Failed to start core: OpenCL device matching slot 3 not foun

[Etherion]

Im having a strange issue. Use to work fine but since last week i am getting this error Failed to start core: OpenCL device matching slot 3 not found, try setting 'opencl-index' manually

if I replace the cuda-index with opencl-index it works but its slow. points per day is all over the place.

Nvidia 440.82 ubuntu 18.04

Code: Select all

miner@WK05[192.168.0.105]/opt/mmp/miners/foldingathome sudo ./FAHClient.sh --user='' --passkey='' --team=''   --amd=0 --nvidia=6 --workdir=/media/usb-drive --allowed_ip='154.119.50.192'
06:42:01:INFO(1):Read GPUs.txt
06:42:02:************************* Folding@home Client *************************
06:42:02:      Website: https://foldingathome.org/
06:42:02:    Copyright: (c) 2009-2018 foldingathome.org
06:42:02:       Author: Joseph Coffland <joseph@cauldrondevelopment.com>
06:42:02:         Args: --config=/media/usb-drive/config.xml --power=full --cpu-usage=10
06:42:02:               --core-dir=/media/usb-drive/cores
06:42:02:               --data-directory=/media/usb-drive --log=/media/usb-drive/log.txt
06:42:02:               --log-color=false --log-rotate=false --log-truncate=true
06:42:02:               --cause=ANY --smp=true
06:42:02:       Config: /media/usb-drive/config.xml
06:42:02:******************************** Build ********************************
06:42:02:      Version: 7.5.1
06:42:02:         Date: May 11 2018
06:42:02:         Time: 19:59:04
06:42:02:   Repository: Git
06:42:02:     Revision: 4705bf53c635f88b8fe85af7675557e15d491ff0
06:42:02:       Branch: master
06:42:02:     Compiler: GNU 6.3.0 20170516
06:42:02:      Options: -std=gnu++98 -O3 -funroll-loops
06:42:02:     Platform: linux2 4.14.0-3-amd64
06:42:02:         Bits: 64
06:42:02:         Mode: Release
06:42:02:******************************* System ********************************
06:42:02:          CPU: Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz
06:42:02:       CPU ID: GenuineIntel Family 6 Model 60 Stepping 3
06:42:02:         CPUs: 8
06:42:02:       Memory: 11.60GiB
06:42:02:  Free Memory: 7.79GiB
06:42:02:      Threads: POSIX_THREADS
06:42:02:   OS Version: 5.4
06:42:02:  Has Battery: false
06:42:02:   On Battery: false
06:42:02:   UTC Offset: 0
06:42:02:          PID: 6132
06:42:02:          CWD: /opt/mmp/miners/foldingathome
06:42:02:           OS: Linux 5.4.32-mmp x86_64
06:42:02:      OS Arch: AMD64
06:42:02:         GPUs: 6
06:42:02:        GPU 0: Bus:1 Slot:0 Func:0 NVIDIA:7 GP104 [GeForce GTX 1070] 6463
06:42:02:        GPU 1: Bus:6 Slot:0 Func:0 NVIDIA:7 GP104 [GeForce GTX 1070] 6463
06:42:02:        GPU 2: Bus:7 Slot:0 Func:0 NVIDIA:7 GP104 [GeForce GTX 1070] 6463
06:42:02:        GPU 3: Bus:8 Slot:0 Func:0 NVIDIA:7 GP104 [GeForce GTX 1070] 6463
06:42:02:        GPU 4: Bus:9 Slot:0 Func:0 NVIDIA:7 GP104 [GeForce GTX 1070] 6463
06:42:02:        GPU 5: Bus:10 Slot:0 Func:0 NVIDIA:7 GP104 [GeForce GTX 1070] 6463
06:42:02:CUDA Device 0: Platform:0 Device:0 Bus:1 Slot:0 Compute:6.1 Driver:10.2
06:42:02:CUDA Device 1: Platform:0 Device:1 Bus:6 Slot:0 Compute:6.1 Driver:10.2
06:42:02:CUDA Device 2: Platform:0 Device:2 Bus:7 Slot:0 Compute:6.1 Driver:10.2
06:42:02:CUDA Device 3: Platform:0 Device:3 Bus:8 Slot:0 Compute:6.1 Driver:10.2
06:42:02:CUDA Device 4: Platform:0 Device:4 Bus:9 Slot:0 Compute:6.1 Driver:10.2
06:42:02:CUDA Device 5: Platform:0 Device:5 Bus:10 Slot:0 Compute:6.1 Driver:10.2
06:42:02:       OpenCL: Not detected: clGetDeviceIDs() returned -1
06:42:02:***********************************************************************
06:42:02:<config>
06:42:02:  <!-- Client Control -->
06:42:02:  <fold-anon v='true'/>
06:42:02:
06:42:02:  <!-- HTTP Server -->
