It seems that a lot of GPU problems revolve around specific versions of drivers. Though AMD has their own support structure, you can often learn from information reported by others who fold.
After getting fahcontrol installed (with much work), the amdgpu-pro drivers installed (with everything installed ./amdgpu-pro-install --opencl=pal,legacy) and adding myself and fahclient to the render group I have been able to get my graphics card identified by fahcontrol. Unfortunately, when I try to use it it attempts to connect to 192.0.2.1:80 and :8080 and is not able download anything. With the following from the log,
ERROR:WU00:FS01:Exception: Failed to connect to 192.0.2.1:80: Connection timed out
09:21:26:WU00:FS01:Connecting to assign1.foldingathome.org:80
09:21:26:WU00:FS01:Assigned to work server 192.0.2.1
09:21:26:WU00:FS01:Requesting new work unit for slot 01: READY gpu:0:Navi 14 [Radeon RX 5500/5500M / Pro 5500M] from 192.0.2.1
09:21:26:WU00:FS01:Connecting to 192.0.2.1:8080
System Info shows the GPU but says that OpenCL is not detected (no libOpenCL.so) but I installed the drivers with opencl requested. Can anyone help so I can get the card folding?
Either you don't have OpenCL installed, or you don't have the right one, or you have several installed.
In all those cases, your GPU is blacklisted and your system is directed to 192.0.2.1, so that it doesn't download a WU which will return errors.
The "right" OpenCL would be the dev package:
iirc you need both the amd drivers and the .dev from what I have seen in other threads ... iirc there is at least one thread specific to building in mint 20 - I'll try to find it ... this may have some stuff in in that might assist viewtopic.php?f=108&t=35820&hilit=mint+20 - realise for the most part it is nvidia but some of it (vendor drivers and the .dev) may hold true
Well I removed the amdgpu-pro drivers and the dev ones and re-installed in that order. Still have the same error messages listed above. Clinfo shows 0 for opencl installed. It showed 1 before when the amdgpu-pro opencl was installed.
And try a former version of the drivers. I have had one case (in Windows) where the newest AMD drivers couldn't help messing with Open CL. And the former ones did work.
Try the All-Open stack, then. If those work, as they don't come with OpenCL at all, it might be better for FAH (still with the dev package, of course).
clinfo
Number of platforms 1
Platform Name AMD Accelerated Parallel Processing
Platform Vendor Advanced Micro Devices, Inc.
Platform Version OpenCL 2.1 AMD-APP (3143.9)
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_icd cl_amd_event_callback cl_amd_offline_devices
Platform Host timer resolution 1ns
Platform Extensions function suffix AMD
Platform Name AMD Accelerated Parallel Processing
Number of devices 1
Device Name gfx1012
Device Vendor Advanced Micro Devices, Inc.
Device Vendor ID 0x1002
Device Version OpenCL 2.0 AMD-APP (3143.9)
Driver Version 3143.9 (PAL,LC)
Device OpenCL C Version OpenCL C 2.0
Device Type GPU
Device Board Name (AMD) Radeon RX 5500 XT
Device Topology (AMD) PCI-E, 09:00.0
Device Profile FULL_PROFILE
Device Available Yes
Compiler Available Yes
Linker Available Yes
Max compute units 11
SIMD per compute unit (AMD) 4
SIMD width (AMD) 32
SIMD instruction width (AMD) 1
Max clock frequency 1900MHz
Graphics IP (AMD) 10.12
Device Partition (core)
Max number of sub-devices 11
Supported partition types None
Supported affinity domains (n/a)
Max work item dimensions 3
Max work item sizes 1024x1024x1024
Max work group size 256
Preferred work group size (AMD) 256
Max work group size (AMD) 1024
Preferred work group size multiple 32
Wavefront width (AMD) 32
Preferred / native vector sizes
char 4 / 4
short 2 / 2
int 1 / 1
long 1 / 1
half 1 / 1 (cl_khr_fp16)
float 1 / 1
double 1 / 1 (cl_khr_fp64)
Half-precision Floating-point support (cl_khr_fp16)
Denormals No
Infinity and NANs No
Round to nearest No
Round to zero No
Round to infinity No
IEEE754-2008 fused multiply-add No
Support is emulated in software No
Single-precision Floating-point support (core)
Denormals Yes
Infinity and NANs Yes
Round to nearest Yes
Round to zero Yes
Round to infinity Yes
IEEE754-2008 fused multiply-add Yes
Support is emulated in software No
Correctly-rounded divide and sqrt operations Yes
Double-precision Floating-point support (cl_khr_fp64)
Denormals Yes
Infinity and NANs Yes
Round to nearest Yes
Round to zero Yes
Round to infinity Yes
IEEE754-2008 fused multiply-add Yes
Support is emulated in software No
Address bits 64, Little-Endian
Global memory size 8573157376 (7.984GiB)
Global free memory (AMD) 8306688 (7.922GiB)
Global memory channels (AMD) 4
Global memory banks per channel (AMD) 4
Global memory bank width (AMD) 256 bytes
Error Correction support No
Max memory allocation 7059013632 (6.574GiB)
Unified memory for Host and Device No
