Page 1 of 1

Semi -Solved // PPD unimaginably low

Posted: Thu May 08, 2014 5:30 am
by superdude
Using : 3x 7950's / win7 x64 and 7.4 FAH client // 13.12 AMD cat

I started with 220k-ish PPD
I had been told that 14.4 cat control yields much better result so i tried that. However, in that setup 1 card would do about 90k-lsh and rest would do 15, so totalling still max of 150PPD. So trying differnt versions untill it came up with 5kPPD total on 3 x 7950.

Im sure you'll agree that 5k on 3 cards is definately not worth the power its getting something must be off. But i dont have a clue what. I tried to start over. And again, and again.Untill i finally decided to wipe my harddisk clean. And start from scratch. Fresh windows. fresh 13.12 cat and fresh client. And about 4 hours later...... still 5k

So im kinda at my wits end here. Only things i can think of is that either something is wrong with my account... maybe got blacklisted cause i retried so many times? Idk i just dont know what else it could be.... Or it has something to do with the project...? It keeps feeding me 13000 and 13001 projects. (also tried removing the work info in appdat... still would get 13000 and 13001 projects.. So still dont know if thats the culprit or not.

Any help is more then welcome cause at this rate im about to abandon folding all together. 5k ppd on 3 cards is definately not worth it (10 days to complete an assignment)
If you need any logs or more specifics please let me know.. Thanks in advance

Re: PPD unimaginably low

Posted: Thu May 08, 2014 5:54 am
by superdude
superdude wrote:Using : 3x 7950's / win7 x64 and 7.4 FAH client // 13.12 AMD cat
Or it has something to do with the project...? It keeps feeding me 13000 and 13001 projects. (also tried removing the work info in appdat... still would get 13000 and 13001 projects.. So still dont know if thats the culprit or not.
Still on low side but PPD just shot up .... Looking at the projects 13000 /13001 now get about 50k each instead of the 17k it was mentioning before.(pretty sure at least... or maybe im just seeing things :)) Still not anywhere near what i was getting before but a good start

Re: PPD unimaginably low

Posted: Thu May 08, 2014 7:14 am
by P5-133XL
The logs with the system and config sections would be rather useful.

Re: PPD unimaginably low

Posted: Thu May 08, 2014 7:39 am
by superdude
Latest log file from appdata FAHclient :

