
[14:12:01] Project: 2096 (Run 27, Clone 12, Gen 2)
[14:12:01]
[14:12:01] Assembly optimizations on if available.
[14:12:01] Entering M.D.
[14:12:21] (Starting from checkpoint)
[14:12:21] Protein: p2096_A21_agbnp_amber99
[14:12:21]
[14:12:21] Writing local files
[14:12:21] GB activated
[14:12:21] Completed 24103 out of 4000000 steps (1)
[14:12:21] Extra SSE boost OK.


That core appears to be the one that usually get's hampered by other uses (gaming etc.) BUT all other projects average at least 60-70, usually 80ish with a peak of anything up to about 140.
 
Last edited:
hmmm tis strange for sure

I got a strange thing happen on friday night - got a random popup which I didn't think much of at the time and just closed it. Then yesterday morning I checked up on the progress of my WU and found it hadn't made any progress at all since the night before.

A quick check of task manager showed that the program which handles my USB wireless data was taking 99% then releasing it to fahcore then taking it again. I scheduled a virus scan at next boot and restarted the machine - seemed to do the trick as it found a couple of suspicious characters

i don't think that's the case here though since yours is obviously just a bit slow rather than stopping totally - see how it goes on the old XP, if it takes even longer per frame then at least you know it's not your machine
 
Project : 2096
Core : GBGromacs
Frames : 100
Credit : 178


-- PWU --

Min. Time / Frame : 38mn 41s - 66.26 ppd
Avg. Time / Frame : 38mn 41s - 66.26 ppd


Go Figure....
 
really really odd - at least it's not much slower than it was going on the dual core

i know that some WUs can have faster and slower frames but I've never heard of them taking twice as long, there have also been WUs where the first generation fold a lot faster/slower than the later generations but once again yours is a gen2 so I wouldn't think that would be an issue

it's gonna majorly suck if it takes a week to do a 178 point WU :o
could be worth deleting the core and seeing if a fresh download of it helps - really scraping the bottom of the barrel now :o
 
I'll leave it' it's a cruncher I don't use much and just get it out of the way. Just stranger a XP1800 is close in performance to one core on my 3800+... May well be quicker on an intel but too much hassle to swap :)
 
You must log in or register to reply here.
Back
Top Bottom