Sub-par GTX970 Linux performance - CUDA Not Detected?

Skufer · Post by **Skufer** » Mon Apr 27, 2015 5:38 pm

I just bought an Nvidia GTX 970 to run on an old PC running Linux Mint 17.1 Rebecca, the CPU is weak but as I'm not running any CPU client I figured it would be okay?

Anyway, the PPD is quite a bit lower than it should be for this card (~181,000) and I am concerned that the FAHClient system info states CUDA is "Not detected". I installed the latest Nvidia proprietary drivers from the CLI, version 349.16. Can anyone please suggest a reason why my GPU is underperforming here?

Code: Select all

17:01:26:Adding folding slot 00: READY gpu:0:GM204 [GeForce GTX 970]
17:01:26:Saving configuration to /etc/fahclient/config.xml
17:01:26:<config>
17:01:26:  <!-- Network -->
17:01:26:  <proxy v=':8080'/>
17:01:26:
17:01:26:  <!-- Slot Control -->
17:01:26:  <power v='full'/>
17:01:26:
17:01:26:  <!-- User Information -->
17:01:26:  <passkey v='********************************'/>
17:01:26:  <team v='212997'/>
17:01:26:  <user v='BestPony'/>
17:01:26:
17:01:26:  <!-- Folding Slots -->
17:01:26:  <slot id='0' type='GPU'/>
17:01:26:</config>
17:01:27:WU00:FS00:Connecting to 171.67.108.200:80
17:01:28:WU00:FS00:Assigned to work server 171.67.108.52
17:01:28:WU00:FS00:Requesting new work unit for slot 00: READY gpu:0:GM204 [GeForce GTX 970] from 171.67.108.52
17:01:28:WU00:FS00:Connecting to 171.67.108.52:8080
17:01:28:WU00:FS00:Downloading 1.52MiB
17:01:31:WU00:FS00:Download complete
17:01:31:WU00:FS00:Received Unit: id:00 state:DOWNLOAD error:NO_ERROR project:9201 run:229 clone:2 gen:347 core:0x17 unit:0x000001e06652edc45399df06b9897d95
17:01:31:WU00:FS00:Downloading core from http://web.stanford.edu/~pande/Linux/AMD64/NVIDIA/Fermi/Core_17.fah
17:01:31:WU00:FS00:Connecting to web.stanford.edu:80
17:01:35:WU00:FS00:FahCore 17: Downloading 3.01MiB
17:01:41:WU00:FS00:FahCore 17: 72.66%
17:01:43:WU00:FS00:FahCore 17: Download complete
17:01:43:WU00:FS00:Valid core signature
17:01:43:WU00:FS00:Unpacked 8.16MiB to cores/web.stanford.edu/~pande/Linux/AMD64/NVIDIA/Fermi/Core_17.fah/FahCore_17
17:01:43:WU00:FS00:Starting
17:01:43:WU00:FS00:Running FahCore: /usr/bin/FAHCoreWrapper /var/lib/fahclient/cores/web.stanford.edu/~pande/Linux/AMD64/NVIDIA/Fermi/Core_17.fah/FahCore_17 -dir 00 -suffix 01 -version 704 -lifeline 1360 -checkpoint 15 -gpu 0 -gpu-vendor nvidia
17:01:43:WU00:FS00:Started FahCore on PID 4478
17:01:43:WU00:FS00:Core PID:4482
17:01:43:WU00:FS00:FahCore 0x17 started
17:01:43:WU00:FS00:0x17:*********************** Log Started 2015-04-27T17:01:43Z ***********************
17:01:43:WU00:FS00:0x17:Project: 9201 (Run 229, Clone 2, Gen 347)
17:01:43:WU00:FS00:0x17:Unit: 0x000001e06652edc45399df06b9897d95
17:01:43:WU00:FS00:0x17:CPU: 0x00000000000000000000000000000000
17:01:43:WU00:FS00:0x17:Machine: 0
17:01:43:WU00:FS00:0x17:Reading tar file state.xml
17:01:43:WU00:FS00:0x17:Reading tar file system.xml
17:01:43:WU00:FS00:0x17:Reading tar file integrator.xml
17:01:43:WU00:FS00:0x17:Reading tar file core.xml
17:01:43:WU00:FS00:0x17:Digital signatures verified
17:02:00:Saving configuration to /etc/fahclient/config.xml
17:02:00:<config>
17:02:00:  <!-- Network -->
17:02:00:  <proxy v=':8080'/>
17:02:00:
17:02:00:  <!-- Slot Control -->
17:02:00:  <power v='full'/>
17:02:00:
17:02:00:  <!-- User Information -->
17:02:00:  <passkey v='********************************'/>
17:02:00:  <team v='212997'/>
17:02:00:  <user v='BestPony'/>
17:02:00:
17:02:00:  <!-- Folding Slots -->
17:02:00:  <slot id='0' type='GPU'/>
17:02:00:</config>
17:02:19:WU00:FS00:0x17:Completed 0 out of 5000000 steps (0%)
17:04:44:WU00:FS00:0x17:Completed 50000 out of 5000000 steps (1%)
17:07:11:WU00:FS00:0x17:Completed 100000 out of 5000000 steps (2%)
17:09:38:WU00:FS00:0x17:Completed 150000 out of 5000000 steps (3%)
17:12:05:WU00:FS00:0x17:Completed 200000 out of 5000000 steps (4%)
17:14:32:WU00:FS00:0x17:Completed 250000 out of 5000000 steps (5%)
17:16:59:WU00:FS00:0x17:Completed 300000 out of 5000000 steps (6%)
17:19:26:WU00:FS00:0x17:Completed 350000 out of 5000000 steps (7%)
17:21:54:WU00:FS00:0x17:Completed 400000 out of 5000000 steps (8%)
17:24:21:WU00:FS00:0x17:Completed 450000 out of 5000000 steps (9%)
17:26:48:WU00:FS00:0x17:Completed 500000 out of 5000000 steps (10%)
17:29:15:WU00:FS00:0x17:Completed 550000 out of 5000000 steps (11%)

