R9 280x rig 100% cpu load, low PPD (80k vs 115k)

[viperbite]

Uhg, I typed up a long post and then accidentally hit the back button on my mouse................ I need a hug

Code: Select all

Computer Specs: 
Windows 7 Pro 64bit
Celeron G1820
MSI R9 280x
Gskill 1866 DDR3 8gb
asrock pro btc
Corsair 1200i 1200w psu
USB x1 to x16 risers
120gb SSD

Network Connection: DSL
Operating System: service packs updated but windows update is off.
Overclocked?: Two cards are +20mhz on gpu core. Two are +60mhz.
Stable?:  24/7 stable. Best speed of card was found while litecoin mining. 
Software: 7.4.4 ? 
WU details: No idea what this is 

One of my 4 card R9 280x rigs is steady at 100% cpu usage.
I do not CPU mine (paused the CPU miner)
It is getting around 80k ppd per card instead of the expected 115k ppd. (i let it normalize, this is after at least 24 hours of folding)
I assume it is because they are using 100% of the cpu load and it cannot keep up.
75% GPU usage in GPU-Z

Instead of throwing $100 at a hyperthreading i3 or an i5, i wanted to get someone to look at my log file and see what they think.

Two of my other rigs are on the same celeron and they do not have this same problem. They are folding at the expected PPD per card.

This rig performed flawlessly litecoin mining for a couple months 24/7
The cards are connected to the mobo via USB x1 to x16 powered risers. (i can test standard ribbon risers to rule that out as a problem)
The motherboard is an AsRock PRO BTC, I am not sure if this is a problem, The other rigs i have are gaming motherboards. (all socket 1150, all 8gb ddr3)



below is my log file.

