Folding Forum

Posted: **Sat Jan 03, 2015 11:36 am**

Hey guys only seem to be getting 72K PPD out of this card with my i4790 on GB z97X-SOC Force using 344.48 drivers.

heres a SS

kind regards
eddy

Posted: **Sat Jan 03, 2015 11:58 am**

Here is my current F@H Log
_________________________________

Code: Select all

15:33:56:Saving configuration to config.xml
15:33:56:<config>
15:33:56:  <!-- Network -->
15:33:56:  <proxy v=':8080'/>
15:33:56:
15:33:56:  <!-- Slot Control -->
15:33:56:  <power v='FULL'/>
15:33:56:
15:33:56:  <!-- User Information -->
15:33:56:  <team v='32'/>
15:33:56:  <user v='eddyg519'/>
15:33:56:
15:33:56:  <!-- Folding Slots -->
15:33:56:  <slot id='0' type='CPU'/>
15:33:56:  <slot id='1' type='GPU'/>
15:33:56:</config>
15:34:09:Saving configuration to config.xml
15:34:09:<config>
15:34:09:  <!-- Network -->
15:34:09:  <proxy v=':8080'/>
15:34:09:
15:34:09:  <!-- Slot Control -->
15:34:09:  <power v='FULL'/>
15:34:09:
15:34:09:  <!-- User Information -->
15:34:09:  <team v='32'/>
15:34:09:  <user v='eddyg519'/>
15:34:09:
15:34:09:  <!-- Folding Slots -->
15:34:09:  <slot id='0' type='CPU'/>
15:34:09:  <slot id='1' type='GPU'/>
15:34:09:</config>
15:36:40:WU01:FS01:0x15:Completed  12800000 out of 40000000 steps (32%).
15:38:25:FS00:Paused
15:38:25:FS01:Paused
15:38:25:FS00:Shutting core down
15:38:25:FS01:Shutting core down
15:38:26:WU01:FS01:FahCore returned: INTERRUPTED (102 = 0x66)
15:38:34:WU00:FS00:0xa3:Client no longer detected. Shutting down core 
15:38:34:WU00:FS00:0xa3:
15:38:34:WU00:FS00:0xa3:Folding@home Core Shutdown: CLIENT_DIED
15:38:35:WU00:FS00:FahCore returned: INTERRUPTED (102 = 0x66)
15:39:14:Saving configuration to config.xml
15:39:14:<config>
15:39:14:  <!-- Network -->
15:39:14:  <proxy v=':8080'/>
15:39:14:
15:39:14:  <!-- Slot Control -->
15:39:14:  <power v='FULL'/>
15:39:14:
15:39:14:  <!-- User Information -->
15:39:14:  <team v='32'/>
15:39:14:  <user v='eddyg519'/>
15:39:14:
15:39:14:  <!-- Folding Slots -->
15:39:14:  <slot id='0' type='CPU'>
15:39:14:    <paused v='true'/>
15:39:14:  </slot>
15:39:14:  <slot id='1' type='GPU'>
15:39:14:    <paused v='true'/>
15:39:14:  </slot>
15:39:14:</config>
15:46:28:FS00:Unpaused
15:46:28:FS01:Unpaused
15:46:28:WU00:FS00:Starting
15:46:28:WU00:FS00:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:/Users/eddy/AppData/Roaming/FAHClient/cores/web.stanford.edu/~pande/Win32/AMD64/Core_a3.fah/FahCore_a3.exe -dir 00 -suffix 01 -version 704 -lifeline 5064 -checkpoint 15 -np 7
15:46:28:WU00:FS00:Started FahCore on PID 1776
15:46:28:WU00:FS00:Core PID:3200