Code: Select all

*********************** Log Started 2014-05-08T07:05:43Z ***********************
07:05:43:************************* Folding@home Client *************************
07:05:43:      Website: http://folding.stanford.edu/
07:05:43:    Copyright: (c) 2009-2014 Stanford University
07:05:43:       Author: Joseph Coffland <joseph@cauldrondevelopment.com>
07:05:43:         Args: --client-type=advanced
07:05:43:       Config: C:/Users/me/AppData/Roaming/FAHClient/config.xml
07:05:43:******************************** Build ********************************
07:05:43:      Version: 7.4.4
07:05:43:         Date: Mar 4 2014
07:05:43:         Time: 20:26:54
07:05:43:      SVN Rev: 4130
07:05:43:       Branch: fah/trunk/client
07:05:43:     Compiler: Intel(R) C++ MSVC 1500 mode 1200
07:05:43:      Options: /TP /nologo /EHa /Qdiag-disable:4297,4103,1786,279 /Ox -arch:SSE
07:05:43:               /QaxSSE2,SSE3,SSSE3,SSE4.1,SSE4.2 /Qopenmp /Qrestrict /MT /Qmkl
07:05:43:     Platform: win32 XP
07:05:43:         Bits: 32
07:05:43:         Mode: Release
07:05:43:******************************* System ********************************
07:05:43:          CPU: AMD Athlon(tm) II X2 260 Processor
07:05:43:       CPU ID: AuthenticAMD Family 16 Model 6 Stepping 3
07:05:43:         CPUs: 2
07:05:43:       Memory: 3.98GiB
07:05:43:  Free Memory: 3.23GiB
07:05:43:      Threads: WINDOWS_THREADS
07:05:43:   OS Version: 6.1
07:05:43:  Has Battery: false
07:05:43:   On Battery: false
07:05:43:   UTC Offset: -7
07:05:43:          PID: 3516
07:05:43:          CWD: C:/Users/me/AppData/Roaming/FAHClient
07:05:43:           OS: Windows 7 Ultimate
07:05:43:      OS Arch: AMD64
07:05:43:         GPUs: 3
07:05:43:        GPU 0: ATI:5 Tahiti PRO [Radeon R9 280/HD 7900/8950]
07:05:43:        GPU 1: ATI:5 Tahiti PRO [Radeon R9 280/HD 7900/8950]
07:05:43:        GPU 2: ATI:5 Tahiti PRO [Radeon R9 280/HD 7900/8950]
07:05:43:         CUDA: Not detected
07:05:43:Win32 Service: false
07:05:43:***********************************************************************
07:05:43:<config>
07:05:43:  <!-- Folding Core -->
07:05:43:  <core-priority v='low'/>
07:05:43:
07:05:43:  <!-- Network -->
07:05:43:  <proxy v=':8080'/>
07:05:43:
07:05:43:  <!-- Slot Control -->
07:05:43:  <power v='full'/>
07:05:43:
07:05:43:  <!-- User Information -->
07:05:43:  <passkey v='********************************'/>
07:05:43:  <team v='224497'/>
07:05:43:  <user v='superdude2'/>
07:05:43:
07:05:43:  <!-- Folding Slots -->
07:05:43:  <slot id='0' type='CPU'>
07:05:43:    <paused v='true'/>
07:05:43:  </slot>
07:05:43:  <slot id='1' type='GPU'/>
07:05:43:  <slot id='2' type='GPU'/>
07:05:43:  <slot id='3' type='GPU'/>
07:05:43:</config>
07:05:43:Trying to access database...
07:05:43:Successfully acquired database lock
07:05:43:Enabled folding slot 00: PAUSED cpu:1 (by user)
07:05:43:Enabled folding slot 01: READY gpu:0:Tahiti PRO [Radeon R9 280/HD 7900/8950]
07:05:43:Enabled folding slot 02: READY gpu:1:Tahiti PRO [Radeon R9 280/HD 7900/8950]
07:05:43:Enabled folding slot 03: READY gpu:2:Tahiti PRO [Radeon R9 280/HD 7900/8950]
07:05:43:WU02:FS02:Starting
07:05:43:WU02:FS02:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:/Users/me/AppData/Roaming/FAHClient/cores/www.stanford.edu/~pande/Win32/AMD64/ATI/R600/Core_17.fah/FahCore_17.exe -dir 02 -suffix 01 -version 704 -lifeline 3516 -checkpoint 15 -gpu 1 -gpu-vendor ati
07:05:43:WU02:FS02:Started FahCore on PID 3580
07:05:43:WU02:FS02:Core PID:3592
07:05:43:WU02:FS02:FahCore 0x17 started
07:05:44:WU03:FS03:Starting
07:05:44:WU03:FS03:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:/Users/me/AppData/Roaming/FAHClient/cores/www.stanford.edu/~pande/Win32/AMD64/ATI/R600/Core_17.fah/FahCore_17.exe -dir 03 -suffix 01 -version 704 -lifeline 3516 -checkpoint 15 -gpu 2 -gpu-vendor ati
07:05:44:WU03:FS03:Started FahCore on PID 3600
07:05:44:WU03:FS03:Core PID:3612
07:05:44:WU03:FS03:FahCore 0x17 started
07:05:44:WU01:FS01:Starting
07:05:44:WU01:FS01:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:/Users/me/AppData/Roaming/FAHClient/cores/www.stanford.edu/~pande/Win32/AMD64/ATI/R600/Core_17.fah/FahCore_17.exe -dir 01 -suffix 01 -version 704 -lifeline 3516 -checkpoint 15 -gpu 0 -gpu-vendor ati
07:05:44:WU01:FS01:Started FahCore on PID 3620
07:05:44:WU01:FS01:Core PID:3632
07:05:44:WU01:FS01:FahCore 0x17 started
07:05:45:WU03:FS03:0x17:*********************** Log Started 2014-05-08T07:05:45Z ***********************
07:05:45:WU03:FS03:0x17:Project: 13000 (Run 1772, Clone 0, Gen 9)
07:05:45:WU03:FS03:0x17:Unit: 0x0000001a538b3db7531190c37a37b325
07:05:45:WU03:FS03:0x17:CPU: 0x00000000000000000000000000000000
07:05:45:WU03:FS03:0x17:Machine: 3
07:05:45:WU03:FS03:0x17:Digital signatures verified
07:05:45:WU03:FS03:0x17:Folding@home GPU core17
07:05:45:WU03:FS03:0x17:Version 0.0.52
07:05:45:WU01:FS01:0x17:*********************** Log Started 2014-05-08T07:05:45Z ***********************
07:05:45:WU01:FS01:0x17:Project: 13000 (Run 1793, Clone 0, Gen 11)
07:05:45:WU01:FS01:0x17:Unit: 0x00000020538b3db7531196bbb2d07db7
07:05:45:WU01:FS01:0x17:CPU: 0x00000000000000000000000000000000
07:05:45:WU01:FS01:0x17:Machine: 1
07:05:45:WU01:FS01:0x17:Digital signatures verified
07:05:45:WU01:FS01:0x17:Folding@home GPU core17
07:05:45:WU01:FS01:0x17:Version 0.0.52
07:05:45:WU02:FS02:0x17:*********************** Log Started 2014-05-08T07:05:45Z ***********************
07:05:45:WU02:FS02:0x17:Project: 13001 (Run 221, Clone 2, Gen 9)
07:05:45:WU02:FS02:0x17:Unit: 0x00000021538b3db753288e5d0f1237f0
07:05:45:WU02:FS02:0x17:CPU: 0x00000000000000000000000000000000
07:05:45:WU02:FS02:0x17:Machine: 2
07:05:45:WU02:FS02:0x17:Digital signatures verified
07:05:45:WU02:FS02:0x17:Folding@home GPU core17
07:05:45:WU02:FS02:0x17:Version 0.0.52
07:05:47:WU03:FS03:0x17:  Found a checkpoint file
07:05:48:WU01:FS01:0x17:  Found a checkpoint file
07:05:48:WU02:FS02:0x17:  Found a checkpoint file
Msi afterbuner is showing full load on all 3 cards. (actually it jumps up and down from 70-100) but one of the cards is still doing nothing even with full load. Other 2 are way under what they could be.
(also --client-type=advanced has been added) Cards are running @ 1025 cclock and 1250 Mclock
If need any other info to pinpoint a problem or if this wasnt the log you were looking for plz let me know..