Any info is helpful.
Thanks,

- Casey

Code: Select all

*********************** Log Started 2014-05-18T06:42:03Z ***********************
06:42:03:************************* Folding@home Client *************************
06:42:03:      Website: http://folding.stanford.edu/
06:42:03:    Copyright: (c) 2009-2014 Stanford University
06:42:03:       Author: Joseph Coffland <joseph@cauldrondevelopment.com>
06:42:03:         Args: 
06:42:03:       Config: C:/Users/MINE RIGGER/AppData/Roaming/FAHClient/config.xml
06:42:03:******************************** Build ********************************
06:42:03:      Version: 7.4.4
06:42:03:         Date: Mar 4 2014
06:42:03:         Time: 20:26:54
06:42:03:      SVN Rev: 4130
06:42:03:       Branch: fah/trunk/client
06:42:03:     Compiler: Intel(R) C++ MSVC 1500 mode 1200
06:42:03:      Options: /TP /nologo /EHa /Qdiag-disable:4297,4103,1786,279 /Ox -arch:SSE
06:42:03:               /QaxSSE2,SSE3,SSSE3,SSE4.1,SSE4.2 /Qopenmp /Qrestrict /MT /Qmkl
06:42:03:     Platform: win32 XP
06:42:03:         Bits: 32
06:42:03:         Mode: Release
06:42:03:******************************* System ********************************
06:42:03:          CPU: Intel(R) Celeron(R) CPU G1820 @ 2.70GHz
06:42:03:       CPU ID: GenuineIntel Family 6 Model 60 Stepping 3
06:42:03:         CPUs: 2
06:42:03:       Memory: 7.94GiB
06:42:03:  Free Memory: 7.14GiB
06:42:03:      Threads: WINDOWS_THREADS
06:42:03:   OS Version: 6.1
06:42:03:  Has Battery: false
06:42:03:   On Battery: false
06:42:03:   UTC Offset: -4
06:42:03:          PID: 2504
06:42:03:          CWD: C:/Users/MINE RIGGER/AppData/Roaming/FAHClient
06:42:03:           OS: Windows 7 Professional
06:42:03:      OS Arch: AMD64
06:42:03:         GPUs: 4
06:42:03:        GPU 0: ATI:5 Tahiti XT [Radeon R9 200/HD 7900/8970]
06:42:03:        GPU 1: ATI:5 Tahiti XT [Radeon R9 200/HD 7900/8970]
06:42:03:        GPU 2: ATI:5 Tahiti XT [Radeon R9 200/HD 7900/8970]
06:42:03:        GPU 3: ATI:5 Tahiti XT [Radeon R9 200/HD 7900/8970]
06:42:03:         CUDA: Not detected
06:42:03:Win32 Service: false
06:42:03:***********************************************************************
06:42:03:<config>
06:42:03:  <!-- Slot Control -->
06:42:03:  <power v='full'/>
06:42:03:
06:42:03:  <!-- User Information -->
06:42:03:  <passkey v='********************************'/>
06:42:03:  <team v='224497'/>
06:42:03:  <user v='viperbite'/>
06:42:03:
06:42:03:  <!-- Folding Slots -->
06:42:03:  <slot id='0' type='CPU'>
06:42:03:    <paused v='true'/>
06:42:03:  </slot>
06:42:03:  <slot id='1' type='GPU'/>
06:42:03:  <slot id='2' type='GPU'/>
06:42:03:  <slot id='3' type='GPU'/>
06:42:03:  <slot id='4' type='GPU'/>
06:42:03:</config>
06:42:03:Trying to access database...
06:42:03:Successfully acquired database lock
06:42:03:Enabled folding slot 00: PAUSED cpu:1 (by user)
06:42:03:Enabled folding slot 01: READY gpu:0:Tahiti XT [Radeon R9 200/HD 7900/8970]
06:42:03:Enabled folding slot 02: READY gpu:1:Tahiti XT [Radeon R9 200/HD 7900/8970]
06:42:03:Enabled folding slot 03: READY gpu:2:Tahiti XT [Radeon R9 200/HD 7900/8970]
06:42:03:Enabled folding slot 04: READY gpu:3:Tahiti XT [Radeon R9 200/HD 7900/8970]
06:42:03:WU03:FS01:Starting
06:42:03:WU03:FS01:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" "C:/Users/MINE RIGGER/AppData/Roaming/FAHClient/cores/web.stanford.edu/~pande/Win32/AMD64/ATI/R600/Core_17.fah/FahCore_17.exe" -dir 03 -suffix 01 -version 704 -lifeline 2504 -checkpoint 15 -gpu 0 -gpu-vendor ati
06:42:03:WU03:FS01:Started FahCore on PID 2772
06:42:03:WU03:FS01:Core PID:2724
06:42:03:WU03:FS01:FahCore 0x17 started
06:42:03:WU06:FS04:Starting
06:42:03:WU06:FS04:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" "C:/Users/MINE RIGGER/AppData/Roaming/FAHClient/cores/web.stanford.edu/~pande/Win32/AMD64/ATI/R600/Core_17.fah/FahCore_17.exe" -dir 06 -suffix 01 -version 704 -lifeline 2504 -checkpoint 15 -gpu 3 -gpu-vendor ati
06:42:03:WU06:FS04:Started FahCore on PID 2824
06:42:03:WU06:FS04:Core PID:2832
06:42:03:WU06:FS04:FahCore 0x17 started
06:42:03:WU07:FS02:Starting
