Folding Forum

Posted: **Fri Feb 28, 2014 1:56 pm**

I benched it. Implicit single precision 74.8 ns/day, explicit 21.3 ns/day.

Sadly the AMD 7970M hasn't been benched on viewtopic.php?f=38&t=24225 which places my GPU between the HD 7790 and the GTX 660 on the explicit single precision, and between the desktop 560 Ti and the GTX 480 ... seems a bit slow for a GCN core even if mobility, no?

Posted: **Fri Feb 28, 2014 2:08 pm**

It all depends on what the maker did to add the "mobility" tag. Downclocks? Less or slower memory? Fewer shaders? Did you use the 14.1 drivers?

Posted: **Fri Feb 28, 2014 2:26 pm**

No largely because on NBR.com those drivers were not recommended for gaming, and I do a fair bit of gaming. 13.12.

Posted: **Tue Mar 04, 2014 10:52 pm**

I haven't tried GPU Folding for well over 2 years. When I upgraded my rigs to AMD 7750 GPUs, they just wouldn't Fold. The 2 CPUs have been Folding without any maintenance, for about 45K PPD total.

Has enough changed since 2012 that the 7750 should Fold without problem? Is there a recommended Catalyst load (using 13.1 and 13.4 currently).

Posted: **Tue Mar 04, 2014 11:20 pm**

Windows, Linux, or MacOS-X?

My ATI HD 7790 folds fine in Windows. I'm running a back-rev of the drivers [13.9] because it gives better performance than the current version but I believe that either should work.

What problem is reported. [Show the first part of your log.]

Posted: **Wed Mar 05, 2014 12:48 am**

Win7-64. Details are in my System Profile.

I haven't tried recently, and am not home to try. I just saw this thread resurrected, and wondered if there was any significant change in drivers or cores that would affect the lower-power AMD GPUs. Also, does the GPU core still rob performance from the SMP core?

IIRC, when I first tried, the core would start and immediately shut down.

Posted: **Wed Mar 05, 2014 1:40 am**

We don't typically recommend beta drivers, but the 14.1 drivers are the best performing AMD drivers for fahcore 17 work units.

Posted: **Wed Mar 05, 2014 2:15 am**

jrweiss wrote:I just saw this thread resurrected, and wondered if there was any significant change in drivers or cores that would affect the lower-power AMD GPUs. Also, does the GPU core still rob performance from the SMP core?

So yes, there has been some development!

The new GPU core 0x17 is based on OpenCL and the AMD's really like it. It still uses some cpu, but not much on my 7970. My NVidia's use a full core, but not my AMD.

So, I think you should give it a try and see how it works for you as your results may vary!

Posted: **Wed Mar 05, 2014 2:41 am**

jrweiss wrote:...Has enough changed since 2012 that the 7750 should Fold without problem? Is there a recommended Catalyst load (using 13.1 and 13.4 currently).

As others have stated, the most significant change is the release of FahCore_17 for AMD GPUs (the same FahCore is also being used by Nvidia GPUs too). You can read about it in the blog posts (1, 2, 3, 4, 5, 6).

With your existing drivers versions taken from your profile, you can download either the latest V7 Public Release (Details) or the Public Beta (Details) and after installation, it should auto detect the GPUs and by default, will fold only when the system is idle (Details). If you want to fold on the GPU even when the system is being used, please move the power slider to Full position. It should be very easy and straight-forward now.

Posted: **Wed Mar 05, 2014 2:52 am**

I'm already on FAH 7.36, so I'll try 7.43.

I just looked at the AMD web site, and 14.1 is no longer listed. 13.12 is the release and 14.2 beta...

Posted: **Wed Mar 05, 2014 3:16 am**

13.9 is no longer listed, but if you dig deep enough you can still find it. I have not tried 14.2-beta.

Using 13.9 drivers on Win7, FahCore_17 does use some CPU but not like later versions that I've tried. I see one CPU fully occupied for a few minutes at the beginning of a WU while it's initializing data files. Then it drops to ~1%. I'm not positive: it might revert to 100% of one CPU during checkpoints but I don't think so.

