It seems that a lot of GPU problems revolve around specific versions of drivers. Though NVidia has their own support structure, you can often learn from information reported by others who fold.
Good afternoon all,
I recently had my old amd radon card die and i replaced it with a gtx 1650 card. I uninstalled the old amd items and installed the latest info for nvidia but still have issues with getting the gpu to fold
these are my logs.
00:46:42:Saving configuration to config.xml
00:46:42:<config>
00:46:42: <!-- Network -->
00:46:42: <proxy v=':8080'/>
00:46:42:
00:46:42: <!-- User Information -->
00:46:42: <team v='12233'/>
00:46:42: <user v='Paranoid56'/>
00:46:42:
00:46:42: <!-- Folding Slots -->
00:46:42: <slot id='0' type='CPU'/>
00:46:42: <slot id='1' type='GPU'/>
00:46:42:</config>
00:47:37:WU02:FS00:0xa7:Completed 15000 out of 125000 steps (12%)
00:49:13:WU02:FS00:0xa7:Completed 16250 out of 125000 steps (13%)
00:49:29:WU00:FS01:Starting
00:49:29:ERROR:WU00:FS01:Failed to start core: OpenCL device matching slot 1 not found, try setting 'opencl-index' manually
00:49:29:WU00:FS01:Sending unit results: id:00 state:SEND error:FAILED project:14163 run:44 clone:23 gen:92 core:0x21 unit:0x000000870002894c5c38bfb6b6daf475
00:49:29:WU00:FS01:Connecting to 155.247.166.220:8080
00:49:29:WU00:FS01:Server responded WORK_ACK (400)
00:49:29:WU00:FS01:Cleaning up
00:49:29:WU01:FS01:Connecting to 65.254.110.245:8080
00:49:30:WU01:FS01:Assigned to work server 155.247.166.220
00:49:30:WU01:FS01:Requesting new work unit for slot 01: READY gpu:0:TU107 from 155.247.166.220
00:49:30:WU01:FS01:Connecting to 155.247.166.220:8080
00:49:31:WU01:FS01:Downloading 500.08KiB
00:49:32:WU01:FS01:Download complete
00:49:32:WU01:FS01:Received Unit: id:01 state:DOWNLOAD error:NO_ERROR project:14163 run:16 clone:75 gen:37 core:0x21 unit:0x000000360002894c5c38bea68d0abc12
00:49:32:WU01:FS01:Starting
00:49:32:ERROR:WU01:FS01:Failed to start core: OpenCL device matching slot 1 not found, try setting 'opencl-index' manually
00:49:32:WU01:FS01:Starting
00:49:32:ERROR:WU01:FS01:Failed to start core: OpenCL device matching slot 1 not found, try setting 'opencl-index' manually
00:50:32:WU01:FS01:Starting
00:50:32:ERROR:WU01:FS01:Failed to start core: OpenCL device matching slot 1 not found, try setting 'opencl-index' manually
00:50:40:WU02:FS00:0xa7:Completed 17500 out of 125000 steps (14%)
00:52:03:WU02:FS00:0xa7:Completed 18750 out of 125000 steps (15%)
00:52:10:WU01:FS01:Starting
00:52:10:ERROR:WU01:FS01:Failed to start core: OpenCL device matching slot 1 not found, try setting 'opencl-index' manually
00:53:25:WU02:FS00:0xa7:Completed 20000 out of 125000 steps (16%)
now i know its a opencl issue but i cant seem to fix it. I am wondering if its some issue with the old amd card drivers or something.
what am i missing? lol i just want this to work lol
Nvidia says it should be enough to edit the khronos registry keys.
HKLM\SOFTWARE\Khronos\OpenCL\Vendors (for 64-bit applications)
or
HKLM\SOFTWARE\Wow6432Node\Khronos\OpenCL\Vendors (for 32-bit applications)
These registry keys should contain one REG_DWORD value for each OpenCL platform. If you change the REG_DWORD value for a platform from 0 to 1 then that platform will no longer enumerate. The nice thing about this method is that if you change your mind, you can re-enable the platform by simply changing the value from 1 back to 0.