Image

Re: PPD unimaginably low

Posted: Thu May 08, 2014 10:16 am
by artoar_11
I see in another thread a similar problem with the same video card (1x 7950).

viewtopic.php?f=38&t=26298

Re: PPD unimaginably low

Posted: Thu May 08, 2014 10:39 am
by EXT64
How is CPU usage looking with all cards running?

Re: PPD unimaginably low

Posted: Thu May 08, 2014 12:52 pm
by superdude
I can see the cpu graph spiking up soon as i open it and activly use desktop enviorment.. spikes up to about 90% usuage. So could very well be a bottleneck thing..
But the strange thing is I had 230k before stable for about 2 days... Up untill when i decided to update to 14.4 cat contrl. and all went downhill from there..
Also memory is 1,6 / 4 used...

Re: PPD unimaginably low

Posted: Thu May 08, 2014 1:28 pm
by 7im
Core_17 on the GPU uses a full CPU core. Feeding 3 GPUs with 2 CPUs is likely going to be a bottleneck.

Experiment running only 2 slots to find your optimal settings, then try adding a third slot and see what happens.

Re: PPD unimaginably low

Posted: Thu May 08, 2014 4:57 pm
by jrweiss
superdude wrote:But the strange thing is I had 230k before stable for about 2 days... Up untill when i decided to update to 14.4 cat contrl. and all went downhill from there..
Also memory is 1,6 / 4 used...
There are reports in other threads that the original Cat 14-4 release was bad, and they re-released it. If you don't have the new, smaller (~256MB) install file, download and try the new one.

Also, did you completely clean out your Registry in between Cat installations? If not, uninstall, clean the Registry with CCleaner or DriverSweeper or similar, and re-install. Reboot after each step, of course.

Re: PPD unimaginably low

Posted: Sun May 11, 2014 11:31 am
by superdude
Also here, thanks for all your help.
What I ended up doing. After didnt know what to do anymore was : Format harddisk (yes extreme measure :) But when in doubt start over :) Then reinstalled all drivers (13.12 cat and sdk 2.9) New folding @ home. And then i just turned away from my pc for half a day. And sure enough im back to 230kish PPD.

IN retrospect, and with your comments here's what i think was happening under 14.x, simple really when you think of it... :)
When i installed 14.x it showed my cards as R9 200's. but still working properly. But when using folding only one card would get 90k, other two would remain at 15k
3 x 7950 went from 75 to 90k PPD so total, so was increase of about 60k PPD.
Im guessing that this increase in PPD was just to much for my cpu to handle and it would start bottlenecking..... And thus it had to choose and keep choosing between cards to feed. But back to the way it was and im pretty sure i wont be tinkering with it for a while now... Ill take my 230k and smile :)

As for reason why at first run at fresh install it didnt want to go any higher then 5k ppd, im still in the dark. Guess would be either my username had some sort of redflag / penalty (since i had to reinstall f@h couple of times during the switching of the cat's. Or maybe my gpu's had to warm up/settle in to their new drivers.. I realloy dont know. Took about 2-3 hours but it started spiking up and after about 4-5 i was back at my original PPD.