Over half of the current WU covering 29H27m, FahCore_17 has accumulated 23m 02s which is just over 1% I let it run while my Quad is set to use all 4 cpus. YMMV.

This DOES depend on the version of the drivers.

Posted: **Fri Mar 07, 2014 12:09 am**

Updated to FAH 7.44 and Cat 13.9. Seems to be working now. I'll see how graphics performance is affected, and whether SMP yield falls. First indications are that FAHcore A4 CPU usage is down to 87-88%, which indicates to me that FAHcore 17 is disallowing use of 1 virtual core on the i7-4770, even though Task Manager sez the System Idle process has the other 11-12% of the CPU...

Should I somehow force the SMP slot to use only 7 CPU cores?

Code: Select all

*********************** Log Started 2014-03-06T21:30:36Z ***********************
21:30:36:************************* Folding@home Client *************************
21:30:36:      Website: http://folding.stanford.edu/
21:30:36:    Copyright: (c) 2009-2014 Stanford University
21:30:36:       Author: Joseph Coffland <joseph@cauldrondevelopment.com>
21:30:36:         Args: 
21:30:36:       Config: C:/Apps/FAH/Data/config.xml
21:30:36:******************************** Build ********************************
21:30:36:      Version: 7.4.4
21:30:36:         Date: Mar 4 2014
21:30:36:         Time: 20:26:54
21:30:36:      SVN Rev: 4130
21:30:36:       Branch: fah/trunk/client
21:30:36:     Compiler: Intel(R) C++ MSVC 1500 mode 1200
21:30:36:      Options: /TP /nologo /EHa /Qdiag-disable:4297,4103,1786,279 /Ox -arch:SSE
21:30:36:               /QaxSSE2,SSE3,SSSE3,SSE4.1,SSE4.2 /Qopenmp /Qrestrict /MT /Qmkl
21:30:36:     Platform: win32 XP
21:30:36:         Bits: 32
21:30:36:         Mode: Release
21:30:36:******************************* System ********************************
21:30:36:          CPU: Intel(R) Core(TM) i7-4770K CPU @ 3.50GHz
21:30:36:       CPU ID: GenuineIntel Family 6 Model 60 Stepping 3
21:30:36:         CPUs: 8
21:30:36:       Memory: 15.69GiB
21:30:36:  Free Memory: 10.16GiB
21:30:36:      Threads: WINDOWS_THREADS
21:30:36:   OS Version: 6.1
21:30:36:  Has Battery: true
21:30:36:   On Battery: false
21:30:36:   UTC Offset: -8
21:30:36:          PID: 4076
21:30:36:          CWD: C:/Apps/FAH/Data
21:30:36:           OS: Windows 7 Professional
21:30:36:      OS Arch: AMD64
21:30:36:         GPUs: 1
21:30:36:        GPU 0: ATI:5 R575A [Radeon R7 250/HD 7700]
21:30:36:         CUDA: Not detected
21:30:36:Win32 Service: false
21:30:36:***********************************************************************
21:30:36:<config>
21:30:36:  <!-- Folding Core -->
21:30:36:  <checkpoint v='30'/>
21:30:36:
21:30:36:  <!-- Network -->
21:30:36:  <proxy v=':8080'/>
21:30:36:
21:30:36:  <!-- Slot Control -->
21:30:36:  <power v='full'/>
21:30:36:
21:30:36:  <!-- User Information -->
21:30:36:  <passkey v='********************************'/>
21:30:36:  <team v='3074'/>
21:30:36:  <user v='jrweiss'/>
21:30:36:
21:30:36:  <!-- Folding Slots -->
21:30:36:  <slot id='1' type='CPU'/>
21:30:36:</config>
21:30:36:Trying to access database...
21:30:36:Successfully acquired database lock
21:30:36:Enabled folding slot 01: READY cpu:8
21:30:36:WU00:FS01:Starting
. . .