Code: Select all

PyON 1 info
[
  [
    "Folding@home Client",
    ["Website", "http://folding.stanford.edu/"],
    ["Copyright", "(c) 2009-2014 Stanford University"],
    ["Author", "Joseph Coffland <joseph@cauldrondevelopment.com>"],
    ["Args", " --child --lifeline 1358 /etc/fahclient/config.xml --run-as fahclient --pid-file=/var/run/fahclient.pid --daemon"],
    ["Config", "/etc/fahclient/config.xml"]
  ],
  [
    "Build",
    ["Version", "7.4.4"],
    ["Date", "Mar  4 2014"],
    ["Time", "12:02:38"],
    ["SVN Rev", "4130"],
    ["Branch", "fah/trunk/client"],
    ["Compiler", "GNU 4.4.7"],
    ["Options", "-std=gnu++98 -O3 -funroll-loops -mfpmath=sse -ffast-math -fno-unsafe-math-optimizations -msse2"],
    ["Platform", "linux2 3.2.0-1-amd64"],
    ["Bits", "64"],
    ["Mode", "Release"]
  ],
  [
    "System",
    ["CPU", "Intel(R) Core(TM)2 Duo CPU     E6750  @ 2.66GHz"],
    ["CPU ID", "GenuineIntel Family 6 Model 15 Stepping 11"],
    ["CPUs", "2"],
    ["Memory", "1.92GiB"],
    ["Free Memory", "1.70GiB"],
    ["Threads", "POSIX_THREADS"],
    ["OS Version", "3.13"],
    ["Has Battery", "false"],
    ["On Battery", "false"],
    ["UTC Offset", "1"],
    ["PID", "1360"],
    ["CWD", "/var/lib/fahclient"],
    ["OS", "Linux 3.13.0-37-generic x86_64"],
    ["OS Arch", "AMD64"],
    ["GPUs", "1"],
    ["GPU 0", "NVIDIA:5 GM204 [GeForce GTX 970]"],
    ["CUDA", "Not detected"]
  ]
]

Thank you for your support as always

Post by **bruce** » Mon Apr 27, 2015 6:07 pm

You were assigned a WU for Core_17. It uses OpenCL instead of CUDA.

Some cores have a CUDA version; some do not. Your client notifies the Servers whether or not you have CUDA support and it won't send you something you can't complete.

Recent Linux proprietary drivers seem to include CUDA; older versions did not. Either way, whoever packaged your drivers might have included it or not. I don't remember which ones fall into those categories.

When drivers do not include CUDA, you should be able to install the CUDA developer's package, though it includes a lot of extras that you won't use -- and as I said, OpenCL will allow you to fold, though not necessarily for as many PPD.