15:46:28:WU00:FS00:FahCore 0xa3 started
15:46:28:WU01:FS01:Starting
15:46:28:WU01:FS01:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:/Users/eddy/AppData/Roaming/FAHClient/cores/web.stanford.edu/~pande/Win32/AMD64/NVIDIA/Fermi/Core_15.fah/FahCore_15.exe -dir 01 -suffix 01 -version 704 -lifeline 5064 -checkpoint 15 -gpu 0 -gpu-vendor nvidia
15:46:28:WU01:FS01:Started FahCore on PID 1684
15:46:28:WU01:FS01:Core PID:4800
15:46:28:WU01:FS01:FahCore 0x15 started
15:46:29:WU00:FS00:0xa3:
15:46:29:WU00:FS00:0xa3:*------------------------------*
15:46:29:WU00:FS00:0xa3:Folding@Home Gromacs SMP Core
15:46:29:WU00:FS00:0xa3:Version 2.27 (Dec. 15, 2010)
15:46:29:WU00:FS00:0xa3:
15:46:29:WU00:FS00:0xa3:Preparing to commence simulation
15:46:29:WU00:FS00:0xa3:- Looking at optimizations...
15:46:29:WU00:FS00:0xa3:- Files status OK
15:46:29:WU00:FS00:0xa3:- Expanded 3793491 -> 4166484 (decompressed 109.8 percent)
15:46:29:WU00:FS00:0xa3:Called DecompressByteArray: compressed_data_size=3793491 data_size=4166484, decompressed_data_size=4166484 diff=0
15:46:29:WU00:FS00:0xa3:- Digital signature verified
15:46:29:WU00:FS00:0xa3:
15:46:29:WU00:FS00:0xa3:Project: 8533 (Run 1, Clone 0, Gen 457)
15:46:29:WU00:FS00:0xa3:
15:46:29:WU00:FS00:0xa3:Assembly optimizations on if available.
15:46:29:WU00:FS00:0xa3:Entering M.D.
15:46:29:WU01:FS01:0x15:
15:46:29:WU01:FS01:0x15:*------------------------------*
15:46:29:WU01:FS01:0x15:Folding@Home GPU Core
15:46:29:WU01:FS01:0x15:Version                2.25 (Wed May 9 17:03:01 EDT 2012)
15:46:29:WU01:FS01:0x15:Build host             AmoebaRemote
15:46:29:WU01:FS01:0x15:Board Type             NVIDIA/CUDA
15:46:29:WU01:FS01:0x15:Core                   15
15:46:29:WU01:FS01:0x15:
15:46:29:WU01:FS01:0x15:Window's signal control handler registered.
15:46:29:WU01:FS01:0x15:Preparing to commence simulation
15:46:29:WU01:FS01:0x15:- Looking at optimizations...
15:46:29:WU01:FS01:0x15:- Files status OK
15:46:29:WU01:FS01:0x15:sizeof(CORE_PACKET_HDR) = 512 file=<>
15:46:29:WU01:FS01:0x15:- Expanded 125036 -> 501826 (decompressed 401.3 percent)
15:46:29:WU01:FS01:0x15:Called DecompressByteArray: compressed_data_size=125036 data_size=501826, decompressed_data_size=501826 diff=0
15:46:29:WU01:FS01:0x15:- Digital signature verified
15:46:29:WU01:FS01:0x15:
15:46:29:WU01:FS01:0x15:Project: 7621 (Run 817, Clone 0, Gen 313)
15:46:29:WU01:FS01:0x15:
15:46:29:WU01:FS01:0x15:Assembly optimizations on if available.
15:46:29:WU01:FS01:0x15:Entering M.D.
15:46:30:WU01:FS01:0x15:Will resume from checkpoint file 01/wudata_01.ckp