*********************** Log Started 2014-03-06T21:30:36Z ***********************
23:06:45:WU00:FS00:Connecting to 171.67.108.201:80
23:06:45:WU00:FS00:Assigned to work server 171.64.65.69
23:06:45:WU00:FS00:Requesting new work unit for slot 00: READY gpu:0:R575A [Radeon R7 250/HD 7700] from 171.64.65.69
23:06:45:WU00:FS00:Connecting to 171.64.65.69:8080
23:06:45:WU00:FS00:Downloading 4.18MiB
23:06:47:WU00:FS00:Download complete
23:06:47:WU00:FS00:Received Unit: id:00 state:DOWNLOAD error:NO_ERROR project:8900 run:194 clone:0 gen:190 core:0x17 unit:0x00000112028c126651a651ad1e3ddb8a
23:06:47:WU00:FS00:Downloading core from http://www.stanford.edu/~pande/Win32/AMD64/ATI/R600/Core_17.fah
23:06:47:WU00:FS00:Connecting to www.stanford.edu:80
23:06:47:WU00:FS00:FahCore 17: Downloading 2.55MiB
23:06:50:WU00:FS00:FahCore 17: Download complete
23:06:50:WU00:FS00:Valid core signature
23:06:50:WU00:FS00:Unpacked 8.60MiB to cores/www.stanford.edu/~pande/Win32/AMD64/ATI/R600/Core_17.fah/FahCore_17.exe
23:06:50:WU00:FS00:Starting
23:06:50:WU00:FS00:Running FahCore: C:\Apps\FAH\Client/FAHCoreWrapper.exe C:/Apps/FAH/Data/cores/www.stanford.edu/~pande/Win32/AMD64/ATI/R600/Core_17.fah/FahCore_17.exe -dir 00 -suffix 01 -version 704 -lifeline 4076 -checkpoint 30 -gpu 0 -gpu-vendor ati
23:06:50:WU00:FS00:Started FahCore on PID 592
23:06:51:WU00:FS00:Core PID:6756
23:06:51:WU00:FS00:FahCore 0x17 started
23:06:51:WU00:FS00:0x17:*********************** Log Started 2014-03-06T23:06:51Z ***********************
23:06:51:WU00:FS00:0x17:Project: 8900 (Run 194, Clone 0, Gen 190)
23:06:51:WU00:FS00:0x17:Unit: 0x00000112028c126651a651ad1e3ddb8a
23:06:51:WU00:FS00:0x17:CPU: 0x00000000000000000000000000000000
23:06:51:WU00:FS00:0x17:Machine: 0
23:06:51:WU00:FS00:0x17:Reading tar file state.xml
23:06:52:WU00:FS00:0x17:Reading tar file system.xml
23:06:52:WU00:FS00:0x17:Reading tar file integrator.xml
23:06:52:WU00:FS00:0x17:Reading tar file core.xml
23:06:52:WU00:FS00:0x17:Digital signatures verified
23:06:52:WU00:FS00:0x17:Folding@home GPU core17
23:06:52:WU00:FS00:0x17:Version 0.0.52
23:09:13:WU00:FS00:0x17:Completed 0 out of 2500000 steps (0%)
23:09:13:WU00:FS00:0x17:Temperature control disabled. Requirements: single Nvidia GPU, tmax must be < 110 and twait >= 900
23:26:19:WU00:FS00:0x17:Completed 25000 out of 2500000 steps (1%)
23:42:41:WU00:FS00:0x17:Completed 50000 out of 2500000 steps (2%)
23:59:18:WU00:FS00:0x17:Completed 75000 out of 2500000 steps (3%)

Posted: **Fri Mar 07, 2014 12:20 am**

With the default settings, FAHClient will automatically reserve 1 CPU per GPU Slot. Since you have a single GPU, the CPU Slot is limited to 7 CPUs with default settings. The reason is that various drivers versions change the CPU Usage of FahCore_17 and to ensure that there wasn't any over-subscription of the CPU, the default settings ensure that each GPU gets 1 CPU.

However, if you have determined that on your set-up, FahCore_17 isn't constantly using 1 CPU, you can always change the default setting from 7 CPUs to 8 CPUs by using Advanced Control (AKA FAHControl).

Posted: **Fri Mar 07, 2014 6:13 am**

That appears to be the case. I'll try it.

Posted: **Sat Mar 08, 2014 3:32 am**

Came back after dinner and found the computer had crashed. Probably not coincidentally, the GPU WU had finished. Here are the relevant logs. Any clues?

