Folding Forum

Posted: **Tue Sep 03, 2013 10:38 pm**

I'm seeing unusually long frame times on this WU. Frame times on this machine are usually around 20-25 minutes, and this one's at close to 2 and a half hours. The only other project I see in my work unit history with times this long was 7039, but that one had a final deadline of 72 days.
There's no way I'm going to hit project 7087's 5-day deadline at this speed -- my ETA is currently 9 days.

Is it possible the deadline was underestimated?

Code: Select all

*********************** Log Started 2013-09-03T19:52:00Z ***********************
19:52:00:************************* Folding@home Client *************************
19:52:00:      Website: http://folding.stanford.edu/
19:52:00:    Copyright: (c) 2009-2012 Stanford University
19:52:00:       Author: Joseph Coffland <joseph@cauldrondevelopment.com>
19:52:00:         Args: --lifeline 2144 --command-port=36330
19:52:00:       Config: C:/Documents and Settings/All Users/Application
19:52:00:               Data/FAHClient/config.xml
19:52:00:******************************** Build ********************************
19:52:00:      Version: 7.1.52
19:52:00:         Date: Mar 20 2012
19:52:00:         Time: 19:37:42
19:52:00:      SVN Rev: 3515
19:52:00:       Branch: fah/trunk/client
19:52:00:     Compiler: Intel(R) C++ MSVC 1500 mode 1200
19:52:00:      Options: /TP /nologo /EHa /Qdiag-disable:4297,4103,1786,279 /Ox -arch:SSE
19:52:00:               /QaxSSE2,SSE3,SSSE3,SSE4.1,SSE4.2 /Qopenmp /Qrestrict /MT
19:52:00:     Platform: win32 XP
19:52:00:         Bits: 32
19:52:00:         Mode: Release
19:52:00:******************************* System ********************************
19:52:00:          CPU: Intel(R) Core(TM)2 CPU T5600 @ 1.83GHz
19:52:00:       CPU ID: GenuineIntel Family 6 Model 15 Stepping 6
19:52:00:         CPUs: 2
19:52:00:       Memory: 2.99GiB
19:52:00:  Free Memory: 2.32GiB
19:52:00:      Threads: WINDOWS_THREADS
19:52:00:   On Battery: false
19:52:00:   UTC offset: -5
19:52:00:          PID: 3744
19:52:00:          CWD: C:/Documents and Settings/All Users/Application Data/FAHClient
19:52:00:           OS: Microsoft Windows XP Service Pack 3
19:52:00:      OS Arch: X86
19:52:00:         GPUs: 0
19:52:00:         CUDA: Not detected
19:52:00:Win32 Service: false
19:52:00:***********************************************************************
19:52:00:<config>
19:52:00:  <!-- Folding Slot Configuration -->
19:52:00:  <smp v='false'/>
19:52:00:
19:52:00:  <!-- Network -->
19:52:00:  <proxy v=':8080'/>
19:52:00:
19:52:00:  <!-- User Information -->
19:52:00:  <passkey v='********************************'/>
19:52:00:  <team v='35054'/>
19:52:00:  <user v='BABackman'/>
19:52:00:
19:52:00:  <!-- Folding Slots -->
19:52:00:  <slot id='0' type='UNIPROCESSOR'>
19:52:00:    <cpu-usage v='90'/>
19:52:00:  </slot>
19:52:00:</config>
19:52:00:Trying to access database...
19:52:01:Successfully acquired database lock
19:52:01:Enabled folding slot 00: READY uniprocessor
19:52:01:WU00:FS00:Starting
19:52:01:WU00:FS00:Running FahCore: "C:\Program Files\FAHClient/FAHCoreWrapper.exe" "C:/Documents and Settings/All Users/Application Data/FAHClient/cores/www.stanford.edu/~pande/Win32/x86/Core_a4.fah/FahCore_a4.exe" -dir 00 -suffix 01 -version 701 -lifeline 3744 -checkpoint 15 -cpu 90
19:52:01:WU00:FS00:Started FahCore on PID 3448
19:52:01:WU00:FS00:Core PID:3476
19:52:01:WU00:FS00:FahCore 0xa4 started
19:52:02:WU00:FS00:0xa4:
19:52:02:WU00:FS00:0xa4:*------------------------------*
19:52:02:WU00:FS00:0xa4:Folding@Home Gromacs GB Core
19:52:02:WU00:FS00:0xa4:Version 2.27 (Dec. 15, 2010)
19:52:02:WU00:FS00:0xa4:
19:52:02:WU00:FS00:0xa4:Preparing to commence simulation
19:52:02:WU00:FS00:0xa4:- Looking at optimizations...
19:52:02:WU00:FS00:0xa4:- Files status OK
19:52:02:Server connection id=1 on 0.0.0.0:36330 from 127.0.0.1
19:52:03:WU00:FS00:0xa4:- Expanded 4288984 -> 5584672 (decompressed 130.2 percent)
19:52:03:WU00:FS00:0xa4:Called DecompressByteArray: compressed_data_size=4288984 data_size=5584672, decompressed_data_size=5584672 diff=0
19:52:03:WU00:FS00:0xa4:- Digital signature verified
19:52:03:WU00:FS00:0xa4:
19:52:03:WU00:FS00:0xa4:Project: 7087 (Run 0, Clone 904, Gen 8)
19:52:03:WU00:FS00:0xa4:
19:52:04:WU00:FS00:0xa4:Assembly optimizations on if available.
19:52:04:WU00:FS00:0xa4:Entering M.D.
19:52:10:WU00:FS00:0xa4:Using Gromacs checkpoints
19:52:11:WU00:FS00:0xa4:Mapping NT from 1 to 1 
19:52:14:WU00:FS00:0xa4:Resuming from checkpoint
19:52:17:WU00:FS00:0xa4:Verified 00/wudata_01.log
19:52:19:WU00:FS00:0xa4:Verified 00/wudata_01.trr
19:52:19:WU00:FS00:0xa4:Verified 00/wudata_01.xtc
19:52:21:WU00:FS00:0xa4:Verified 00/wudata_01.edr
19:52:25:WU00:FS00:0xa4:Completed 59390 out of 500000 steps  (11%)
20:15:27:WU00:FS00:0xa4:Completed 60000 out of 500000 steps  (12%)
22:23:27:Server connection id=2 on 0.0.0.0:36330 from 127.0.0.1