15:46:30:WU01:FS01:0x15:Tpr hash 01/wudata_01.tpr:  1243417863 2449323202 3053042270 3134505598 1483402944
15:46:30:WU01:FS01:0x15:GPU device id=0
15:46:30:WU01:FS01:0x15:Working on Protein
15:46:30:WU01:FS01:0x15:Client config unavailable.
15:46:31:WU01:FS01:0x15:Starting GUI Server
15:46:34:WU00:FS00:0xa3:Using Gromacs checkpoints
15:46:35:WU00:FS00:0xa3:Mapping NT from 7 to 7 
15:46:35:WU00:FS00:0xa3:Resuming from checkpoint
15:46:35:WU00:FS00:0xa3:Verified 00/wudata_01.log
15:46:35:WU00:FS00:0xa3:Verified 00/wudata_01.trr
15:46:35:WU00:FS00:0xa3:Verified 00/wudata_01.edr
15:46:35:WU00:FS00:0xa3:Completed 187155 out of 500000 steps  (37%)
15:47:22:Saving configuration to config.xml
15:47:22:<config>
15:47:22:  <!-- Network -->
15:47:22:  <proxy v=':8080'/>
15:47:22:
15:47:22:  <!-- Slot Control -->
15:47:22:  <power v='FULL'/>
15:47:22:
15:47:22:  <!-- User Information -->
15:47:22:  <team v='32'/>
15:47:22:  <user v='eddyg519'/>
15:47:22:
15:47:22:  <!-- Folding Slots -->
15:47:22:  <slot id='0' type='CPU'/>
15:47:22:  <slot id='1' type='GPU'/>
15:47:22:</config>
15:47:32:WU01:FS01:0x15:Resuming from checkpoint
15:47:32:WU01:FS01:0x15:fcCheckPointResume: retreived and current tpr file hash:
15:47:32:WU01:FS01:0x15:   0   1243417863   1243417863
15:47:32:WU01:FS01:0x15:   1   2449323202   2449323202
15:47:32:WU01:FS01:0x15:   2   3053042270   3053042270
15:47:32:WU01:FS01:0x15:   3   3134505598   3134505598
15:47:32:WU01:FS01:0x15:   4   1483402944   1483402944
15:47:32:WU01:FS01:0x15:fcCheckPointResume: file hashes same.
15:47:32:WU01:FS01:0x15:fcCheckPointResume: state restored.
15:47:32:WU01:FS01:0x15:fcCheckPointResume: name 01/wudata_01.log Verified 01/wudata_01.log
15:47:32:WU01:FS01:0x15:fcCheckPointResume: name 01/wudata_01.trr Verified 01/wudata_01.trr
15:47:32:WU01:FS01:0x15:fcCheckPointResume: name 01/wudata_01.xtc Verified 01/wudata_01.xtc
15:47:32:WU01:FS01:0x15:fcCheckPointResume: name 01/wudata_01.edr Verified 01/wudata_01.edr
15:47:32:WU01:FS01:0x15:fcCheckPointResume: state restored 2
15:47:32:WU01:FS01:0x15:Resumed from checkpoint
15:47:32:WU01:FS01:0x15:Setting checkpoint frequency: 400000
15:47:32:WU01:FS01:0x15:Completed  12800001 out of 40000000 steps (32%).
15:47:33:WARNING:WU01:FS01:Detected clock skew (1 mins 05 secs), adjusting time estimates
15:50:28:WU01:FS01:0x15:Completed  13200000 out of 40000000 steps (33%).
15:53:22:WU01:FS01:0x15:Completed  13600000 out of 40000000 steps (34%).
15:55:49:WU00:FS00:0xa3:Completed 190000 out of 500000 steps  (38%)
15:56:13:WU01:FS01:0x15:Completed  14000000 out of 40000000 steps (35%).