06:42:02:  <allow v='127.0.0.1 154.119.50.192'/>
06:42:02:
06:42:02:  <!-- Remote Command Server -->
06:42:02:  <command-allow-no-pass v='127.0.0.1 154.119.50.192'/>
06:42:02:
06:42:02:  <!-- User Information -->
06:42:02:  <passkey v=''/>
06:42:02:  <team v=''/>
06:42:02:  <user v=''/>
06:42:02:
06:42:02:  <!-- Web Server -->
06:42:02:  <web-allow v='127.0.0.1 154.119.50.192'/>
06:42:02:
06:42:02:  <!-- Folding Slots -->
06:42:02:  <slot id='0' type='GPU'>
06:42:02:    <cuda-index v='0'/>
06:42:02:  </slot>
06:42:02:  <slot id='1' type='GPU'>
06:42:02:    <cuda-index v='1'/>
06:42:02:  </slot>
06:42:02:  <slot id='2' type='GPU'>
06:42:02:    <cuda-index v='2'/>
06:42:02:  </slot>
06:42:02:  <slot id='3' type='GPU'>
06:42:02:    <cuda-index v='3'/>
06:42:02:  </slot>
06:42:02:  <slot id='4' type='GPU'>
06:42:02:    <cuda-index v='4'/>
06:42:02:  </slot>
06:42:02:  <slot id='5' type='GPU'>
06:42:02:    <cuda-index v='5'/>
06:42:02:  </slot>
06:42:02:</config>
06:42:02:Trying to access database...
06:42:02:Successfully acquired database lock
06:42:02:Enabled folding slot 00: READY gpu:0:GP104 [GeForce GTX 1070] 6463
06:42:02:Enabled folding slot 01: READY gpu:1:GP104 [GeForce GTX 1070] 6463
06:42:02:Enabled folding slot 02: READY gpu:2:GP104 [GeForce GTX 1070] 6463
06:42:02:Enabled folding slot 03: READY gpu:3:GP104 [GeForce GTX 1070] 6463
06:42:02:Enabled folding slot 04: READY gpu:4:GP104 [GeForce GTX 1070] 6463
06:42:02:Enabled folding slot 05: READY gpu:5:GP104 [GeForce GTX 1070] 6463
06:42:02:ERROR:No compute devices matched GPU #1 NVIDIA:7 GP104 [GeForce GTX 1070] 6463.  You may need to update your graphics drivers.
06:42:02:ERROR:No compute devices matched GPU #2 NVIDIA:7 GP104 [GeForce GTX 1070] 6463.  You may need to update your graphics drivers.
06:42:02:ERROR:No compute devices matched GPU #3 NVIDIA:7 GP104 [GeForce GTX 1070] 6463.  You may need to update your graphics drivers.
06:42:02:ERROR:No compute devices matched GPU #4 NVIDIA:7 GP104 [GeForce GTX 1070] 6463.  You may need to update your graphics drivers.
06:42:02:ERROR:No compute devices matched GPU #5 NVIDIA:7 GP104 [GeForce GTX 1070] 6463.  You may need to update your graphics drivers.
06:42:02:WU02:FS02:Starting
06:42:02:ERROR:WU02:FS02:Failed to start core: OpenCL device matching slot 2 not found, try setting 'opencl-index' manually
06:42:02:WU03:FS03:Starting
06:42:02:ERROR:WU03:FS03:Failed to start core: OpenCL device matching slot 3 not found, try setting 'opencl-index' manually
06:42:02:WU04:FS04:Starting
06:42:02:ERROR:WU04:FS04:Failed to start core: OpenCL device matching slot 4 not found, try setting 'opencl-index' manually
06:42:02:WU01:FS01:Starting
06:42:02:ERROR:WU01:FS01:Failed to start core: OpenCL device matching slot 1 not found, try setting 'opencl-index' manually
06:42:02:WU02:FS02:Starting
06:42:02:ERROR:WU02:FS02:Failed to start core: OpenCL device matching slot 2 not found, try setting 'opencl-index' manually
06:42:02:WU03:FS03:Starting
06:42:02:ERROR:WU03:FS03:Failed to start core: OpenCL device matching slot 3 not found, try setting 'opencl-index' manually
06:42:02:WU04:FS04:Starting
06:42:02:ERROR:WU04:FS04:Failed to start core: OpenCL device matching slot 4 not found, try setting 'opencl-index' manually
06:42:02:WU01:FS01:Starting
06:42:02:ERROR:WU01:FS01:Failed to start core: OpenCL device matching slot 1 not found, try setting 'opencl-index' manually
06:42:03:WU00:FS00:Connecting to 65.254.110.245:8080

[Etherion]

it seems to be the same problem as these people are reporting but the work around does not work for me https://bbs.archlinux.org/viewtopic.php?id=254001

[PantherX]

Please note that --cpu-usage=10 isn't useful as you don't have any CPU slots.

