Project 8046 (9, 28, 33)

[Jeremey]

Hi Guys I was wondering if this behavior is at all normal? As you see the time it takes to complete a frame has gone up and down all over the place.

Code: Select all

*********************** Log Started 2012-06-27T13:44:48Z ***********************
13:44:49:WU01:FS01:Connecting to assign-GPU.stanford.edu:80
13:44:53:WU01:FS01:Connecting to assign3.stanford.edu:8080
13:44:56:WU01:FS01:News: Welcome to Folding@Home
13:44:56:WU01:FS01:Assigned to work server 171.67.108.60
13:44:56:WU01:FS01:Requesting new work unit for slot 01: READY smp:6 from 171.67.108.60
13:44:56:WU01:FS01:Connecting to 171.67.108.60:8080
13:45:02:WU01:FS01:Downloading 468.55KiB
13:45:07:WU01:FS01:Download complete
13:45:07:WU01:FS01:Received Unit: id:01 state:DOWNLOAD error:OK project:8046 run:9 clone:28 gen:33 core:0xa4 unit:0x000000286652edcc4fd934a7bf6388b2
13:45:08:WU01:FS01:Downloading core from http://www.stanford.edu/~pande/Win32/AMD64/Core_a4.fah
13:45:08:WU01:FS01:Connecting to www.stanford.edu:80
13:45:09:WU01:FS01:FahCore a4: Downloading 2.89MiB
13:45:16:WU01:FS01:FahCore a4: 4.33%
13:45:22:WU01:FS01:FahCore a4: 17.31%
13:45:28:WU01:FS01:FahCore a4: 38.96%
13:45:35:WU01:FS01:FahCore a4: 54.10%
13:45:41:WU01:FS01:FahCore a4: 64.93%
13:45:47:WU01:FS01:FahCore a4: 82.24%
13:45:53:WU01:FS01:FahCore a4: 97.39%
13:45:54:WU01:FS01:FahCore a4: Download complete
13:45:54:WU01:FS01:Valid core signature
13:45:54:WU01:FS01:Unpacked 9.59MiB to cores/www.stanford.edu/~pande/Win32/AMD64/Core_a4.fah/FahCore_a4.exe
13:45:54:WU01:FS01:Starting
13:45:54:WU01:FS01:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:/Users/Jem/AppData/Roaming/FAHClient/cores/www.stanford.edu/~pande/Win32/AMD64/Core_a4.fah/FahCore_a4.exe -dir 01 -suffix 01 -version 701 -lifeline 3432 -checkpoint 15 -np 6
13:45:54:WU01:FS01:Started FahCore on PID 6140
13:45:56:WU01:FS01:Core PID:2496
13:45:56:WU01:FS01:FahCore 0xa4 started
13:45:56:WU01:FS01:0xa4:
13:45:56:WU01:FS01:0xa4:*------------------------------*
13:45:56:WU01:FS01:0xa4:Folding@Home Gromacs GB Core
13:45:56:WU01:FS01:0xa4:Version 2.27 (Dec. 15, 2010)
13:45:56:WU01:FS01:0xa4:
13:45:56:WU01:FS01:0xa4:Preparing to commence simulation
13:45:56:WU01:FS01:0xa4:- Looking at optimizations...
13:45:56:WU01:FS01:Downloading project 8046 description
13:45:56:WU01:FS01:Connecting to fah-web.stanford.edu:80
13:45:56:WU01:FS01:0xa4:- Created dyn
13:45:56:WU01:FS01:0xa4:- Files status OK
13:45:56:WU01:FS01:0xa4:- Expanded 479279 -> 1077644 (decompressed 224.8 percent)
13:45:56:WU01:FS01:0xa4:Called DecompressByteArray: compressed_data_size=479279 data_size=1077644, decompressed_data_size=1077644 diff=0
13:45:56:WU01:FS01:0xa4:- Digital signature verified
13:45:56:WU01:FS01:0xa4:
13:45:56:WU01:FS01:0xa4:Project: 8046 (Run 9, Clone 28, Gen 33)
13:45:56:WU01:FS01:0xa4:
13:45:56:WU01:FS01:0xa4:Assembly optimizations on if available.
13:45:56:WU01:FS01:0xa4:Entering M.D.
13:45:57:WU01:FS01:Project 8046 description downloaded successfully
13:46:02:WU01:FS01:0xa4:Mapping NT from 6 to 6 
13:46:03:WU01:FS01:0xa4:Completed 0 out of 250000 steps  (0%)
13:48:49:WU01:FS01:0xa4:Completed 2500 out of 250000 steps  (1%)
13:50:22:WU01:FS01:0xa4:Completed 5000 out of 250000 steps  (2%)
13:54:08:WU01:FS01:0xa4:Completed 7500 out of 250000 steps  (3%)
13:56:18:WU01:FS01:0xa4:Completed 10000 out of 250000 steps  (4%)
13:59:19:WU01:FS01:0xa4:Completed 12500 out of 250000 steps  (5%)
14:01:45:WU01:FS01:0xa4:Completed 15000 out of 250000 steps  (6%)
14:04:21:WU01:FS01:0xa4:Completed 17500 out of 250000 steps  (7%)
14:18:15:WU01:FS01:0xa4:Completed 20000 out of 250000 steps  (8%)
14:51:41:WU01:FS01:0xa4:Completed 22500 out of 250000 steps  (9%)
15:08:46:WU01:FS01:0xa4:Completed 25000 out of 250000 steps  (10%)
15:43:37:WU01:FS01:0xa4:Completed 27500 out of 250000 steps  (11%)
15:54:55:WU01:FS01:0xa4:Completed 30000 out of 250000 steps  (12%)
15:57:02:WU01:FS01:0xa4:Completed 32500 out of 250000 steps  (13%)
15:59:26:WU01:FS01:0xa4:Completed 35000 out of 250000 steps  (14%)
16:01:28:WU01:FS01:0xa4:Completed 37500 out of 250000 steps  (15%)
16:03:29:WU01:FS01:0xa4:Completed 40000 out of 250000 steps  (16%)
16:22:50:WU01:FS01:0xa4:Completed 42500 out of 250000 steps  (17%)
16:24:47:WU01:FS01:0xa4:Completed 45000 out of 250000 steps  (18%)
16:26:39:WU01:FS01:0xa4:Completed 47500 out of 250000 steps  (19%)
16:28:45:WU01:FS01:0xa4:Completed 50000 out of 250000 steps  (20%)
16:30:47:WU01:FS01:0xa4:Completed 52500 out of 250000 steps  (21%)
16:32:38:WU01:FS01:0xa4:Completed 55000 out of 250000 steps  (22%)
17:04:41:WU01:FS01:0xa4:Completed 57500 out of 250000 steps  (23%)
17:20:33:WU01:FS01:0xa4:Completed 60000 out of 250000 steps  (24%)
17:22:54:WU01:FS01:0xa4:Completed 62500 out of 250000 steps  (25%)
17:37:33:WU01:FS01:0xa4:Completed 65000 out of 250000 steps  (26%)
18:50:59:WU01:FS01:0xa4:Completed 67500 out of 250000 steps  (27%)
19:21:44:WU01:FS01:0xa4:Completed 70000 out of 250000 steps  (28%)
19:33:02:WU01:FS01:0xa4:Completed 72500 out of 250000 steps  (29%)
******************************** Date: 27/06/12 ********************************
20:05:29:WU01:FS01:0xa4:Completed 75000 out of 250000 steps  (30%)
20:34:24:WU01:FS01:0xa4:Completed 77500 out of 250000 steps  (31%)
20:36:13:WU01:FS01:0xa4:Completed 80000 out of 250000 steps  (32%)
20:38:43:WU01:FS01:0xa4:Completed 82500 out of 250000 steps  (33%)
20:52:27:WU01:FS01:0xa4:Completed 85000 out of 250000 steps  (34%)
21:27:25:WU01:FS01:0xa4:Completed 87500 out of 250000 steps  (35%)
21:29:39:WU01:FS01:0xa4:Completed 90000 out of 250000 steps  (36%)
22:06:10:WU01:FS01:0xa4:Completed 92500 out of 250000 steps  (37%)
22:36:11:WU01:FS01:0xa4:Completed 95000 out of 250000 steps  (38%)
23:32:08:WU01:FS01:0xa4:Completed 97500 out of 250000 steps  (39%)
23:38:41:WU01:FS01:0xa4:Completed 100000 out of 250000 steps  (40%)
00:05:44:WU01:FS01:0xa4:Completed 102500 out of 250000 steps  (41%)
00:19:46:WU01:FS01:0xa4:Completed 105000 out of 250000 steps  (42%)
00:37:23:WU01:FS01:0xa4:Completed 107500 out of 250000 steps  (43%)
00:50:40:WU01:FS01:0xa4:Completed 110000 out of 250000 steps  (44%)
00:53:29:WU01:FS01:0xa4:Completed 112500 out of 250000 steps  (45%)
00:56:31:WU01:FS01:0xa4:Completed 115000 out of 250000 steps  (46%)
01:07:00:WU01:FS01:0xa4:Completed 117500 out of 250000 steps  (47%)
01:24:53:WU01:FS01:0xa4:Completed 120000 out of 250000 steps  (48%)
01:34:04:WU01:FS01:0xa4:Completed 122500 out of 250000 steps  (49%)
******************************** Date: 28/06/12 ********************************
02:19:26:WU01:FS01:0xa4:Completed 125000 out of 250000 steps  (50%)
02:36:36:WU01:FS01:0xa4:Completed 127500 out of 250000 steps  (51%)