Skufer · Post by **Skufer** » Mon Apr 27, 2015 9:59 pm

I had followed the guide here almost exactly to the letter viewtopic.php?f=80&t=27040 and used the package NVIDIA-Linux-x86_64-346.59.run to install it.

It didn't appear to have included the CUDA developers package, nvcc --version showed nothing was installed. Upon trying to install the CUDA developers package my whole install is bricked and I will be wiping the PC and starting again shortly.

I don't quite understand what you meant about OpenCL, does Core_17 use it instead of CUDA when CUDA is not available? And how can I install the required CUDA components without bricking the PC again?

davidcoton · Post by **davidcoton** » Mon Apr 27, 2015 11:39 pm

BestPony wrote:
I don't quite understand what you meant about OpenCL, does Core_17 use it instead of CUDA when CUDA is not available? And how can I install the required CUDA components without bricking the PC again?

Core_17 uses OpenCL instead of CUDA. even if CUDA is available. This makes maintenance of nVidia and AMD versions simpler.
Bruce seems to have a loose mouse on his keyboard.

AFAIK the CUDA (non-developer) parts are installed by the nVidia installer -- it may be a non-default option. I haven't done a Linux install recently, and when I did it was Ubuntu not Mint, quite an old version because my Linux system has a pre-Maxwell GPU. You shouldn't need the developer package for FaH, but I don't know why it might brick you system if you need it for another reason.

If you have problems with the rebuild, please post the first parts of the log (see link on Bruce's sig, section 6.)

Skufer · Post by **Skufer** » Tue Apr 28, 2015 12:56 pm

I've re-build the entire PC, using Ubuntu 14.04 LTS this time around, same Nvidia drivers.

I'm getting even fewer PPD than I was before, could anyone please clarify the following?

- Would the power of the CPU have an effect on the efficiency of my GPU?

- Would the fact that my GPU is plugged into a PCI-E 1.0 x1 lane affect it's efficiency to fold?

Up till now I didn't think that either such bottlenecks would really have an impact on folding, but I'm beginning to have my doubts...