Posted: **Tue Sep 03, 2013 11:32 pm**

Are any other background processes running which could have a negative impact on the TPF? Have you considered restarting the system and seeing if the TPF is shorter?

Posted: **Tue Sep 03, 2013 11:39 pm**

PantherX wrote:Are any other background processes running which could have a negative impact on the TPF? Have you considered restarting the system and seeing if the TPF is shorter?

No unusual background processes. The fahcore seems to be taking the same amount of CPU it usually does. I tried rebooting based on some other threads here, but that didn't have a noticeable effect.

Posted: **Tue Sep 03, 2013 11:45 pm**

Unusually long frame times are only meaningful in relation to the deadlines. If the ETAs are predictions and they may or may not be accurate. Assuming it is a good prediction and your hardware can't meet the deadline, how many hours per day have you been folding?

I'm not saying that the problem does not exist, but we do need additional information.

You've probably been getting a lot of assignments for FahCore_a4 which have less strenuous deadlines. The Txxxx CPUs are often the slowest two-core CPUs and the Pande Group may need to make some additional allowances for them. (I have one, too, and need to check on it.)

You are not alone. I saw a similar report a few hours ago: viewtopic.php?f=61&t=24877

Posted: **Wed Sep 04, 2013 9:35 am**

bruce wrote:Unusually long frame times are only meaningful in relation to the deadlines. If the ETAs are predictions and they may or may not be accurate. Assuming it is a good prediction and your hardware can't meet the deadline, how many hours per day have you been folding?

I'm folding 24/7. When I checked the frame time last night, it had crept down to "only" 1:55:00. I was hoping that was going to be a trend, but now its back up to 2:20. HFM's ETA calculations have been consistently close. I don't think it could be wrong enough to bring a 9/12 finish inside the 9/7 deadline. I guess I'll just hope that assignments aren't so tight by then and the odd projects aren't getting handed out so much, so when this one comes in late it's not redundant.

bruce wrote:You've probably been getting a lot of assignments for FahCore_a4 which have less strenuous deadlines. The Txxxx CPUs are often the slowest two-core CPUs and the Pande Group may need to make some additional allowances for them. (I have one, too, and need to check on it.)

You are not alone. I saw a similar report a few hours ago: viewtopic.php?f=61&t=24877

You're right about the A4's. Thanks for the explanation on the other thread on why unusual assignments are happening.

Posted: **Wed Sep 04, 2013 2:05 pm**

My old laptop has a T2060 which has two cores and runs at 1.60 GHz. I got a new assignment of a p8579 last night and it has now completed 7% at a frame rate of about 1h 33m. That means that it can finish that project in 6.3 days which is less than the 8.0 day timeout and significantly less than the 13.3 day expiration. The baseline points are 1657 and the kfactor is 3. By running 24x7 I'll be earning a bonus. While those characteristics are not identical to p7087, the kfactor is similar so I expect that the level of difficulty isn't significantly different.

This is very similar to what this machine has been getting for many months, now. I conclude that the server relocation has not introduced any new issues (though I'm open to new evidence). My choices for this machine have always been (A) The default configuration of SMP:2 can beat the timeout by a small factor as long as it runs very nearly 24x7 or (B) I can configure folding to use only one CPU which will force it to receive WUs from projects which earn lower points per day and which require a lower level of performance.

Posted: **Thu Sep 05, 2013 10:56 am**

There are two things you can do:

First of all: set CPU-usage to 100% - you set it to 90%.
Second one: Do not use uniprocessor, use flag -smp to use both CPU-Cores (and delete the setting: smp=false) but - check the temperatures of the CPU

Posted: **Thu Sep 05, 2013 1:12 pm**

Step 1 should be to update to the newest version of the client. 7.3.6

Folding Forum

Project 7087 (Run 0, Clone 904, Gen 8)

Project 7087 (Run 0, Clone 904, Gen 8)

Re: Project 7087 (Run 0, Clone 904, Gen 8)

Re: Project 7087 (Run 0, Clone 904, Gen 8)

Re: Project 7087 (Run 0, Clone 904, Gen 8)

Re: Project 7087 (Run 0, Clone 904, Gen 8)

Re: Project 7087 (Run 0, Clone 904, Gen 8)

Re: Project 7087 (Run 0, Clone 904, Gen 8)

Re: Project 7087 (Run 0, Clone 904, Gen 8)