Shared Virtual Memory (SVM) capabilities (core)
Coarse-grained buffer sharing Yes
Fine-grained buffer sharing Yes
Fine-grained system sharing No
Atomics No
Minimum alignment for any data type 128 bytes
Alignment of base address 2048 bits (256 bytes)
Preferred alignment for atomics
SVM 0 bytes
Global 0 bytes
Local 0 bytes
Max size for global variable 6353112064 (5.917GiB)
Preferred total size of global vars 8573157376 (7.984GiB)
Global Memory cache type Read/Write
Global Memory cache size 16384 (16KiB)
Global Memory cache line size 64 bytes
Image support Yes
Max number of samplers per kernel 16
Max size for 1D images from buffer 134217728 pixels
Max 1D or 2D image array size 2048 images
Base address alignment for 2D image buffers 256 bytes
Pitch alignment for 2D image buffers 256 pixels
Max 2D image size 16384x16384 pixels
Max 3D image size 2048x2048x2048 pixels
Max number of read image args 128
Max number of write image args 64
Max number of read/write image args 64
Max number of pipe args 16
Max active pipe reservations 16
Max pipe packet size 2764046336 (2.574GiB)
Local memory type Local
Local memory size 65536 (64KiB)
Local memory syze per CU (AMD) 65536 (64KiB)
Local memory banks (AMD) 32
Max number of constant args 8
Max constant buffer size 7059013632 (6.574GiB)
Preferred constant buffer size (AMD) 16384 (16KiB)
Max size of kernel argument 1024
Queue properties (on host)
Out-of-order execution No
Profiling Yes
Queue properties (on device)
Out-of-order execution Yes
Profiling Yes
Preferred size 262144 (256KiB)
Max size 8388608 (8MiB)
Max queues on device 1
Max events on device 1024
Prefer user sync for interop Yes
Number of P2P devices (AMD) 0
P2P devices (AMD) <printDeviceInfo:147: get number of CL_DEVICE_P2P_DEVICES_AMD : error -30>
Profiling timer resolution 1ns
Profiling timer offset since Epoch (AMD) 1597005654959312626ns (Sun Aug 9 15:40:54 2020)
Execution capabilities
Run OpenCL kernels Yes
Run native kernels No
Thread trace supported (AMD) Yes
Number of async queues (AMD) 4
Max real-time compute queues (AMD) 1
Max real-time compute units (AMD) 0
printf() buffer size 4194304 (4MiB)
Built-in kernels (n/a)
Device Extensions cl_khr_fp64 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_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_fp16 cl_khr_gl_sharing cl_khr_gl_depth_images cl_amd_device_attribute_query cl_amd_media_ops cl_amd_media_ops2 cl_khr_image2d_from_buffer cl_khr_subgroups cl_khr_gl_event cl_khr_depth_images cl_khr_mipmap_image cl_khr_mipmap_image_writes cl_amd_copy_buffer_p2p
NULL platform behavior
clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...) No platform
clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...) No platform
clCreateContext(NULL, ...) [default] No platform
clCreateContext(NULL, ...) [other] Success [AMD]
clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT) Success (1)
Platform Name AMD Accelerated Parallel Processing
Device Name gfx1012
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) Success (1)
Platform Name AMD Accelerated Parallel Processing
Device Name gfx1012
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ACCELERATOR) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CUSTOM) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ALL) Success (1)
Platform Name AMD Accelerated Parallel Processing
Device Name gfx1012
sudo -u fahclient clinfo
[sudo] password for jeffmr:
Number of platforms 1
Platform Name AMD Accelerated Parallel Processing
Platform Vendor Advanced Micro Devices, Inc.
Platform Version OpenCL 2.1 AMD-APP (3143.9)
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_icd cl_amd_event_callback cl_amd_offline_devices
Platform Host timer resolution 1ns
Platform Extensions function suffix AMD
Platform Name AMD Accelerated Parallel Processing
Number of devices 0
NULL platform behavior
clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...) No platform
clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...) No platform
clCreateContext(NULL, ...) [default] No platform
clCreateContext(NULL, ...) [other] No platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ACCELERATOR) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CUSTOM) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_ALL) No devices found in platform
This is all with the install you mention and with ocl-icd-opencl-dev.
The clinfo command shows that this is a permissions issue with newer versions of Ubuntu and Mint, but it can be solved:
Add fahclient to the video and render groups:
If it still doesn't work with your card, there are reports that ROCM works with some AMD cards now, so in that case you might try to replace AMD's OpenCL with ROCM's - but try the above first.
Online: GTX 1660 Super + occasional CPU folding in the cold.
Offline: Radeon HD 7770, GTX 1050 Ti 4G OC, RX580