Code: Select all

*********************** Log Started 2015-04-28T12:37:30Z ***********************
12:37:30:************************* Folding@home Client *************************
12:37:30:    Website: http://folding.stanford.edu/
12:37:30:  Copyright: (c) 2009-2014 Stanford University
12:37:30:     Author: Joseph Coffland <joseph@cauldrondevelopment.com>
12:37:30:       Args: --child --lifeline 1132 /etc/fahclient/config.xml --run-as
12:37:30:             fahclient --pid-file=/var/run/fahclient.pid --daemon
12:37:30:     Config: /etc/fahclient/config.xml
12:37:30:******************************** Build ********************************
12:37:30:    Version: 7.4.4
12:37:30:       Date: Mar 4 2014
12:37:30:       Time: 12:02:38
12:37:30:    SVN Rev: 4130
12:37:30:     Branch: fah/trunk/client
12:37:30:   Compiler: GNU 4.4.7
12:37:30:    Options: -std=gnu++98 -O3 -funroll-loops -mfpmath=sse -ffast-math
12:37:30:             -fno-unsafe-math-optimizations -msse2
12:37:30:   Platform: linux2 3.2.0-1-amd64
12:37:30:       Bits: 64
12:37:30:       Mode: Release
12:37:30:******************************* System ********************************
12:37:30:        CPU: Intel(R) Core(TM)2 Duo CPU E6750 @ 2.66GHz
12:37:30:     CPU ID: GenuineIntel Family 6 Model 15 Stepping 11
12:37:30:       CPUs: 2
12:37:30:     Memory: 1.92GiB
12:37:30:Free Memory: 1.69GiB
12:37:30:    Threads: POSIX_THREADS
12:37:30: OS Version: 3.16
12:37:30:Has Battery: false
12:37:30: On Battery: false
12:37:30: UTC Offset: 1
12:37:30:        PID: 1134
12:37:30:        CWD: /var/lib/fahclient
12:37:30:         OS: Linux 3.16.0-30-generic x86_64
12:37:30:    OS Arch: AMD64
12:37:30:       GPUs: 1
12:37:30:      GPU 0: NVIDIA:5 GM204 [GeForce GTX 970]
12:37:30:       CUDA: 5.2
12:37:30:CUDA Driver: 7000
12:37:30:***********************************************************************
12:37:30:<config>
12:37:30:  <!-- Folding Slot Configuration -->
12:37:30:  <client-type v='advanced'/>
12:37:30:
12:37:30:  <!-- Network -->
12:37:30:  <proxy v=':8080'/>
12:37:30:
12:37:30:  <!-- Slot Control -->
12:37:30:  <power v='full'/>
12:37:30:
12:37:30:  <!-- User Information -->
12:37:30:  <passkey v='********************************'/>
12:37:30:  <team v='212997'/>
12:37:30:  <user v='BestPony'/>
12:37:30:
12:37:30:  <!-- Folding Slots -->
12:37:30:</config>
12:37:30:Switching to user fahclient
12:37:30:Trying to access database...
12:37:30:Successfully acquired database lock
12:37:30:Enabled folding slot 00: READY cpu:1
12:37:30:Enabled folding slot 01: READY gpu:0:GM204 [GeForce GTX 970]
12:37:30:WU00:FS00:Connecting to 171.67.108.200:8080
12:37:34:WU01:FS01:Connecting to 171.67.108.200:80
12:37:35:WU00:FS00:Assigned to work server 171.64.65.124
12:37:35:WU00:FS00:Requesting new work unit for slot 00: READY cpu:1 from 171.64.65.124
12:37:35:WU00:FS00:Connecting to 171.64.65.124:8080
12:37:35:WU01:FS01:Assigned to work server 171.64.65.56
12:37:35:WU01:FS01:Requesting new work unit for slot 01: READY gpu:0:GM204 [GeForce GTX 970] from 171.64.65.56
12:37:35:WU01:FS01:Connecting to 171.64.65.56:8080
12:37:36:WU01:FS01:Downloading 889.95KiB
12:37:36:WU00:FS00:Downloading 808.42KiB
12:37:38:WU00:FS00:Download complete
12:37:38:WU00:FS00:Received Unit: id:00 state:DOWNLOAD error:NO_ERROR project:9007 run:1236 clone:2 gen:160 core:0xa4 unit:0x000000c6664f2de453d81999b4a631d5
12:37:38:WU01:FS01:Download complete
12:37:39:WU00:FS00:Starting
12:37:39:WU00:FS00:Running FahCore: /usr/bin/FAHCoreWrapper /var/lib/fahclient/cores/web.stanford.edu/~pande/Linux/AMD64/Core_a4.fah/FahCore_a4 -dir 00 -suffix 01 -version 704 -lifeline 1134 -checkpoint 15
12:37:39:WU00:FS00:Started FahCore on PID 1376
12:37:40:WU00:FS00:Core PID:1409
12:37:40:WU00:FS00:FahCore 0xa4 started
12:37:41:WU01:FS01:Received Unit: id:01 state:DOWNLOAD error:NO_ERROR project:9411 run:2060 clone:0 gen:40 core:0x17 unit:0x00000031ab40413854d27e3eebc730d3
12:37:41:WU01:FS01:Downloading core from http://web.stanford.edu/~pande/Linux/AMD64/NVIDIA/Fermi/Core_17.fah
12:37:41:WU01:FS01:Connecting to web.stanford.edu:80
12:37:41:WU00:FS00:0xa4:
