You are not logged in.
I have the following AMD graphics card and I have been trying to use it for GPU accelerated computing using OpenCL mesa driver. I am trying to use ArrayFire for the computation and after posting a couple of issues about it in the ArrayFire github page, found that the issue is with the Mesa driver for Arch.
0d:00.0 Display controller: Advanced Micro Devices, Inc. [AMD/ATI] Sun XT [Radeon HD 8670A/8670M/8690M / R5 M330 / M430 / Radeon 520 Mobile] (rev 83)
Subsystem: Hewlett-Packard Company Device 80c2
Flags: bus master, fast devsel, latency 0, IRQ 52
Memory at b0000000 (64-bit, prefetchable) [size=256M]
Memory at d4000000 (64-bit, non-prefetchable) [size=256K]
I/O ports at 4000 [size=256]
Expansion ROM at d4040000 [disabled] [size=128K]
Capabilities: <access denied>
Kernel driver in use: radeon
Kernel modules: radeon, amdgpu
I tried using the mesa driver with OpenCl support in the official repositories, but found that it is buggy and would always result in some error. Also, it does not support any higher version other than OpenCL 1.1. After a distro hop to Ubuntu, I used the mesa driver from Oibaf's PPA and it seemed to work pretty well. After some searching, I found that an equivalent in Arch is the mesa-git driver. Now, in my Arch, I installed the compiled mesa-git and opencl-mesa-git from Laurent Carlier's repository. After, testing it with Julia I find the following errors.
julia> a = AFArray(rand(4,4))
AFArray: 4×4 Array{Float64,2}:
0.910863 0.528568 0.769485 0.262519
0.758505 0.59156 0.464172 0.498584
0.888846 0.839448 0.482488 0.179536
0.470455 0.0313732 0.784979 0.995334
julia> a+1
AFArray: Error showing value of type AFArray{Float64,2}:
ERROR: ArrayFire Error (998) : Internal error
OpenCL Error (-11): Build Program Failure when calling clBuildProgram
Stacktrace:
[1] _error(::UInt32, ::Bool) at /home/<username>/.julia/packages/ArrayFire/wYxcd/src/util.jl:96
[2] _error at /home/<username>/.julia/packages/ArrayFire/wYxcd/src/util.jl:85 [inlined]
[3] get_data_ptr at /home/<username>/.julia/packages/ArrayFire/wYxcd/src/wrap.jl:748 [inlined]
[4] convert_array(::AFArray{Float64,2}) at /home/<username>/.julia/packages/ArrayFire/wYxcd/src/util.jl:216
[5] Array at /home/<username>/.julia/packages/ArrayFire/wYxcd/src/array.jl:33 [inlined]
[6] toa(::AFArray{Float64,2}) at /home/<username>/.julia/packages/ArrayFire/wYxcd/src/util.jl:48
[7] show(::IOContext{REPL.Terminals.TTYTerminal}, ::MIME{Symbol("text/plain")}, ::AFArray{Float64,2}) at /home/<username>/.julia/packages/ArrayFire/wYxcd/src/util.jl:52
[8] display(::REPL.REPLDisplay, ::MIME{Symbol("text/plain")}, ::Any) at /build/julia/src/julia-1.4.2/usr/share/julia/stdlib/v1.4/REPL/src/REPL.jl:137
[9] display(::REPL.REPLDisplay, ::Any) at /build/julia/src/julia-1.4.2/usr/share/julia/stdlib/v1.4/REPL/src/REPL.jl:141
[10] display(::Any) at ./multimedia.jl:323
[11] #invokelatest#1 at ./essentials.jl:712 [inlined]
[12] invokelatest at ./essentials.jl:711 [inlined]
[13] print_response(::IO, ::Any, ::Bool, ::Bool, ::Any) at /build/julia/src/julia-1.4.2/usr/share/julia/stdlib/v1.4/REPL/src/REPL.jl:161
[14] print_response(::REPL.AbstractREPL, ::Any, ::Bool, ::Bool) at /build/julia/src/julia-1.4.2/usr/share/julia/stdlib/v1.4/REPL/src/REPL.jl:146
[15] (::REPL.var"#do_respond#38"{Bool,REPL.var"#48#57"{REPL.LineEditREPL,REPL.REPLHistoryProvider},REPL.LineEditREPL,REPL.LineEdit.Prompt})(::Any, ::Any, ::Any) at /build/julia/src/julia-1.4.2/usr/share/julia/stdlib/v1.4/REPL/src/REPL.jl:729
So, it seems the mesa-git is also buggy in my case. My question is why are the drivers on Mesa buggy and can it be fixed? Also, is there any way that I can install the drivers from Oibaf's PPA(which are for Ubuntu) on Arch as they seem to work without any errors? You help here would be great as I will then be able to do my calculations without the need to switch over from Arch.
