p8090

[rickoic]

Just started my new rig to folding yesterday.
ASUS ASMB4 with 4GB per cpu 2 Operton 6238 2.60GHz with total 24 cores Win 7 64-bit

First wu I caught for it was a p8090 (Run 50, Clone 4, Gen 11)
First frame took 38 minutes to complete.
Frames since have dropped down into the 11-15 minute range.
Was wondering if this is typical with this type wu?
Base Credit 925
Estimated Credit 3319
Estimated PPD 3649
Estimated TPF 13 mins 06 secs

Tks
Rick

[Napoleon]

Sounds EXTREMELY slow for a machine of that class, viewtopic.php?f=66&t=24057. Please post config and log.

[Joe_H]

Is that time for the first frame from your log or from what was displayed in Web Control or FAHControl? Initial estimates of frame times are known to be inaccurate when displayed in one of the control screens. What else is running on the server? Relatively minor continuous uses of CPU time by other processes can severely impact folding of a SMP WU. Often reducing the core count given to folding can improve the overall frame rate by leaving cores available for other processes.

[bruce]

What else is running on that machine?

[rickoic]

I have 2 GPU's running also (GTX 670 and GTX 650) they are producing as expected in excess of 10K PPD.
I've tried some different things to see if anything would help.
1. Pausing 1 and then both GPU's had no effect.
2. Pausing CPU and dropping cores from 24 to 22. Just lengthened frame time and dropped the points.

I'm sure some of the first frame time was from everything getting set up, and I was playing with it a little, but it still was a long frame time.
I paused it at 18 or so minutes and restarted it to see what would happen. Showed it had compleded 0.30 of first frame 380 of 1000000.
Then overnight Win 7 restarted with all the new updates that a new install gathers but here are the frame times since it restarted at 10:32:55

Code: Select all

10:32:55: 130350 out of 1000000 steps (13%)
11:00:02: 140000 out of 1000000 steps (14%)
11:12:56: 150000 out of 1000000 steps (15%)
11:24:38: 160000 out of 1000000 steps (16%)
11:38:09: 170000 out of 1000000 steps (17%)
11:49:26: 180000 out of 1000000 steps (18%)
12:01:19: 190000 out of 1000000 steps (19%)
12:14:18: 200000 out of 1000000 steps (20%)
12:27:57: 210000 out of 1000000 steps (21%)
12:39:11: 220000 out of 1000000 steps (22%)
12:51:32: 230000 out of 1000000 steps (23%)
13:06:30: 240000 out of 1000000 steps (24%)
13:19:32: 250000 out of 1000000 steps (25%)
13:31:59: 260000 out of 1000000 steps (26%)
13:47:11: 270000 out of 1000000 steps (27%)
14:00:17: 280000 out of 1000000 steps (28%)
14:15:16: 290000 out of 1000000 steps (29%)
14:35:05: 300000 out of 1000000 steps (30%)
Heres where I paused it to drop it to 22 cores.
14:50:59: 299240 out of 1000000 steps (29%)
14:51:20: 300000 out of 1000000 steps (30%)
15:05:15: 310000 out of 1000000 steps (31%)
15:20:10: 320000 out of 1000000 steps (32%)
15:35:04: 330000 out of 1000000 steps (33%)
15:46:14: 340000 out of 1000000 steps (34%)
15:58:50: 350000 out of 1000000 steps (35%)
Heres where I paused it to go back to 24 cores.
Which I had just did and its just completed 1 frame since then.
16:16:35: 354650 out of 1000000 steps (35%)
Printed date time in here.
17:03:05 3600000 out of 1000000 steps (36%)

Display shows it's finished 37% now but hasn't updated log.
Something happened here however as now it showing this:
Base Credit 925
Estimated Credit 3616
Estimated PPD 12751
Estimated TPF 4 mins 05 secs

17:15:24 3700000 out of 1000000 steps (37%)

At 16:53:40 My GTX 670 finished a p8071
At 16:54:57 it stated another p8071

So maybe pausing it doesn't release anything, just makes it idle.
Removed the GTX 670 from folding instead of pausing it to see what that will do.

17:27:23: 3800000 out of 1000000 steps (38%)

Nothing happening yet so maybe resources the GTX 670 was using just dropped to the GTX 650 which as 33 minutes to compete its wu.
When it finishes I'll remove it from folding also and come back later to let you know what's happening.

Tks
Rick

Mod Edit: Added Code Tags - PantherX

[rickoic]

Ok, got all GPU's removed from folding and heres the CPU log file.