12:37:41:WU00:FS00:0xa4:*------------------------------*
12:37:41:WU00:FS00:0xa4:Folding@Home Gromacs GB Core
12:37:41:WU00:FS00:0xa4:Version 2.27 (Dec. 15, 2010)
12:37:41:WU00:FS00:0xa4:
12:37:41:WU00:FS00:0xa4:Preparing to commence simulation
12:37:41:WU00:FS00:0xa4:- Looking at optimizations...
12:37:41:WU00:FS00:0xa4:- Created dyn
12:37:41:WU00:FS00:0xa4:- Files status OK
12:37:41:WU00:FS00:0xa4:- Expanded 827306 -> 1400900 (decompressed 169.3 percent)
12:37:41:WU00:FS00:0xa4:Called DecompressByteArray: compressed_data_size=827306 data_size=1400900, decompressed_data_size=1400900 diff=0
12:37:41:WU00:FS00:0xa4:- Digital signature verified
12:37:41:WU00:FS00:0xa4:
12:37:41:WU00:FS00:0xa4:Project: 9007 (Run 1236, Clone 2, Gen 160)
12:37:41:WU00:FS00:0xa4:
12:37:41:WU00:FS00:0xa4:Assembly optimizations on if available.
12:37:41:WU00:FS00:0xa4:Entering M.D.
12:37:42:WU01:FS01:FahCore 17: Downloading 3.01MiB
12:37:48:WU01:FS01:FahCore 17: 85.12%
12:37:49:WU01:FS01:FahCore 17: Download complete
12:37:49:WU01:FS01:Valid core signature
12:37:49:WU01:FS01:Unpacked 8.16MiB to cores/web.stanford.edu/~pande/Linux/AMD64/NVIDIA/Fermi/Core_17.fah/FahCore_17
12:37:50:WU01:FS01:Starting
12:37:50:WU01:FS01:Running FahCore: /usr/bin/FAHCoreWrapper /var/lib/fahclient/cores/web.stanford.edu/~pande/Linux/AMD64/NVIDIA/Fermi/Core_17.fah/FahCore_17 -dir 01 -suffix 01 -version 704 -lifeline 1134 -checkpoint 15 -gpu 0 -gpu-vendor nvidia
12:37:50:WU01:FS01:Started FahCore on PID 1972
12:37:50:WU01:FS01:Core PID:1976
12:37:50:WU01:FS01:FahCore 0x17 started
12:37:50:WU01:FS01:0x17:*********************** Log Started 2015-04-28T12:37:50Z ***********************
12:37:50:WU01:FS01:0x17:Project: 9411 (Run 2060, Clone 0, Gen 40)
12:37:50:WU01:FS01:0x17:Unit: 0x00000031ab40413854d27e3eebc730d3
12:37:50:WU01:FS01:0x17:CPU: 0x00000000000000000000000000000000
12:37:50:WU01:FS01:0x17:Machine: 1
12:37:50:WU01:FS01:0x17:Reading tar file state.xml
12:37:50:WU01:FS01:0x17:Reading tar file system.xml
12:37:50:WU01:FS01:0x17:Reading tar file integrator.xml
12:37:50:WU01:FS01:0x17:Reading tar file core.xml
12:37:50:WU01:FS01:0x17:Digital signatures verified
12:37:50:WU00:FS00:0xa4:Completed 0 out of 250000 steps  (0%)
12:38:20:Saving configuration to /etc/fahclient/config.xml
12:38:20:<config>
12:38:20:  <!-- Folding Slot Configuration -->
12:38:20:  <client-type v='advanced'/>
12:38:20:
12:38:20:  <!-- Network -->
12:38:20:  <proxy v=':8080'/>
12:38:20:
12:38:20:  <!-- Slot Control -->
12:38:20:  <power v='full'/>
12:38:20:
12:38:20:  <!-- User Information -->
12:38:20:  <passkey v='********************************'/>
12:38:20:  <team v='212997'/>
12:38:20:  <user v='BestPony'/>
12:38:20:
12:38:20:  <!-- Folding Slots -->
12:38:20:  <slot id='1' type='GPU'/>
12:38:20:</config>
12:38:20:FS00:Shutting core down
12:38:21:WU00:FS00:FahCore returned: INTERRUPTED (102 = 0x66)
12:38:21:WARNING:WU00:Slot ID 0 no longer exists and there are no other matching slots, dumping
12:38:21:WU00:Sending unit results: id:00 state:SEND error:DUMPED project:9007 run:1236 clone:2 gen:160 core:0xa4 unit:0x000000c6664f2de453d81999b4a631d5
12:38:21:WU00:Connecting to 171.64.65.124:8080
12:38:22:WU00:Server responded WORK_ACK (400)
12:38:22:WU00:Cleaning up
12:38:23:WU01:FS01:0x17:Completed 0 out of 16000000 steps (0%)
12:38:31:Saving configuration to /etc/fahclient/config.xml
12:38:31:<config>
12:38:31:  <!-- Folding Slot Configuration -->
12:38:31:  <client-type v='advanced'/>
12:38:31:
12:38:31:  <!-- Network -->
12:38:31:  <proxy v=':8080'/>
12:38:31:
12:38:31:  <!-- Slot Control -->
12:38:31:  <power v='full'/>
12:38:31:
12:38:31:  <!-- User Information -->
12:38:31:  <passkey v='********************************'/>
12:38:31:  <team v='212997'/>
12:38:31:  <user v='BestPony'/>
12:38:31:
12:38:31:  <!-- Folding Slots -->
12:38:31:  <slot id='1' type='GPU'/>
12:38:31:</config>
12:44:29:WU01:FS01:0x17:Completed 160000 out of 16000000 steps (1%)
12:50:28:WU01:FS01:0x17:Completed 320000 out of 16000000 steps (2%)
12:56:25:WU01:FS01:0x17:Completed 480000 out of 16000000 steps (3%)

