7970 making just 35k PPD

[senuketo]

Hey,

Client 7.4.4
Windows 7 Ultimate 64 bit
2Gb RAM DDR3 1600Mhz
Celeron G1620 2.7Ghz dual core
Asrock z77 extreme 3

ATI 7970 XFX DD overclocked
engine 950
mem 1450
undervolted 1,050V

GPU-z showing 95% gpu load
client showing FahCore 0x17
Folding Power on Full
I've paused the CPU: paused by user status

I saw that the card should be getting about 100k PPD and it's getting only 30% of that. I can see that it's trying to crawl up a bit but still far way go.
How much time does it need to show maximum reading for the PPD (top right in the client)?

I read that I need to have at least one CPU core for each GPU. Is that written in stone or an advanced setup could be made for more GPUs?

Is the protein folding algo RAM thirsty?

Will post the log file shortly.

Thanks.

[senuketo]

Code: Select all

*********************** Log Started 2014-05-16T10:31:12Z ***********************
10:31:12:************************* Folding@home Client *************************
10:31:12:      Website: http://folding.stanford.edu/
10:31:12:    Copyright: (c) 2009-2014 Stanford University
10:31:12:       Author: Joseph Coffland <joseph@cauldrondevelopment.com>
10:31:12:         Args: 
10:31:12:       Config: C:/Users/asen2/AppData/Roaming/FAHClient/config.xml
10:31:12:******************************** Build ********************************
10:31:12:      Version: 7.4.4
10:31:12:         Date: Mar 4 2014
10:31:12:         Time: 20:26:54
10:31:12:      SVN Rev: 4130
10:31:12:       Branch: fah/trunk/client
10:31:12:     Compiler: Intel(R) C++ MSVC 1500 mode 1200
10:31:12:      Options: /TP /nologo /EHa /Qdiag-disable:4297,4103,1786,279 /Ox -arch:SSE
10:31:12:               /QaxSSE2,SSE3,SSSE3,SSE4.1,SSE4.2 /Qopenmp /Qrestrict /MT /Qmkl
10:31:12:     Platform: win32 XP
10:31:12:         Bits: 32
10:31:12:         Mode: Release
10:31:12:******************************* System ********************************
10:31:12:          CPU: Intel(R) Celeron(R) CPU G1620 @ 2.70GHz
10:31:12:       CPU ID: GenuineIntel Family 6 Model 58 Stepping 9
10:31:12:         CPUs: 2
10:31:12:       Memory: 1.90GiB
10:31:12:  Free Memory: 1.30GiB
10:31:12:      Threads: WINDOWS_THREADS
10:31:12:   OS Version: 6.1
10:31:12:  Has Battery: false
10:31:12:   On Battery: false
10:31:12:   UTC Offset: 3
10:31:12:          PID: 2328
10:31:12:          CWD: C:/Users/asen2/AppData/Roaming/FAHClient
10:31:12:           OS: Windows 7 Ultimate
10:31:12:      OS Arch: AMD64
10:31:12:         GPUs: 1
10:31:12:        GPU 0: ATI:5 Tahiti XT [Radeon R9 200/HD 7900/8970]
10:31:12:         CUDA: Not detected
10:31:12:Win32 Service: false
10:31:12:***********************************************************************
10:31:12:<config>
10:31:12:  <!-- User Information -->
10:31:12:  <passkey v='********************************'/>
10:31:12:  <team v='224497'/>
10:31:12:  <user v='senuketo'/>
10:31:12:
10:31:12:  <!-- Folding Slots -->
10:31:12:  <slot id='0' type='CPU'>
10:31:12:    <paused v='true'/>
10:31:12:  </slot>
10:31:12:  <slot id='1' type='GPU'/>
10:31:12:</config>
10:31:12:Trying to access database...
10:31:19:Successfully acquired database lock
10:31:19:Enabled folding slot 00: PAUSED cpu:1 (by user)
10:31:19:Enabled folding slot 01: READY gpu:0:Tahiti XT [Radeon R9 200/HD 7900/8970]
10:31:20:WU01:FS01:Starting
10:31:20:WU01:FS01:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" C:/Users/asen2/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 2328 -checkpoint 15 -gpu 0 -gpu-vendor ati
10:31:20:WU01:FS01:Started FahCore on PID 2788
10:31:23:WU01:FS01:Core PID:2780
10:31:23:WU01:FS01:FahCore 0x17 started
10:31:28:WU01:FS01:0x17:*********************** Log Started 2014-05-16T10:31:28Z ***********************
10:31:28:WU01:FS01:0x17:Project: 9408 (Run 897, Clone 0, Gen 22)
10:31:28:WU01:FS01:0x17:Unit: 0x0000001b0a3b1e5c5342e81a0c53bffd
10:31:28:WU01:FS01:0x17:CPU: 0x00000000000000000000000000000000
10:31:28:WU01:FS01:0x17:Machine: 1
10:31:28:WU01:FS01:0x17:Digital signatures verified
10:31:28:WU01:FS01:0x17:Folding@home GPU core17
10:31:28:WU01:FS01:0x17:Version 0.0.52
10:34:22:Saving configuration to config.xml
10:34:22:<config>
10:34:22:  <!-- Slot Control -->
10:34:22:  <power v='full'/>
10:34:22:
10:34:22:  <!-- User Information -->
10:34:22:  <passkey v='********************************'/>
10:34:22:  <team v='224497'/>
10:34:22:  <user v='senuketo'/>
10:34:22:
10:34:22:  <!-- Folding Slots -->
10:34:22:  <slot id='0' type='CPU'>
10:34:22:    <paused v='true'/>
10:34:22:  </slot>
10:34:22:  <slot id='1' type='GPU'/>
10:34:22:</config>
10:34:23:WU01:FS01:0x17:Completed 0 out of 5000000 steps (0%)
10:34:23:WU01:FS01:0x17:Temperature control disabled. Requirements: single Nvidia GPU, tmax must be < 110 and twait >= 900
10:43:48:WU01:FS01:0x17:Completed 50000 out of 5000000 steps (1%)
10:51:17:17:127.0.0.1:New Web connection
10:53:08:WU01:FS01:0x17:Completed 100000 out of 5000000 steps (2%)
11:02:23:WU01:FS01:0x17:Completed 150000 out of 5000000 steps (3%)
11:11:26:WU01:FS01:0x17:Completed 200000 out of 5000000 steps (4%)
11:20:28:WU01:FS01:0x17:Completed 250000 out of 5000000 steps (5%)
11:29:30:WU01:FS01:0x17:Completed 300000 out of 5000000 steps (6%)
11:38:32:WU01:FS01:0x17:Completed 350000 out of 5000000 steps (7%)
11:47:35:WU01:FS01:0x17:Completed 400000 out of 5000000 steps (8%)
11:56:39:WU01:FS01:0x17:Completed 450000 out of 5000000 steps (9%)
11:59:10:44:127.0.0.1:New Web connection
12:05:41:WU01:FS01:0x17:Completed 500000 out of 5000000 steps (10%)