Linux <username> 5.7.3-arch1-1 #1 SMP PREEMPT Wed, 17 Jun 2020 19:42:12 +0000 x86_64 GNU/Linux
Further information: I am using the latest Arch kernel and have just updated all my drivers after a fresh reinstall a couple of days ago. I would be happy to provide any other information as needed. Thank you
Last edited by Lattitude (2020-06-21 04:23:29)
Offline
There's a package "opencl-amd" in the AUR. That package downloads the "AMDGPU-PRO" driver package and then extracts just the OpenCL driver part out of it. Perhaps try that opencl-amd package and see if it works for you. Your GPU is old so it might not work. Remove the opencl-mesa package while you try opencl-amd.
You will need to run the "amdgpu" kernel module while you try this. Your system currently uses "radeon". You need to force a switch to amdgpu. Check the ArchWiki amdgpu article to see how to do that.
Offline
Thank you for the reply. I actually tried that too. I did a complete switch to amdgpu by blacklisting the radeon module in the mkinitcpio and setting the required kernel parameters. Then, I installed opencl-amd which supports the mesa driver from amdgpu pro. This did not work either. I also found that my card is not supported by the amdgpu-pro driver. The only option which did work was using the updated Clover drivers from OiBaf's PPA as I said earlier, but obviously only in Ubuntu. I think I want a similar driver for Arch if possible.
Offline
julia> a = AFArray(rand(4,4))
ERROR: UndefVarError: AFArray not defined
Stacktrace:
[1] top-level scope at REPL[1]:1
Please provide a clear testcase suitable for someone that has no experience with julia and arrayfire .
Falling back to troubleshooting with limited info.
Mesa OpenCL uses libclc .
The libclc version Oibaf provides is 0.2.0+git20200209.1921.9aa6f3~oibaf , which builds from exactly the same commit (9aa6f3 ) as the archlinux libclc pacakge .
Looking further at oibaf ppa they build mesa-git against llvm 10 , the mesa-git in Lordheavy's repo is build against llvm git (currently at 11) .
My aur mesa-git package[1] allows selecting against which llvm version it should be build , repo llvm is one of the choices.
Try building that and verify if it solves your problems.
If it doesn't help, please link to the issues in arrayfire github about this.
They may provide enough info to figure out what's causing the problem.
[1] https://aur.archlinux.org/packages/mesa-git/
Last edited by Lone_Wolf (2020-06-20 12:26:35)
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
Thank you very much for the reply, Lone_Wolf. I would try building mesa-git as you suggested and get back to you. I will also raise an issue in ArrayFire github page and see if they can shine some light on the issue. Is it okay if I keep this thread open until then?
Offline
This package works fine with my RX570, Open CL working in Darktable and Libre Office.
Offline
@Arup, Thanks. But, my card is R5 M330. I tried to use the proprietary drivers too as I said earlier. But, it did not help.