Have you installed the OpenCL package? Nvidia GPUs need their proprietary drivers and OpenCL too to function correctly AFAIK.

[Etherion]

I installed Nvidia 440.82.

[Etherion]

If i remove the cuda-index and replace it with opencl-index it worked. but it is 1 point/s vs 8p/s when using cuda as before

[Etherion]

#sudo clinfo
Number of platforms: 2
Platform Profile: FULL_PROFILE
Platform Version: OpenCL 2.1 AMD-APP (2906.7)
Platform Name: AMD Accelerated Parallel Processing
Platform Vendor: Advanced Micro Devices, Inc.
Platform Extensions: cl_khr_icd cl_amd_event_callback cl_amd_offline_devices
Platform Profile: FULL_PROFILE
Platform Version: OpenCL 1.2 CUDA 10.2.159
Platform Name: NVIDIA CUDA
Platform Vendor: NVIDIA Corporation
Platform Extensions: cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_fp64 cl_khr_byte_addressable_store cl_khr_icd cl_khr_gl_sharing cl_nv_compiler_options cl_nv_device_attribute_query cl_nv_pragma_unroll cl_nv_copy_opts cl_nv_create_buffer cl_khr_int64_base_atomics cl_khr_int64_extended_atomics


opencl is installed

[PantherX]

Sorry, I am not a Linux expert but this line suggest that OpenCL is not being detected by the client which is causing issues:
06:42:02: OpenCL: Not detected: clGetDeviceIDs() returned -1

For Nvidia GPUs, you need CUDA and OpenCL for the GPU to fold successfully.

I am not sure what you mean by this "but it is 1 point/s vs 8p/s when using cuda as before" so can you please elaborate it? There is a CUDA FahCore in development (no ETA) so the current FahCore can't use CUDA.

[Etherion]

Nvidia driver has cuda 1.2 installed. Platform Version: OpenCL 1.2 CUDA 10.2.159

if I replace the cuda-index v= with opencl=index the it works but is runs 1/8 the number of points per day.

06:42:02: <!-- Folding Slots -->
06:42:02: <slot id='0' type='GPU'>
06:42:02: <cuda-index v='0'/>
06:42:02: </slot>
06:42:02: <slot id='1' type='GPU'>
06:42:02: <cuda-index v='1'/>
06:42:02: </slot>
06:42:02: <slot id='2' type='GPU'>
06:42:02: <cuda-index v='2'/>
06:42:02: </slot>
06:42:02: <slot id='3' type='GPU'>
06:42:02: <cuda-index v='3'/>
06:42:02: </slot>
06:42:02: <slot id='4' type='GPU'>
06:42:02: <cuda-index v='4'/>
06:42:02: </slot>
06:42:02: <slot id='5' type='GPU'>
06:42:02: <cuda-index v='5'/>
06:42:02: </slot>
06:42:02:</config

[PantherX]

Can you just try to pause all the GPU Slots except 1 and see how it performs? Keep doing that until the performance tanks. Theoretically, you have 8 CPUs for 6 GPUs so it may not be a CPU bottleneck. It could be a PCIe bandwidth issue but apart from physically removing the GPUs or disabling the PCIe Slot in the BIOS, don't know how you can test it via the OS.

Also, what's the temperatures like? Do you have sufficient cooling? How about PSU as 6 GPUs is a significant load and folding is very stressful.

[Etherion]

nothing is hot. one does the same as 6. I also tests this on vastai. get the same issue. once I put opencl-index in it runs but at 1/8 what it use to run. I use to get 8-9 Points/S.I have test this on 5 different systems so far. All does the same. I think most other folder don't watch the speed at all so they don't know. If you use --gpu=true then it will auto-configure

[Joe_H]

First, fix the OpenCL problem. People have reported needing to install the nVidia driver, nVidia OpenCL runtime support, and the OpenCL dev package.

Delete this - '--cpu-usage=10' - from your command. It does nothing for controlling CPU usage currently, that option was for an older CPU folding core that is no longer being used.

The folding core need to be directed to opencl-index numbers on a working OpenCL install. What the folding cores are doing otherwise I have only the guess that they might be running OpenCL code on your CPU instead.