06:42:03:WU07:FS02:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" "C:/Users/MINE RIGGER/AppData/Roaming/FAHClient/cores/web.stanford.edu/~pande/Win32/AMD64/ATI/R600/Core_17.fah/FahCore_17.exe" -dir 07 -suffix 01 -version 704 -lifeline 2504 -checkpoint 15 -gpu 1 -gpu-vendor ati
06:42:03:WU07:FS02:Started FahCore on PID 2804
06:42:03:WU07:FS02:Core PID:2856
06:42:03:WU07:FS02:FahCore 0x17 started
06:42:03:WU05:FS03:Starting
06:42:03:WU05:FS03:Running FahCore: "C:\Program Files (x86)\FAHClient/FAHCoreWrapper.exe" "C:/Users/MINE RIGGER/AppData/Roaming/FAHClient/cores/web.stanford.edu/~pande/Win32/AMD64/ATI/R600/Core_17.fah/FahCore_17.exe" -dir 05 -suffix 01 -version 704 -lifeline 2504 -checkpoint 15 -gpu 2 -gpu-vendor ati
06:42:03:WU05:FS03:Started FahCore on PID 1044
06:42:03:WU05:FS03:Core PID:2892
06:42:03:WU05:FS03:FahCore 0x17 started
06:42:03:WU06:FS04:0x17:*********************** Log Started 2014-05-18T06:42:03Z ***********************
06:42:03:WU06:FS04:0x17:Project: 13001 (Run 91, Clone 6, Gen 11)
06:42:03:WU06:FS04:0x17:Unit: 0x00000028538b3db753286996ef556401
06:42:03:WU06:FS04:0x17:CPU: 0x00000000000000000000000000000000
06:42:03:WU06:FS04:0x17:Machine: 4
06:42:03:WU06:FS04:0x17:Digital signatures verified
06:42:03:WU06:FS04:0x17:Folding@home GPU core17
06:42:03:WU06:FS04:0x17:Version 0.0.52
06:42:03:WU07:FS02:0x17:*********************** Log Started 2014-05-18T06:42:03Z ***********************
06:42:03:WU07:FS02:0x17:Project: 13001 (Run 178, Clone 4, Gen 4)
06:42:03:WU07:FS02:0x17:Unit: 0x0000000a538b3db753288236ce513dd8
06:42:03:WU07:FS02:0x17:CPU: 0x00000000000000000000000000000000
06:42:03:WU07:FS02:0x17:Machine: 2
06:42:03:WU07:FS02:0x17:Digital signatures verified
06:42:03:WU07:FS02:0x17:Folding@home GPU core17
06:42:03:WU07:FS02:0x17:Version 0.0.52
06:42:03:WU05:FS03:0x17:*********************** Log Started 2014-05-18T06:42:03Z ***********************
06:42:03:WU05:FS03:0x17:Project: 13001 (Run 135, Clone 7, Gen 23)
06:42:03:WU05:FS03:0x17:Unit: 0x00000031538b3db753287612b6645ded
06:42:03:WU05:FS03:0x17:CPU: 0x00000000000000000000000000000000
06:42:03:WU05:FS03:0x17:Machine: 3
06:42:03:WU05:FS03:0x17:Digital signatures verified
06:42:03:WU05:FS03:0x17:Folding@home GPU core17
06:42:03:WU05:FS03:0x17:Version 0.0.52
06:42:04:WU03:FS01:0x17:*********************** Log Started 2014-05-18T06:42:03Z ***********************
06:42:04:WU03:FS01:0x17:Project: 13001 (Run 252, Clone 7, Gen 13)
06:42:04:WU03:FS01:0x17:Unit: 0x00000020538b3db75328974c94cab1fd
06:42:04:WU03:FS01:0x17:CPU: 0x00000000000000000000000000000000
06:42:04:WU03:FS01:0x17:Machine: 1
06:42:04:WU03:FS01:0x17:Digital signatures verified
06:42:04:WU03:FS01:0x17:Folding@home GPU core17
06:42:04:WU03:FS01:0x17:Version 0.0.52
06:42:04:WU06:FS04:0x17:  Found a checkpoint file
06:42:05:WU07:FS02:0x17:  Found a checkpoint file
06:42:05:WU03:FS01:0x17:  Found a checkpoint file
06:42:06:WU05:FS03:0x17:  Found a checkpoint file
06:45:53:WU05:FS03:0x17:Completed 4625000 out of 5000000 steps (92%)
06:45:53:WU05:FS03:0x17:Temperature control disabled. Requirements: single Nvidia GPU, tmax must be < 110 and twait >= 900
06:48:18:WU07:FS02:0x17:Completed 4500000 out of 5000000 steps (90%)
06:48:18:WU07:FS02:0x17:Temperature control disabled. Requirements: single Nvidia GPU, tmax must be < 110 and twait >= 900
06:48:33:WU03:FS01:0x17:Completed 4375000 out of 5000000 steps (87%)
06:48:34:WU03:FS01:0x17:Temperature control disabled. Requirements: single Nvidia GPU, tmax must be < 110 and twait >= 900
06:48:56:WU06:FS04:0x17:Completed 4375000 out of 5000000 steps (87%)
06:48:56:WU06:FS04:0x17:Temperature control disabled. Requirements: single Nvidia GPU, tmax must be < 110 and twait >= 900
06:52:42:WU05:FS03:0x17:Completed 4650000 out of 5000000 steps (93%)
06:53:27:WU03:FS01:0x17:Completed 4400000 out of 5000000 steps (88%)
06:53:42:WU06:FS04:0x17:Completed 4400000 out of 5000000 steps (88%)
06:57:57:WU07:FS02:0x17:Completed 4550000 out of 5000000 steps (91%)
07:01:39:WU05:FS03:0x17:Completed 4700000 out of 5000000 steps (94%)