Posted: **Sat Jan 03, 2015 12:19 pm**

Welcome to the folding@home support forum eddyg519.

Your current log shows you are running a core 15 project. Because these do not receive Quick Return Bonus they do produce lower PPD than some other projects. When you start receiving core 17 work you will see higher PPD assuming you have a qualified Passkey which I don't see in your log. Without a qualified passkey you will only receive the base credit for your returns and thus much lower than possible PPD. Monitoring programs do not know if you have a qualified passkey so they will display your estimated PPD as if you do have one.

Adding the client-type option of advanced to your GPU slot configuration may help in getting core 17 work but there are no guarantees ... the core 15 work must be finished too. Linux users are only being assigned core 17 work ( limited to one project at this time ) so they do not see any core 15 projects and are unlikely to since there is no core 15 for linux.

Please see https://www.reddit.com/r/foldingathome/ ... _projects/ for a discussion of the core 15 projects and their PPD.

Posted: **Sat Jan 03, 2015 12:49 pm**

I sort of understand whats going on here. I personally dont really care what PPD Im getting as long as Im getting the most out of the cards. I understand some work returns less but thats just how the world spins. how do I add the client type advance to my config? Also how can I get a qualified passkey?

Thanks in advance

EDIT

___
ok i added my passkey into F@H. I Also went to edit on my GPU and added slot client-advanced Should I do this on the CPU also ?

Posted: **Sat Jan 03, 2015 1:31 pm**

It's really not necessary to add it to your cpu slot too but if at some time you'd like to see if it makes any difference you can always give it a try.

The rest was typed before I saw your edits:

As long as you entered your passkey using Advanced Control (aka FAHControl) thru the Identity tab and clicked on Save you should be okay for future work and should start receiving QRB after successfully completing 10 qualified work units and have maintained an 80% success rate. It should show up in the config/User Information section of your log as
<passkey v='********************************'/> the next time the client displays your configuration or you can click on Refresh in the Log view and scroll up to the top. Passkeys are usually hidden for security reasons.

Basically you want to try adding client-type advanced to the Extra slot options to see if it helps to receive other core projects like 17. Again you use Advanced Control:

Configure > Slots > double click on gpu slot - click on Add and enter:

Name: client-type
Value: advanced

Click on Ok Ok and Save.

https://fah-web.stanford.edu/projects/F ... l#SlotsTab

If you continue to get core 15 work even with the advanced client-type you can remove it by highlighting it and clicking on Remove.

Posted: **Sat Jan 03, 2015 2:23 pm**

beauty I think its working ill let it fold for a day or so. What is the best way to add this 760GTX into the mix? keep the current drivers? i only have two six + two pins (970 uses 1 six + two) and then one adapter. the 760 uses two six pins. so i basically have to use the other "piggy backed" six pin to power the other card. Would that limit anything? I should probably go and buy another adapter vs using the one ribbon for both cards? Hopefully ya follow me lol

EDIT

Added the second card in and i was upto 140k PPD 70K on each card then the 970 just dropped down to 12K I saw this in the log

Code: Select all

19:56:38:WU01:FS01:0x17:Temperature control disabled. Requirements: single Nvidia GPU, tmax must be < 110 and twait >= 900

its almost like the two cards switched spots in FAH the 760 is getting 70K and the 970 much lower.

Reset all clocks to default and ramped up fan speeds ill let it go thru the motions tonight and tomorrow and report back.

-eddy

Posted: **Sun Jan 04, 2015 1:03 am**

Reports from the client do get interchanged, at times, particularly if you add a new GPU. The total performance should still be right.

Ignore the message about Temperature Control.

Posted: **Sun Jan 04, 2015 8:28 am**

eddyg519 wrote:beauty I think its working ill let it fold for a day or so. What is the best way to add this 760GTX into the mix? keep the current drivers? i only have two six + two pins (970 uses 1 six + two) and then one adapter. the 760 uses two six pins. so i basically have to use the other "piggy backed" six pin to power the other card. Would that limit anything? I should probably go and buy another adapter vs using the one ribbon for both cards? Hopefully ya follow me lol

-SNIP

-eddy

If you are having to use adapters on your PSU cables its usually because your PSU is not very powerful, I would recommend against using any adapters at any time and purchasing a decent PSU of about 650-750w. If you try and pull more power than either the adapter or PSU can cope with you run the risk of damaging components

Posted: **Sun Jan 04, 2015 1:26 pm**

eddyg519 wrote:
its almost like the two cards switched spots in FAH the 760 is getting 70K and the 970 much lower.

Reset all clocks to default and ramped up fan speeds ill let it go thru the motions tonight and tomorrow and report back.