To add further, this is the information from clinfo:
Number of platforms 1
Platform Name Clover
Platform Vendor Mesa
Platform Version OpenCL 1.1 Mesa 20.2.0-devel (git-e94a122642)
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_icd
Platform Extensions function suffix MESA
Platform Name Clover
Number of devices 1
Device Name AMD HAINAN (DRM 2.50.0, 5.7.3-arch1-1, LLVM 11.0.0)
Device Vendor AMD
Device Vendor ID 0x1002
Device Version OpenCL 1.1 Mesa 20.2.0-devel (git-e94a122642)
Driver Version 20.2.0-devel
Device OpenCL C Version OpenCL C 1.1
Device Type GPU
Device Profile FULL_PROFILE
Device Available Yes
Compiler Available Yes
Max compute units 5
Max clock frequency 1030MHz
Max work item dimensions 3
Max work item sizes 256x256x256
Max work group size 256
=== CL_PROGRAM_BUILD_LOG ===
fatal error: cannot open file '/usr//usr/lib/clc/hainan-amdgcn-mesa-mesa3d.bc': No such file or directory
Preferred work group size multiple <getWGsizes:1256: create kernel : error -46>
Preferred / native vector sizes
char 16 / 16
short 8 / 8
int 4 / 4
long 2 / 2
half 0 / 0 (n/a)
float 4 / 4
double 2 / 2 (cl_khr_fp64)
Half-precision Floating-point support (n/a)
Single-precision Floating-point support (core)
Denormals No
Infinity and NANs Yes
Round to nearest Yes
Round to zero No
Round to infinity No
IEEE754-2008 fused multiply-add No
Support is emulated in software No
Correctly-rounded divide and sqrt operations No
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 2147483648 (2GiB)
Error Correction support No
Max memory allocation 1503238553 (1.4GiB)
Unified memory for Host and Device No
Minimum alignment for any data type 128 bytes
Alignment of base address 32768 bits (4096 bytes)
Global Memory cache type None
Image support No
Local memory type Local
Local memory size 32768 (32KiB)
Max number of constant args 16
Max constant buffer size 1503238400 (1.4GiB)
Max size of kernel argument 1024
Queue properties
Out-of-order execution No
Profiling Yes
Profiling timer resolution 0ns
Execution capabilities
Run OpenCL kernels Yes
Run native kernels No
Device Extensions cl_khr_byte_addressable_store 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_fp64
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 [MESA]
clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT) Success (1)
Platform Name Clover
Device Name AMD HAINAN (DRM 2.50.0, 5.7.3-arch1-1, LLVM 11.0.0)
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) Success (1)
Platform Name Clover
Device Name AMD HAINAN (DRM 2.50.0, 5.7.3-arch1-1, LLVM 11.0.0)
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 Clover
Device Name AMD HAINAN (DRM 2.50.0, 5.7.3-arch1-1, LLVM 11.0.0)
OpenCL Error (-11): Build Program Failure when calling clBuildProgram
And in the above code I see that it is unable to get the file hainan-amdgcn-mesa-mesa3d.bc. And from preliminary search, I found that it is available in libclc-amdgcn_0.2.0 in some ubuntu packages. Is there any way this error can be averted? Or is this file available in the libclc package from the base repository?
@Lone_Wolf I have built the mesa-git with libclc, llvm-libs=10.0.0. But, is there an opencl-mesa package for it? The official opencl-mesa does not seem to work with mesa-git package built from your repo. ArrayFire uses OpenCL from Mesa Clover for the calculations. I see that the clinfo shows zero platforms after installing mesa-git from your aur.
Last edited by Lattitude (2020-06-20 21:31:51)
Offline
My mesa-git package combines the functionality of repo mesa, opencl-mesa, vulkan-intel, vulkan-radeon, vulkan-mesa-layer, libva-mesa-driver and mesa-vdpau stock packages into one package .
To use it you have to answer yes to all conflicts upon install.
You may not need to test with my mesa-git package though, the clinfo gives a clue about what may be wrong.
/usr//usr/lib/clc/hainan-amdgcn-mesa-mesa3d.bc
clinfo is looking at the wrong path , the file is at /usr/lib/clc/hainan-amdgcn-mesa-mesa3d.bc .
This suggests something is wrong with your ICD loader setup or ld.so searchpath.
post
$ pacman -Qs icd
$ ls -l /etc/ld.so.conf.d/
Last edited by Lone_Wolf (2020-06-20 21:44:55)
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
@Lattitude if its any help here is my clinfo
[arup@ryzen-arup ~]$ clinfo
Number of platforms 2
Platform Name Clover
Platform Vendor Mesa
Platform Version OpenCL 1.1 Mesa 20.1.1
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_icd
Platform Extensions function suffix MESA
Platform Name AMD Accelerated Parallel Processing
Platform Vendor Advanced Micro Devices, Inc.