Code: Select all

00:39:58:WU00:FS00:0x17:*********************** Log Started 2014-03-08T00:39:58Z ***********************
00:39:58:WU00:FS00:0x17:Project: 8900 (Run 194, Clone 0, Gen 190)
00:39:58:WU00:FS00:0x17:Unit: 0x00000112028c126651a651ad1e3ddb8a
00:39:58:WU00:FS00:0x17:CPU: 0x00000000000000000000000000000000
00:39:58:WU00:FS00:0x17:Machine: 0
00:39:58:WU00:FS00:0x17:Digital signatures verified
00:39:58:WU00:FS00:0x17:Folding@home GPU core17
00:39:58:WU00:FS00:0x17:Version 0.0.52
00:39:58:WU00:FS00:0x17:  Found a checkpoint file
00:40:04:WU01:FS01:0xa3:Using Gromacs checkpoints
00:40:04:WU01:FS01:0xa3:Mapping NT from 8 to 8 
00:40:06:WU01:FS01:0xa3:Resuming from checkpoint
00:40:06:WU01:FS01:0xa3:Verified 01/wudata_01.log
00:40:06:WU01:FS01:0xa3:Verified 01/wudata_01.trr
00:40:06:WU01:FS01:0xa3:Verified 01/wudata_01.edr
00:40:06:WU01:FS01:0xa3:Completed 209240 out of 500000 steps  (41%)
00:42:13:WU00:FS00:0x17:Completed 2300000 out of 2500000 steps (92%)
00:42:13:WU00:FS00:0x17:Temperature control disabled. Requirements: single Nvidia GPU, tmax must be < 110 and twait >= 900
00:44:25:WU01:FS01:0xa3:Completed 210000 out of 500000 steps  (42%)
00:54:15:WU01:FS01:0xa3:Completed 215000 out of 500000 steps  (43%)
00:59:01:WU00:FS00:0x17:Completed 2325000 out of 2500000 steps (93%)
01:02:05:WU01:FS01:0xa3:Completed 220000 out of 500000 steps  (44%)
01:09:52:WU01:FS01:0xa3:Completed 225000 out of 500000 steps  (45%)
01:15:44:WU00:FS00:0x17:Completed 2350000 out of 2500000 steps (94%)
01:18:34:WU01:FS01:0xa3:Completed 230000 out of 500000 steps  (46%)
01:26:55:WU01:FS01:0xa3:Completed 235000 out of 500000 steps  (47%)
01:32:40:WU00:FS00:0x17:Completed 2375000 out of 2500000 steps (95%)
01:35:24:WU01:FS01:0xa3:Completed 240000 out of 500000 steps  (48%)
01:43:18:WU01:FS01:0xa3:Completed 245000 out of 500000 steps  (49%)
01:49:24:WU00:FS00:0x17:Completed 2400000 out of 2500000 steps (96%)
01:51:25:WU01:FS01:0xa3:Completed 250000 out of 500000 steps  (50%)
01:59:12:WU01:FS01:0xa3:Completed 255000 out of 500000 steps  (51%)
02:06:22:WU00:FS00:0x17:Completed 2425000 out of 2500000 steps (97%)
02:07:01:WU01:FS01:0xa3:Completed 260000 out of 500000 steps  (52%)
02:14:50:WU01:FS01:0xa3:Completed 265000 out of 500000 steps  (53%)
02:22:39:WU01:FS01:0xa3:Completed 270000 out of 500000 steps  (54%)
02:23:03:WU00:FS00:0x17:Completed 2450000 out of 2500000 steps (98%)
02:30:41:WU01:FS01:0xa3:Completed 275000 out of 500000 steps  (55%)
02:38:27:WU01:FS01:0xa3:Completed 280000 out of 500000 steps  (56%)
02:40:02:WU00:FS00:0x17:Completed 2475000 out of 2500000 steps (99%)
02:46:16:WU01:FS01:0xa3:Completed 285000 out of 500000 steps  (57%)
02:54:05:WU01:FS01:0xa3:Completed 290000 out of 500000 steps  (58%)
02:56:45:WU00:FS00:0x17:Completed 2500000 out of 2500000 steps (100%)
02:56:46:WU02:FS00:Connecting to 171.67.108.201:80
02:56:46:WU02:FS00:Assigned to work server 171.64.65.69