17:03:05: 360000 out of 1000000 steps (36%)
17:15:20: 370000 out of 1000000 steps (37%)
17:27:23: 380000 out of 1000000 steps (38%)
17:41:53: 390000 out of 1000000 steps (39%)
17:54:13: 400000 out of 1000000 steps (40%)
18:05:54: 410000 out of 1000000 steps (41%)
It was about 40.60% when I removed the last GPU from folding.
18:10:10: 420000 out of 1000000 steps (42%)
18:14:29: 430000 out of 1000000 steps (43%)
18:18:49: 440000 out of 1000000 steps (44%)
18:23:08: 450000 out of 1000000 steps (45%)
18:27:24: 460000 out of 1000000 steps (46%)

Deffinite improvement with frame times about 1/3 of before.
Base Credit 925
Estimated Credit 3579
Estimated PPD 11,938
Estimated TPF 4 mins 19 secs.

So guess I won't be doing any GPU folding on this rig.

Tks for looking and trying to help.
Rick

[TheWolf]

You should lock affinity of each GPU running to a free unused CPU core.
This should improve all folding times across the board.
CPU TPF may go up just a tad.

Edit: here is a example of how to for W7 64bit: http://www.evga.com/forums/fb.ashx?m=1930636
Don't pay much attention to the rest of the post only the picture.
Almost forgot, you'll have to be a member of the forum in order to see the picture.

[EXT64]

That still seems really slow for a machine of that caliber.

[Napoleon]

This just doesn't compute on my brain. rickoic please post a full log, along these lines (using my dualcore E2220 Mint14 64bit as an example):