Platform Version OpenCL 2.1 AMD-APP (3110.6)
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 Clover
Number of devices 1
Device Name Radeon RX 570 Series (POLARIS10, DRM 3.37.0, 5.7.4-arch1-1, LLVM 10.0.0)
Device Vendor AMD
Device Vendor ID 0x1002
Device Version OpenCL 1.1 Mesa 20.1.1
Driver Version 20.1.1
Device OpenCL C Version OpenCL C 1.1
Device Type GPU
Device Profile FULL_PROFILE
Device Available Yes
Compiler Available Yes
Max compute units 32
Max clock frequency 1268MHz
Max work item dimensions 3
Max work item sizes 256x256x256
Max work group size 256
Preferred work group size multiple 64
Preferred / native vector sizes
char 16 / 16
short 8 / 8
int 4 / 4
long 2 / 2
half 0 / 0 (n/a)
float 4 / 4
double 2 / 2 (cl_khr_fp64)
Half-precision Floating-point support (n/a)
Single-precision Floating-point support (core)
Denormals No
Infinity and NANs Yes
Round to nearest Yes
Round to zero No
Round to infinity No
IEEE754-2008 fused multiply-add No
Support is emulated in software No
Correctly-rounded divide and sqrt operations No
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 8589934592 (8GiB)
Error Correction support No
Max memory allocation 6871947673 (6.4GiB)
Unified memory for Host and Device No
Minimum alignment for any data type 128 bytes
Alignment of base address 32768 bits (4096 bytes)
Global Memory cache type None
Image support No
Local memory type Local
Local memory size 32768 (32KiB)
Max number of constant args 16
Max constant buffer size 2147483392 (2GiB)
Max size of kernel argument 1024
Queue properties
Out-of-order execution No
Profiling Yes
Profiling timer resolution 0ns
Execution capabilities
Run OpenCL kernels Yes
Run native kernels No
Device Extensions cl_khr_byte_addressable_store 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_fp64
Platform Name AMD Accelerated Parallel Processing
Number of devices 1
Device Name Ellesmere
Device Vendor Advanced Micro Devices, Inc.
Device Vendor ID 0x1002
Device Version OpenCL 1.2 AMD-APP (3110.6)
Driver Version 3110.6
Device OpenCL C Version OpenCL C 1.2
Device Type GPU
Device Board Name (AMD) Radeon RX 570 Series
Device Topology (AMD) PCI-E, 08:00.0
Device Profile FULL_PROFILE
Device Available Yes
Compiler Available Yes
Linker Available Yes
Max compute units 32
SIMD per compute unit (AMD) 4
SIMD width (AMD) 16
SIMD instruction width (AMD) 1
Max clock frequency 1268MHz
Graphics IP (AMD) 8.0
Device Partition (core)
Max number of sub-devices 32
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 64
Wavefront width (AMD) 64
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 No
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 8259293184 (7.692GiB)
Global free memory (AMD) 8046132 (7.673GiB)
Global memory channels (AMD) 8
Global memory banks per channel (AMD) 16
Global memory bank width (AMD) 256 bytes
Error Correction support No
Max memory allocation 4244635648 (3.953GiB)
Unified memory for Host and Device No
Minimum alignment for any data type 128 bytes
Alignment of base address 2048 bits (256 bytes)
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 8
Local memory type Local
Local memory size 32768 (32KiB)
Local memory syze per CU (AMD) 65536 (64KiB)
Local memory banks (AMD) 32
Max number of constant args 8
Max constant buffer size 4244635648 (3.953GiB)
Preferred constant buffer size (AMD) 16384 (16KiB)
Max size of kernel argument 1024
Queue properties
Out-of-order execution No
Profiling Yes
Prefer user sync for interop Yes
Profiling timer resolution 1ns
Profiling timer offset since Epoch (AMD) 1592680572751492895ns (Sun Jun 21 03:16:12 2020)
Execution capabilities
Run OpenCL kernels Yes
Run native kernels No
Thread trace supported (AMD) Yes
Number of async queues (AMD) 2
Max real-time compute queues (AMD) 0
Max real-time compute units (AMD) 0
SPIR versions 1.2
printf() buffer size 4194304 (4MiB)
Built-in kernels (n/a)
Device Extensions cl_khr_fp64 cl_amd_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_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_media_ops2 cl_amd_popcnt cl_khr_image2d_from_buffer cl_amd_bus_addressable_memory cl_khr_spir cl_khr_gl_event
NULL platform behavior
clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...) Clover
clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...) Success [MESA]
clCreateContext(NULL, ...) [default] Success [MESA]
clCreateContext(NULL, ...) [other] Success [AMD]
clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT) Success (1)
Platform Name Clover
Device Name Radeon RX 570 Series (POLARIS10, DRM 3.37.0, 5.7.4-arch1-1, LLVM 10.0.0)
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) Success (1)
Platform Name Clover
Device Name Radeon RX 570 Series (POLARIS10, DRM 3.37.0, 5.7.4-arch1-1, LLVM 10.0.0)
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 Clover
Device Name Radeon RX 570 Series (POLARIS10, DRM 3.37.0, 5.7.4-arch1-1, LLVM 10.0.0)
ICD loader properties
ICD loader Name OpenCL ICD Loader
ICD loader Vendor OCL Icd free software
ICD loader Version 2.2.12
ICD loader Profile OpenCL 2.2
[arup@ryzen-arup ~]$
Last edited by Arup (2020-06-20 22:00:59)
Offline
I see that the clinfo shows zero platforms after installing mesa-git from your aur.