-eddy

When using 9xx cards and 7xx GPU in same rig. This problem occurred all the time. It is not just the fact that it is confusing to users, it is also far from optimal. Your client tells Fah that your 970 is a 7xx card, and the server sends proper WU adapted 7xx cards and the other way around.

It is not a big of a deal if you do not care whether you get optimal PPD. The science vil benefit from your effort anyhow.

Posted: **Sun Jan 04, 2015 1:54 pm**

Sn1ken wrote:
eddyg519 wrote:
its almost like the two cards switched spots in FAH the 760 is getting 70K and the 970 much lower.

Reset all clocks to default and ramped up fan speeds ill let it go thru the motions tonight and tomorrow and report back.

-eddy
When using 9xx cards and 7xx GPU in same rig. This problem occurred all the time. It is not just the fact that it is confusing to users, it is also far from optimal. Your client tells Fah that your 970 is a 7xx card, and the server sends proper WU adapted 7xx cards and the other way around.

It is not a big of a deal if you do not care whether you get optimal PPD. The science vil benefit from your effort anyhow.

That's the indexing of your cards getting mixed up, what F@H thinks is card one may not be what you/your OS thinks is card one. Run a quick search on the forum as there are lots of posts to show you how to change it

Posted: **Mon Jan 05, 2015 8:40 am**

I did some digging and from what I understand I need to just uninstall and remove all data and reinstall the client to hopefully fix the indexing issues?

This PSU is a smart 650w from Thermaltake it seems to be holding up I will be looking into a better one shortly.

I found out I should dedicate a cpu core per GPU? Theres lots of talk about it and its done in the slots but I dont really understand exactly how?

is there anything else i need to do to make this rig perform correctly?

Posted: **Mon Jan 05, 2015 9:15 am**

Yes it does look like an indexing issue. Write down your settings (especially any remote access). Set all WUs to Finish, when they are done uninstall FAH (including data) and reinstall. Reconfigure with your user name and passkey, set up any remote access again. Then see if the GPUs respond correctly. If not you may have to set the indices manually -- it's described somewhere here but I don't have a link offhand.

You will need to experiment to find out whether you need one or two CPU cores to feed your two GPUs. You can configure the CPU slot to use seven or six cores respectively. At present it's seven, if you reinstall and it finds both GPUs, it should set six automatically. But I don't recall anyone posting results as to which setting is best (the differences are almost certainly small and difficult to measure amongst the random variations between WUs).

More fundamentally, with mixes generation GPUs in one PC, one GPU will have a non-optimal driver. The Maxwell card must have a recent driver, but the 760 works better with a 327 (or earlier) driver. Unfortunately that can't be solved until nVidia restore the older 760 performance on a newer driver (and may not be possible with mixed GPUs in any case). The only full solution is to run only one generation of GPU in each PC -- and yes I know that may not be realistic.

As a side note there are exceptions to the expected driver limitations -- 750Ti is a Maxwell despite its designation, 780Ti is not Maxwell, and both are happy with later drivers.

Posted: **Mon Jan 05, 2015 11:01 am**

A reinstall should solve your problem but here are a couple of examples showing how to set indices manually:

viewtopic.php?p=264564#p264564
viewtopic.php?p=199379#p199379

Posted: **Wed Jan 07, 2015 8:24 am**

I played around with this long enough lol Ive came to the conclusion its just not worth it to fold with these two cards. I pretty much get better PPD with just the 970 so im just going to up the psu and buy another one. Ill pick up a couple parts and put the 760 in my old 4100+ OC rig.

I figure its this driver issue. I need the newest drivers for the 970 but they dont seem to work on the older gen 760 so much. no biggie

Thanks for the help guys.

Posted: **Wed Jan 07, 2015 3:02 pm**

I need the newest drivers for the 970 but they dont seem to work on the older gen 760 so much. no biggie

YUP.
Try the 347.09 ...incremental increase in speed on the 970.

However, does not fix OPEN CL bug on core 18 Yet.

Folding Forum

Strix Nvidia 970GTX lowww PPD ? 344.48

Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48

Re: Strix Nvidia 970GTX lowww PPD ? 344.48