7im · Post by **7im** » Tue Apr 28, 2015 1:27 pm

Yes, the older PCIe version and x1 slot will bottleneck performance. I speak from personal experience.

Skufer · Post by **Skufer** » Tue Apr 28, 2015 2:09 pm

Damn, that's a pain, I had thought that the data throughput required for folding wouldn't be all that much.

I do have a PC with PCI-E 2.0 which I was hoping not to have to use for this, but it would still be connected via a x1 connector. Would folding on a GTX 970 really need more that 4 Gbit/s?

7im · Post by **7im** » Tue Apr 28, 2015 4:06 pm

PCIe 2.0 x16 slot with a x1 bandwidth slowed my GTX 760 about 10-15% in PPD. x4 or above is the minimum, IMO.

Skufer · Post by **Skufer** » Tue Apr 28, 2015 5:01 pm

According to nvidia-settings I am only using a mere 22% of the available bandwidth, this makes me think that the problem lies more with the CPU:

Can this figure be trusted?

Post by **bruce** » Tue Apr 28, 2015 6:13 pm

FAH doesn't do anything all the time. It spends some time waiting for the GPU to finish processing some data, it spends some time waiting for data to be transferred, it spends some time waiting for the CPU do do things, it may spend some time waiting for the internet to download/upload data, etc. The calculations involved in waiting are complicated because sometimes it's doing more than one thing at a time and sometimes it's not.

Suppose it spends 100% of it's time computing on a very fast PCIe connection. Now suppose it spends 22% more time waiting on a slower data transfer. It now takes as much as 122% as long to finish the WU so PPD can go down as much as 1 - 1/1.22 or 18% slower. It might not be that bad, but it also might be, especially if the bandwidth is shared with screen updates or whatever. It's a very statistical result so 7im statement that he sees about 10-15% loss in PPD is probably realistic. Moreover, your GTX 970 probably spends less time processing bigger WUs at a faster rate, making the bandwidth a more important factor than it is for his GTX 760. Faster GPUs are always more dependent on fast data transfers.

Skufer · Post by **Skufer** » Tue Apr 28, 2015 7:19 pm

7im wrote:PCIe 2.0 x16 slot with a x1 bandwidth slowed my GTX 760 about 10-15% in PPD. x4 or above is the minimum, IMO.

Sorry 7im, I missed this post you made. And thank you for the advice so far too Bruce, to give you a bit of the backstory, I am running this GTX 970 in an external chassis connected to the PC via an x1 PCI-E connector. Would you agree that at this stage it is a bit of a lost cause to think that I can get the most out of my GPU in this circumstance regardless of what revision of PCI-E I'm running on?

I suspect I might be flogging a dead horse.

7im · Post by **7im** » Tue Apr 28, 2015 7:48 pm

Mine was on a Core 2 Duo at 2.0 GHz. With a CPU core dedicated to GPU folding, and not CPU folding, the CPU speed was not the bottleneck.

IIRC, PCIe 2.0 should be faster than 1.0, even on a 1x connection.

Post by **bruce** » Tue Apr 28, 2015 7:53 pm

The most recent WUs from each of your slots are:
1332.87 p9007 (1375.3,47)
2417.52 p6398 (74,18,2)
A: 0 p9201 (981,3,340)
B: 0 p9411 (1789,0,78)
C: 0 p9201 (906,3,307)

The first two got bonus points but the last three were received with an error indication for 0 points.

WU A was reassigned and another person completed it successfully.
WU B has probably been reassigned, but there's no second return yet.
WU C has been reassigned twice. Two failed and one was completed successfully.

I'm not sure if this is useful, but it does indicate some kind of a problem.

Skufer · Post by **Skufer** » Tue Apr 28, 2015 7:58 pm

I think that's probably my fault, getting F@H running then messing with drivers, breaking my Linux install part way through. On several occasions the slot has failed because of this.

At this point in time I'm only getting around 170k PPD but the WU will finish. Could you please give your opinion of whether the PCIE x1 connector is a show stopper? And also where I can see my recent WU's like you just found? Thanks again.

Post by **bruce** » Tue Apr 28, 2015 8:26 pm

Officially, PCIe 1x is supposed to work.

Personally, I would never put a recent generation GPU on one, but I might consider something old and slow.

Folding Forum

Sub-par GTX970 Linux performance - CUDA Not Detected?

Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?

Re: Sub-par GTX970 Linux performance - CUDA Not Detected?