It would be helpful if you included the applicable portions of FAH's log. (see below) When FAHClient initiqalizes, it should report the presence or absence of supporting software. In my case, it looks like this:
22:35:20:******************************* System ********************************
22:35:20: CPU: Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz
22:35:20: CPU ID: GenuineIntel Family 6 Model 26 Stepping 4
22:35:20: CPUs: 8
22:35:20: Memory: 5.99GiB
22:35:20: Free Memory: 4.91GiB
22:35:20: Threads: WINDOWS_THREADS
22:35:20: OS Version: 6.1
22:35:20: Has Battery: false
22:35:20: On Battery: false
22:35:20: UTC Offset: -7
22:35:20: PID: 6076
22:35:20: CWD: C:\Users\bruce\AppData\Roaming\FAHClient
22:35:20: OS: Windows 7 Home Premium
22:35:20: OS Arch: AMD64
22:35:20: GPUs: 2
22:35:20: GPU 0: Bus:3 Slot:0 Func:0 NVIDIA:4 GM107 [GeForce GTX 750 Ti] 1306
22:35:20: GPU 1: Bus:4 Slot:0 Func:0 NVIDIA:5 GM206 [GeForce GTX 960] 2308
22:35:20: CUDA Device 0: Platform:0 Device:0 Bus:4 Slot:0 Compute:5.2 Driver:9.1
22:35:20: CUDA Device 1: Platform:0 Device:1 Bus:3 Slot:0 Compute:5.0 Driver:9.1
22:35:20:OpenCL Device 0: Platform:0 Device:0 Bus:4 Slot:0 Compute:1.2 Driver:391.35
22:35:20:OpenCL Device 1: Platform:0 Device:1 Bus:3 Slot:0 Compute:1.2 Driver:391.35
22:35:20: Win32 Service: false
22:35:20:***********************************************************************
The driver versions of OpenCL and CUDA are important information. What does yours show?
I expect that the easiest way to fix this problem is to download and install the recommended drivers from nVidia. That's a lot safer than fixing the applicable registry keys manually -- just in case that there's another step that we forgot to have you change, too. (Otherwise, you'll need to post the applicable registry keys associated with your OpenCL installation and we'll have to go from there.)
Do you have a passkey entered into the folding client? If so, has the client run and processed at least 4-5% of the WU's? The estimates are based on whether or not a passkey is entered into the client, and the client takes a few percent of progress to settle on an accurate PPD estimate. It can take longer to get an accurate estimate if the client has not processed a WU form a particular project before.
iMac 2.8 i7 12 GB smp8, Mac Pro 2.8 quad 12 GB smp6
MacBook Pro 2.9 i7 8 GB smp3
Joe_H wrote:Do you have a passkey entered into the folding client? If so, has the client run and processed at least 4-5% of the WU's? The estimates are based on whether or not a passkey is entered into the client, and the client takes a few percent of progress to settle on an accurate PPD estimate. It can take longer to get an accurate estimate if the client has not processed a WU form a particular project before.
I have not, just our team number and name. I will let it got a week and see what it shows.
Without a passkey you will just get the base points awarded. The Quick Return Bonus (QRB) makes up a large portion of the points a fast card like a 1650 will get. It does take the return of 10 WU's to qualify a passkey and username pairing.
iMac 2.8 i7 12 GB smp8, Mac Pro 2.8 quad 12 GB smp6
MacBook Pro 2.9 i7 8 GB smp3
Joe_H wrote:Without a passkey you will just get the base points awarded. The Quick Return Bonus (QRB) makes up a large portion of the points a fast card like a 1650 will get. It does take the return of 10 WU's to qualify a passkey and username pairing.
Thanks, i have no idea what some of that means, so let me go read up on getting a passkey after 10wu
wow, that bumped it up to 330000 lol.
i have a little over 2000 WU's so was able to get that passkey, i need to add that to my other systems running that now lol
thanks for the help
interesting!
I think you're one of the first to report PPD on the 1650 over here.
~330k for a 1650.
It's about on par with a 1060 6GB (or slightly faster than a 1060 3GB).