[ChristianVirtual]

normally each GPU card need one CPU core, too. That would mean you run a bit in an overallocation; seems still ok as per your other systems.

Your MB has 1 x PCIe 2.0 x16, 5 x PCIe 2.0 x1; x1 is a performance blocker; I would swap that board with something else (or accept the lower throughput)

[viperbite]

Thanks for the quick reply.

I wish my original was more thorough. But alas, the back button on my mouse betrays me.

I hate to just write the problem off as throughput on x1 as 3 of my other rigs do not appear to be suffering. Every card I have in folding rigs are using X1 slots...

My other R9 280x rig is doing between 110k to 120k per card. :/


Though, for some reason I really want to blame the AsRock pro btc motherboard because it is the only difference between my other rigs...
are there different projects that end up utilizing more CPU usage than others?
All of my other rigs are using less than 30% of either that same G1820 celeron or a pentium.

[bollix47]

Please check your PPD using the frame times in your log and the online Bonus Point Calculator. I did check one of them from the log you posted and the PPD was around 109K but with the restarts in your log it might not be totally accurate. Let them run for a few frames each before calculating the TPF and use that in the points calculator.

[viperbite]

Ah ok.

Thanks, I will try to figure that calculator out right now.

[ChristianVirtual]

... I really want to blame the AsRock pro btc motherboard because it is the only difference between my other rigs...

you still could blame the board. Maybe try one of your cards in the x16 slot and see if that one is different in PPD. I bet it will be. The implementation of PCIe on this specific MB and the used chipset would easy be good for a difference.

But also bollix47 suggestion is worth looking for (and might help already, too).

[viperbite]

I just calculated my PPD using the calculator. It ended up being around 100k on the several different instances I checked on 2 of the cards.

I then noticed that one of the cards had started new a project and had low ppd estimate. The other 3 are cruising along at 100k ppd.

CPU still 100%

But the GPU's are all around 96% now....

I guess I just need to watch this rig for a few more days and see how it does.

[PantherX]

Welcome to the F@H Forum viperbite,

Sorry to hear about your lost post.

While performing a comparison of PPD across identical GPUs, please note that the Project number should be the same to get a reasonable comparison. Moreover, if the Project is the same, some WUs may still exhibit PPD variation due to the varying positions of the atoms and their forces. This is something that happens in protein folding so you may occasionally see a "low PPD" WU. Sometimes, if the variation in PPD is extreme, the researcher may consider splitting up the original Project into multiple Projects to keep the variation in PPD to a reasonable amount (viewtopic.php?p=263454#p263454).