The majority of mesa runtime functionality is provided by llvm-libs, but for opencl you need a bit more.
Use pacman -Qi mesa-git and check the optional dependencies.
Last edited by Lone_Wolf (2020-06-20 21:58:16)
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
The majority of mesa runtime functionality is provided by llvm-libs, but for opencl you need a bit more.
Use pacman -Qi mesa-git and check the optional dependencies.
Thank you very much for this advice. Seems, like I missed the packages clang and compiler-rt. Once I installed them, I could get the opencl-mesa to work with the following info.
Number of platforms 1
Platform Name Clover
Platform Vendor Mesa
Platform Version OpenCL 1.1 Mesa 20.2.0-devel (git-0c32403c73)
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_icd
Platform Extensions function suffix MESA
Platform Name Clover
Number of devices 1
Device Name AMD HAINAN (DRM 2.50.0, 5.7.3-arch1-1, LLVM 10.0.0)
Device Vendor AMD
Device Vendor ID 0x1002
Device Version OpenCL 1.1 Mesa 20.2.0-devel (git-0c32403c73)
Driver Version 20.2.0-devel
Device OpenCL C Version OpenCL C 1.1
Device Type GPU
Device Profile FULL_PROFILE
Device Available Yes
Compiler Available Yes
Max compute units 5
Max clock frequency 1030MHz
Max work item dimensions 3
Max work item sizes 256x256x256
Max work group size 256
Preferred work group size multiple 64
Preferred / native vector sizes
char 16 / 16
short 8 / 8
int 4 / 4
long 2 / 2
half 0 / 0 (n/a)
float 4 / 4
double 2 / 2 (cl_khr_fp64)
Half-precision Floating-point support (n/a)
Single-precision Floating-point support (core)
Denormals No
Infinity and NANs Yes
Round to nearest Yes
Round to zero No
Round to infinity No
IEEE754-2008 fused multiply-add No
Support is emulated in software No
Correctly-rounded divide and sqrt operations No
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 2147483648 (2GiB)
Error Correction support No
Max memory allocation 1503238553 (1.4GiB)
Unified memory for Host and Device No
Minimum alignment for any data type 128 bytes
Alignment of base address 32768 bits (4096 bytes)
Global Memory cache type None
Image support No
Local memory type Local
Local memory size 32768 (32KiB)
Max number of constant args 16
Max constant buffer size 1503238400 (1.4GiB)
Max size of kernel argument 1024
Queue properties
Out-of-order execution No
Profiling Yes
Profiling timer resolution 0ns
Execution capabilities
Run OpenCL kernels Yes
Run native kernels No
Device Extensions cl_khr_byte_addressable_store 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_fp64
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 [MESA]
clCreateContextFromType(NULL, CL_DEVICE_TYPE_DEFAULT) Success (1)
Platform Name Clover
Device Name AMD HAINAN (DRM 2.50.0, 5.7.3-arch1-1, LLVM 10.0.0)
clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU) No devices found in platform
clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU) Success (1)
Platform Name Clover
Device Name AMD HAINAN (DRM 2.50.0, 5.7.3-arch1-1, LLVM 10.0.0)
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 Clover
Device Name AMD HAINAN (DRM 2.50.0, 5.7.3-arch1-1, LLVM 10.0.0)
With the above change, I am now able to run ArrayFire in Julia without any errors. Seems like my only step was to compile mesa-git from your aur with the default packages and install clang and compiler-rt.
post
$ pacman -Qs icd
local/ocl-icd 2.2.12-4
OpenCL ICD Bindings
local/vulkan-icd-loader 1.2.141-1
Vulkan Installable Client Driver (ICD) Loader
ls -l /etc/ld.so.conf.d/
total 12
-rw-r--r-- 1 root root 21 Jun 19 09:20 arrayfire.conf
-rw-r--r-- 1 root root 21 Nov 13 2019 fakeroot.conf
-rw-r--r-- 1 root root 17 Jun 14 16:12 openmpi.conf
I will now mark this as solved.
Offline