02:56:46:WU02:FS00:Requesting new work unit for slot 00: RUNNING gpu:0:R575A [Radeon R7 250/HD 7700] from 171.64.65.69
02:56:46:WU02:FS00:Connecting to 171.64.65.69:8080
02:56:46:WU02:FS00:Downloading 4.17MiB
02:56:49:WU02:FS00:Download complete
02:56:49:WU02:FS00:Received Unit: id:02 state:DOWNLOAD error:NO_ERROR project:8900 run:593 clone:1 gen:123 core:0x17 unit:0x000000d1028c126651a6aa8df66229ba
02:56:58:WU00:FS00:0x17:Saving result file logfile_01.txt
02:56:58:WU00:FS00:0x17:Saving result file checkpointState.xml
02:57:00:WU00:FS00:0x17:Saving result file checkpt.crc
02:57:00:WU00:FS00:0x17:Saving result file log.txt
02:57:00:WU00:FS00:0x17:Saving result file positions.xtc
02:57:02:WU00:FS00:0x17:Folding@home Core Shutdown: FINISHED_UNIT
. . .
[crash here]
. . .
*********************** Log Started 2014-03-08T03:26:13Z ***********************
03:26:13:************************* Folding@home Client *************************
03:26:13:      Website: http://folding.stanford.edu/
03:26:13:    Copyright: (c) 2009-2014 Stanford University
03:26:13:       Author: Joseph Coffland <joseph@cauldrondevelopment.com>
03:26:13:         Args: 
03:26:13:       Config: C:/Apps/FAH/Data/config.xml
03:26:13:******************************** Build ********************************
03:26:13:      Version: 7.4.4
03:26:13:         Date: Mar 4 2014
03:26:13:         Time: 20:26:54
03:26:13:      SVN Rev: 4130
03:26:13:       Branch: fah/trunk/client
03:26:13:     Compiler: Intel(R) C++ MSVC 1500 mode 1200
03:26:13:      Options: /TP /nologo /EHa /Qdiag-disable:4297,4103,1786,279 /Ox -arch:SSE
03:26:13:               /QaxSSE2,SSE3,SSSE3,SSE4.1,SSE4.2 /Qopenmp /Qrestrict /MT /Qmkl
03:26:13:     Platform: win32 XP
03:26:13:         Bits: 32
03:26:13:         Mode: Release
03:26:13:******************************* System ********************************
03:26:13:          CPU: Intel(R) Core(TM) i7-4770K CPU @ 3.50GHz
03:26:13:       CPU ID: GenuineIntel Family 6 Model 60 Stepping 3
03:26:13:         CPUs: 8
03:26:13:       Memory: 15.69GiB
03:26:13:  Free Memory: 9.93GiB
03:26:13:      Threads: WINDOWS_THREADS
03:26:13:   OS Version: 6.1
03:26:13:  Has Battery: true
03:26:13:   On Battery: false
03:26:13:   UTC Offset: -8
03:26:13:          PID: 4332
03:26:13:          CWD: C:/Apps/FAH/Data
03:26:13:           OS: Windows 7 Professional
03:26:13:      OS Arch: AMD64
03:26:13:         GPUs: 1
03:26:13:        GPU 0: ATI:5 R575A [Radeon R7 250/HD 7700]
03:26:13:         CUDA: Not detected
03:26:13:Win32 Service: false
03:26:13:***********************************************************************
03:26:13:<config>
03:26:13:  <!-- Folding Core -->
03:26:13:  <checkpoint v='30'/>
03:26:13:
03:26:13:  <!-- Network -->
03:26:13:  <proxy v=':8080'/>
03:26:13:
03:26:13:  <!-- Slot Control -->
03:26:13:  <power v='full'/>
03:26:13:
03:26:13:  <!-- User Information -->