[viperbite]

Thanks!

There is a big difference between excitement and knowledge!

I am excited to be folding, and probably obsessed with efficiency.

I appreciate all the help and info

[PantherX]

What drivers are you using? Do note that the latest AMD 14.4 Drivers may provide a nice boost in PPD (https://folding.stanford.edu/home/new-a ... rformance/).

Regarding the PCI-E lanes, you may find this post helpful (viewtopic.php?p=254816#p254816).

Regarding the CPU Usage, generally, AMD GPUs don't require a dedicated CPU per FahCore_17 instance. Instead, they require some CPU cycles during the initial start-up and checkpoints. However, this may change at any time with the driver version. To see if the CPU is being a bottleneck, what you can do is to pause all GPUs except one and see how the CPU Usage and the GPU's PPD/TPF changes (use the log file for accurate results). If no changes occur for the GPU, start-up another GPU. Continue until you see a change. If you notice it, then you have found the bottleneck. In your case, it could a combination of CPU and PCI-E lanes.

BTW, if case you dont know what WU means, you can have a look at this post (viewtopic.php?p=261080&f=24#p261080).

[viperbite]

Thanks PantherX

I am using driver version 13.12 right now. I think I will try out 14.4.

I am going to try the troubleshooting method you suggested.

After reading the post by nookiebandit, Pending the outcome of troubleshooting a suspected cpu bottleneck, I may also end up upgrading this computer to a motherboard that supports PCI-E 3.0 and just continue using 1x risers.


edit

I have a basic idea of what WU is but folding is still new to me. Picking it up as I go

[viperbite]

For those of you who have been helping

Forced bios to select pci-e 3.0 (gpu-z still sees it as 1.1 or 2.0 depending on load) and updated drivers to 14.4

The rig in question is now at 127 to 132ppd per card!

Still going to test for bottleneck on CPU next.

Thanks for everyone's help!!!!

[bruce]

There are many, many reports where folks were running both GPU and CPU (SMP) folding. We're certain that if all of your CPU threads are devoted to Folding and/or other CPU-intensive processing plus the GPU driver needed CPU processing (over-committed), this will adversely affect the CPU/SMP folding very significantly.

Until recently, there were a relatively small number of reports from people with multiple GPUs. If the GPU drivers or FahCore can't get all of the CPU resources that it wants, it stands to reason that GPU folding will be degraded, other factors being equal. (With different speed slots, other factors re NOT equal.) We don't have a lot of quantitative reports. Please let us know whatever you find.

[viperbite]

There are many, many reports where folks were running both GPU and CPU (SMP) folding. We're certain that if all of your CPU threads are devoted to Folding and/or other CPU-intensive processing plus the GPU driver needed CPU processing (over-committed), this will adversely affect the CPU/SMP folding very significantly.

Until recently, there were a relatively small number of reports from people with multiple GPUs. If the GPU drivers or FahCore can't get all of the CPU resources that it wants, it stands to reason that GPU folding will be degraded, other factors being equal. (With different speed slots, other factors re NOT equal.) We don't have a lot of quantitative reports. Please let us know whatever you find.

Thank you,

I have done some random calculations using the PPD calculator. These calculations gave me the results I expected per card on all of my rigs. Since, I have learned to just look at TPF and what project I am on to determine if my PPD was near what it should be.

Additionally, I seem to have determined my CPU bottleneck theory was unfounded.

I have an old Opteron 185 socket 939 AMD processor inside an asus a8n-sli premium.
4gb ddr1
650w 80+ silver antec psu
2.5" notebook 40gb IDE HDD
(One) R9 270x. O/C to 1125 core / 1500 ram
I am getting 80,000ppd
TPF is ~11:05 on project 13000


I will compile my results for all of my individual rigs more thoroughly when I am a little less sleep deprived