[bollix47]

If you check your TPFs using the bonus calculator you will see that your PPD is up around 85K. I suspect getting 100K or more would require an overclock but doing so could result in failed work units. I do get over 100K PPD on my 7970 but it was factory overclocked to 1100MHz. It has been running for over a year and has been stable.

[senuketo]

thanks bollix
where am I able to see the failed WUs?
do you know how much watt does your single 7970 pull?
I presume 1100 is your engine, and how much is your memory?

[bollix47]

You're welcome.

You can open FAHControl and select the log tab ... then click on Warnings & Errors. That should list any current failures. If you need a specific work unit checked a mod or admin can do that for you if you supply the PRCG numbers as displayed in the log ... they can also check your completion rate which has to be at least 80% to qualify for the Quick Return Bonus. Currently you have successfully completed 6 of the required 10 work units before QRB is applied to your credits.

I do not know how many watts are being used but I'm sure others will be able to answer that question.

Core:1100 Memory:1375 Fan:auto Temperature:68C ... all figures being reported by MSI Afterburner.

Please remember we are not an overclocking support site ... the settings that work on my GPU may not work on yours ... you may get better advice using one of the many sites that discuss overclocking.

[7im]

Overclocking the memory just wastes electricity when Folding. Folding speed is not responsive to faster or larger video memory. Memory speed is not the limiting speed factor with fah on GPUs, although it may improve the speed of your games.

[senuketo]

thanks for the extremely helpful info, guys!
I have so many questions. Isn't there some kind of advanced guide how to optimize your machine like
do you need lots of RAM
can you make 1 CPU core meet the requirements of 2 GPUs
how to clock the cards

[7im]

Don't need lots of ram, so won't be in any guide.
OverClocking cards is covered on a 1000 websites.
It may be possible to have one CPU feed two GPU but that either works or doesn't work with each new driver update or each new fahcore update. You are welcomed to try, and report your results, but everyone's hardware is different and won't work for everyone. There is little need to document was does not help.

The client is already well optimized, and any tips that people can actually use are covered in the install guide and client FAQ.

[senuketo]

Thanks.
I asked about the overclocking because the different algos respond differently to engine clock, mem clock and their ratio.
Let's see what happens

[PantherX]

Please note that different GPU Projects may have different levels of stress so just because you can fold the current GPU Projects with an overclocked GPU without issues, doesn't guarantee that future Projects can fold successfully on that overclocked GPU. Thus, you might have to monitor your GPU especially if it is overclocked.

[senuketo]

Thanks!