03:26:13:  <passkey v='********************************'/>
03:26:13:  <team v='3074'/>
03:26:13:  <user v='jrweiss'/>
03:26:13:
03:26:13:  <!-- Folding Slots -->
03:26:13:  <slot id='1' type='CPU'>
03:26:13:    <cpus v='8'/>
03:26:13:  </slot>
03:26:13:  <slot id='0' type='GPU'/>
03:26:13:</config>
03:26:13:Trying to access database...
03:26:13:Successfully acquired database lock
03:26:13:Enabled folding slot 01: READY cpu:8
03:26:13:Enabled folding slot 00: READY gpu:0:R575A [Radeon R7 250/HD 7700]
03:26:13:WU01:FS01:Starting
03:26:13:WU01:FS01:Running FahCore: C:\Apps\FAH\Client/FAHCoreWrapper.exe C:/Apps/FAH/Data/cores/www.stanford.edu/~pande/Win32/AMD64/Core_a3.fah/FahCore_a3.exe -dir 01 -suffix 01 -version 704 -lifeline 4332 -checkpoint 30 -np 8
03:26:13:WU01:FS01:Started FahCore on PID 4192
03:26:13:WU01:FS01:Core PID:4368
03:26:13:WU01:FS01:FahCore 0xa3 started
03:26:13:WU00:FS00:Sending unit results: id:00 state:SEND error:NO_ERROR project:8900 run:194 clone:0 gen:190 core:0x17 unit:0x00000112028c126651a651ad1e3ddb8a
03:26:13:WU00:FS00:Uploading 12.96MiB to 171.64.65.69
03:26:13:WU00:FS00:Connecting to 171.64.65.69:8080
03:26:13:WU02:FS00:Starting
03:26:13:WU02:FS00:Running FahCore: C:\Apps\FAH\Client/FAHCoreWrapper.exe C:/Apps/FAH/Data/cores/www.stanford.edu/~pande/Win32/AMD64/ATI/R600/Core_17.fah/FahCore_17.exe -dir 02 -suffix 01 -version 704 -lifeline 4332 -checkpoint 30 -gpu 0 -gpu-vendor ati
03:26:13:WU02:FS00:Started FahCore on PID 4496
03:26:13:WU02:FS00:Core PID:4340
03:26:13:WU02:FS00:FahCore 0x17 started
03:26:13:WU02:FS00:0x17:*********************** Log Started 2014-03-08T03:26:13Z ***********************
03:26:13:WU02:FS00:0x17:Project: 8900 (Run 593, Clone 1, Gen 123)
03:26:13:WU02:FS00:0x17:Unit: 0x000000d1028c126651a6aa8df66229ba
03:26:13:WU02:FS00:0x17:CPU: 0x00000000000000000000000000000000
03:26:13:WU02:FS00:0x17:Machine: 0
03:26:13:WU02:FS00:0x17:Reading tar file state.xml
03:26:14:WU02:FS00:0x17:Reading tar file system.xml
03:26:14:WU02:FS00:0x17:Reading tar file integrator.xml
03:26:14:WU02:FS00:0x17:Reading tar file core.xml
03:26:14:WU02:FS00:0x17:Digital signatures verified
03:26:14:WU02:FS00:0x17:Folding@home GPU core17
03:26:14:WU02:FS00:0x17:Version 0.0.52
03:26:19:WU00:FS00:Upload 13.50%
03:26:25:WU00:FS00:Upload 27.48%
03:26:31:WU00:FS00:Upload 41.46%
03:26:37:WU00:FS00:Upload 55.44%
03:26:43:WU00:FS00:Upload 68.46%
03:26:49:WU00:FS00:Upload 80.51%
03:26:55:WU00:FS00:Upload 93.04%
03:26:58:WU00:FS00:Upload complete
03:26:58:WU00:FS00:Server responded WORK_QUIT (404)
03:26:58:WARNING:WU00:FS00:Server did not like results, dumping
03:26:58:WU00:FS00:Cleaning up
03:28:25:WU02:FS00:0x17:Completed 0 out of 2500000 steps (0%)
03:28:25:WU02:FS00:0x17:Temperature control disabled. Requirements: single Nvidia GPU, tmax must be < 110 and twait >= 900

Folding Forum

Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..

Re: Been trying to fold for years..