So.. thanks for all your help and imput. Who knows, maybe this post will help someone in the future and serve its purpose... thanks

Re: PPD unimaginably low

Posted: Sun May 11, 2014 1:08 pm
by PantherX
superdude wrote:...As for reason why at first run at fresh install it didnt want to go any higher then 5k ppd, im still in the dark. Guess would be either my username had some sort of redflag / penalty (since i had to reinstall f@h couple of times during the switching of the cat's. Or maybe my gpu's had to warm up/settle in to their new drivers.. I realloy dont know. Took about 2-3 hours but it started spiking up and after about 4-5 i was back at my original PPD...
The reason is that with a fresh installation, there isn't any historical data for the Client to use to display the information. Thus, it uses the "worst case" information for the initial display (the previous releases wouldn't display any information which made donors think that nothing was working and attempted to "fix" it). However, once it finishes a few frames (IIRC, between 1% to 5%), then the values will be automatically reflected with actual data. While this approach is better than the previous releases, hopefully it can be further improved (https://fah.stanford.edu/projects/FAHClient/ticket/1129).

BTW, on systems with multiple non-identical GPUs, i.e. GTX 660 and GTX 650 on my system, what happens is when both GPUs get assigned WUs from the same Project for the first time, the estimates for GTX 660 is used for GTX 650 which leads to an artificial inflation. However, once GTX 650 has folded few frames, the actual estimates are then used. Moreover, if the system is restarted, the estimates for my GPUs are incorrect until the first few frames are folded.

Re: Semi -Solved // PPD unimaginably low

Posted: Sun May 11, 2014 2:37 pm
by 7im
No such thing as a user based red flag/penalty! As PX, said, just had to get some history. I'm sure if you looked at the frame times and calculated the PPD manually, it was never running at 5K PPD.

Re: Semi -Solved // PPD unimaginably low

Posted: Thu Jun 05, 2014 2:52 pm
by Krusher33
Opposite is happening for me. I was getting a good PPD on my single 290X rig. But I borked up my OS playing around with overclocks and had to reinstall.

Now the client says I'm getting just 26k PPD? And the GPU usage in afterburner looks exactly like the one in this thread. I'm using same drivers and client as prior install. I can't seem to figure out what's going on with it.

Re: Semi -Solved // PPD unimaginably low

Posted: Thu Jun 05, 2014 3:04 pm
by bruce
Krusher33 wrote:Opposite is happening for me. I was getting a good PPD on my single 290X rig. But I borked up my OS playing around with overclocks and had to reinstall.

Now the client says I'm getting just 26k PPD? And the GPU usage in afterburner looks exactly like the one in this thread. I'm using same drivers and client as prior install. I can't seem to figure out what's going on with it.
You absolutely must ignore any PPD predictions which are obtained before the client has been running long enough to complete a few percent of the WU. If you look at the numbers before there's enough of a history of your progress, there's nothing on which to base the prediction.

Re: Semi -Solved // PPD unimaginably low

Posted: Thu Jun 05, 2014 3:26 pm
by Krusher33
bruce wrote:
Krusher33 wrote:Opposite is happening for me. I was getting a good PPD on my single 290X rig. But I borked up my OS playing around with overclocks and had to reinstall.

Now the client says I'm getting just 26k PPD? And the GPU usage in afterburner looks exactly like the one in this thread. I'm using same drivers and client as prior install. I can't seem to figure out what's going on with it.
You absolutely must ignore any PPD predictions which are obtained before the client has been running long enough to complete a few percent of the WU. If you look at the numbers before there's enough of a history of your progress, there's nothing on which to base the prediction.
It's not so much that being what I'm concerned with. I can mostly tell something's not right when Afterburner was showing weird fluctuations in the GPU usage.

I think I fixed it though. I changed settings in Afterburner to "unlock voltage control", "unlock voltage monitoring", "extend Official overclocking limits", "Disable ULPS", and "Unofficial overclocking mode without PowerPlay support". It prompted me to restart machine and now Afterburner is showing a more stable GPU usage. Rather than the up-downs like in the screen shot in this thread, it's more of 100% with an occasional downspike to 80%.

Still not as stable as I remember with my previous windows installation though.