I am not sure if you need any more details, I am Running GPU and Smp with a 1090t and HD6850

[Joe_H]

GPU folding on an ATI video card uses most of one CPU core to process the data. That makes the SMP folding wait some of the time to keep the threads synchronized, and the TPF will be variable. The SMP folding slot can work better with the number of cores reduced, a setting of 4 or possibly 5. There have been a few reports of some WU's having problems with SMP settings to odd or prime numbers higher than 3, that is why I mention 5 being "possible". One way to test if the TPF settles down is to pause the GPU slot and see if the time between each percent becomes more regular.

[Jeremey]

It settled down to faster and more regular times when the last GPU unit finished and was being uploaded. I also set the cores to 4, when I start up again if it is still a little funny I will pause the GPU and see how it goes. It is looking a lot better now already. Though it does not seem the 4 processor change has actually taken effect I guess it will on a restart.

Thank you for the help Joe.

[bruce]

The change in number of SMP cores will take effect when the next WU starts or when you (Pause, then Fold) the SMP slot as well as if you actually restart.

An i7 running ATI GPU + SMP will attempt to fold using the hyperthreaded virtual cores. When running only SMP, if you reduce the SMP setting by either 1 or by 2, there will be almost no reduction in SMP performance. That means you'll have one or two free virtual cores to support the GPU and total performance will go up, as you've already found.

It's a good idea to avoid ALL odd numbers, especially when we're talking about a Hyperthreaded CPU. You may successfully run with 5 virtual cores for a long time and then you may run into a project that has lots of crashes and by then you'll have forgotten that somebody mentioned running with 4.

[Joe_H]

Less of an issue with a true hex-core processor such as the Phenom II 1090t being used to fold here, but definitely an issue to be aware of.