Code: Select all

*********************** Log Started 2013-05-13T15:11:21Z ***********************
15:11:21:************************* Folding@home Client *************************
15:11:21:    Website: http://folding.stanford.edu/
15:11:21:  Copyright: (c) 2009-2013 Stanford University
15:11:21:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
15:11:21:       Args: 
15:11:21:     Config: /home/napoleon/fah/config.xml
15:11:21:******************************** Build ********************************
15:11:21:    Version: 7.3.6
15:11:21:       Date: Feb 18 2013
15:11:21:       Time: 07:24:08
15:11:21:    SVN Rev: 3923
15:11:21:     Branch: fah/trunk/client
15:11:21:   Compiler: GNU 4.4.7
15:11:21:    Options: -std=gnu++98 -O3 -funroll-loops -mfpmath=sse -ffast-math
15:11:21:             -fno-unsafe-math-optimizations -msse2
15:11:21:   Platform: linux2 3.2.0-1-amd64
15:11:21:       Bits: 64
15:11:21:       Mode: Release
15:11:21:******************************* System ********************************
15:11:21:        CPU: Intel(R) Pentium(R) Dual CPU E2220 @ 2.40GHz
15:11:21:     CPU ID: GenuineIntel Family 6 Model 15 Stepping 13
15:11:21:       CPUs: 2
15:11:21:     Memory: 3.86GiB
15:11:21:Free Memory: 3.01GiB
15:11:21:    Threads: POSIX_THREADS
15:11:21:Has Battery: false
15:11:21: On Battery: false
15:11:21: UTC offset: 3
15:11:21:        PID: 2400
15:11:21:        CWD: /home/napoleon/fah
15:11:21:         OS: Linux 3.5.0-17-generic x86_64
15:11:21:    OS Arch: AMD64
15:11:21:       GPUs: 1
15:11:21:      GPU 0: NVIDIA:1 G96 [GeForce 9400 GT]
15:11:21:       CUDA: 1.1
15:11:21:CUDA Driver: 5000
15:11:21:***********************************************************************
15:11:21:<config>
15:11:21:  <!-- Folding Core -->
15:11:21:  <checkpoint v='30'/>
15:11:21:
15:11:21:  <!-- Folding Slot Configuration -->
15:11:21:  <power v='full'/>
15:11:21:
15:11:21:  <!-- HTTP Server -->
15:11:21:  <allow v='192.168.0.100-192.168.0.199'/>
15:11:21:
15:11:21:  <!-- Logging -->
15:11:21:  <log-rotate-max v='1000'/>
15:11:21:
15:11:21:  <!-- Network -->
15:11:21:  <proxy v=':8080'/>
15:11:21:
15:11:21:  <!-- Remote Command Server -->
15:11:21:  <command-allow-no-pass v='192.168.0.100-192.168.0.199'/>
15:11:21:  <password v='***'/>
15:11:21:
15:11:21:  <!-- Slot Control -->
15:11:21:  <pause-on-battery v='false'/>
15:11:21:  <pause-on-start v='true'/>
15:11:21:
15:11:21:  <!-- User Information -->
15:11:21:  <passkey v='********************************'/>
15:11:21:  <team v='191980'/>
15:11:21:  <user v='GREYHOUND_SMP'/>
15:11:21:
15:11:21:  <!-- Work Unit Control -->
15:11:21:  <next-unit-percentage v='100'/>
15:11:21:
15:11:21:  <!-- Folding Slots -->
15:11:21:  <slot id='0' type='CPU'>
15:11:21:    <cpus v='2'/>
15:11:21:  </slot>
15:11:21:</config>
15:11:21:Trying to access database...
15:11:21:Successfully acquired database lock
15:11:21:Enabled folding slot 00: PAUSED cpu:2 (paused)
15:44:20:FS00:Unpaused
15:44:20:WU00:FS00:Connecting to assign3.stanford.edu:8080
15:44:21:WU00:FS00:News: Welcome to Folding@Home
15:44:21:WU00:FS00:Assigned to work server 171.67.108.60
15:44:21:WU00:FS00:Requesting new work unit for slot 00: READY cpu:2 from 171.67.108.60
15:44:21:WU00:FS00:Connecting to 171.67.108.60:8080
15:44:23:WU00:FS00:Downloading 1.11MiB
15:44:29:WU00:FS00:Download 56.26%
15:44:33:WU00:FS00:Download complete
15:44:33:WU00:FS00:Received Unit: id:00 state:DOWNLOAD error:NO_ERROR project:8090 run:347 clone:4 gen:9 core:0xa4 unit:0x0000000a6652edcc516747300904ce73
15:44:34:WU00:FS00:Starting
15:44:34:WU00:FS00:Running FahCore: /home/napoleon/fah/FAHCoreWrapper /home/napoleon/fah/cores/www.stanford.edu/~pande/Linux/AMD64/beta/Core_a4.fah/FahCore_a4 -dir 00 -suffix 01 -version 703 -lifeline 2400 -checkpoint 30 -np 2
15:44:34:WU00:FS00:Started FahCore on PID 2599
15:44:34:WU00:FS00:Core PID:2603
15:44:34:WU00:FS00:FahCore 0xa4 started
15:44:34:WU00:FS00:0xa4:
15:44:34:WU00:FS00:0xa4:*------------------------------*
15:44:34:WU00:FS00:0xa4:Folding@Home Gromacs GB Core
15:44:34:WU00:FS00:0xa4:Version 2.27 (Dec. 15, 2010)
15:44:34:WU00:FS00:0xa4:
15:44:34:WU00:FS00:0xa4:Preparing to commence simulation
15:44:34:WU00:FS00:0xa4:- Looking at optimizations...
15:44:34:WU00:FS00:0xa4:- Created dyn
15:44:34:WU00:FS00:0xa4:- Files status OK
15:44:34:WU00:FS00:0xa4:- Expanded 1164322 -> 3098240 (decompressed 266.0 percent)
15:44:34:WU00:FS00:0xa4:Called DecompressByteArray: compressed_data_size=1164322 data_size=3098240, decompressed_data_size=3098240 diff=0
15:44:34:WU00:FS00:0xa4:- Digital signature verified
15:44:34:WU00:FS00:0xa4:
15:44:34:WU00:FS00:0xa4:Project: 8090 (Run 347, Clone 4, Gen 9)
15:44:34:WU00:FS00:0xa4:
15:44:34:WU00:FS00:0xa4:Assembly optimizations on if available.
15:44:34:WU00:FS00:0xa4:Entering M.D.
15:44:42:WU00:FS00:0xa4:Completed 0 out of 1000000 steps  (0%)
16:19:21:WU00:FS00:0xa4:Completed 10000 out of 1000000 steps  (1%)
16:53:16:WU00:FS00:0xa4:Completed 20000 out of 1000000 steps  (2%)
17:27:11:WU00:FS00:0xa4:Completed 30000 out of 1000000 steps  (3%)
18:01:07:WU00:FS00:0xa4:Completed 40000 out of 1000000 steps  (4%)
18:35:03:WU00:FS00:0xa4:Completed 50000 out of 1000000 steps  (5%)
19:09:01:WU00:FS00:0xa4:Completed 60000 out of 1000000 steps  (6%)
19:42:59:WU00:FS00:0xa4:Completed 70000 out of 1000000 steps  (7%)
20:16:56:WU00:FS00:0xa4:Completed 80000 out of 1000000 steps  (8%)
20:50:54:WU00:FS00:0xa4:Completed 90000 out of 1000000 steps  (9%)
******************************* Date: 2013-05-13 *******************************
21:24:52:WU00:FS00:0xa4:Completed 100000 out of 1000000 steps  (10%)
21:58:47:WU00:FS00:0xa4:Completed 110000 out of 1000000 steps  (11%)
22:32:46:WU00:FS00:0xa4:Completed 120000 out of 1000000 steps  (12%)

and by "my dualcore" I mean cpu:2, not cpu:24 or cpu:22.