I have only tried compiling kernels using pyopencl, but I can only seem to be able to use OpenCl C. Looking at clinfo, I only see support for CLC listed, heres some truncated output from my pc:
Platform Name AMD Accelerated Parallel Processing
Platform Vendor Advanced Micro Devices, Inc.
Platform Version OpenCL 2.1 AMD-APP (3423.0)
Platform Profile FULL_PROFILE
Platform Extensions cl_khr_icd cl_amd_event_callback
Platform Extensions function suffix AMD
Platform Host timer resolution 1ns
Platform Name AMD Accelerated Parallel Processing
Number of devices 1
Device Name gfx1031
Device Vendor Advanced Micro Devices, Inc.
Device Vendor ID 0x1002
Device Version OpenCL 2.0
Driver Version 3423.0 (HSA1.1,LC)
Device OpenCL C Version OpenCL C 2.0
Device Type GPU
Device Board Name (AMD) AMD Radeon RX 6700 XT
Device PCI-e ID (AMD) 0x73df
Device Topology (AMD) PCI-E, 0000:2f:00.0
Device Profile FULL_PROFILE
Device Available Yes
Compiler Available Yes
Linker Available Yes
Max compute units 20
I am using a rocm driver compiled from the AUR, I tried to also install the mesa driver alongside but could not do so (perhaps I need to uninstall the other, but I dread having to recompile it if mesa fails).
My laptop (intel hd graphics) seems to support OpenCL 3.0 but also only lists CLC support. What am I missing, is this not implemented yet? I saw something somewhere about "offline compilation" and maybe using a "clc++" option with clang but can someone elaborate?
C++ for OpenCL can be used in two ways:
If OpenCL device supports cl_ext_cxx_for_opencl, it is possible to compile a program written using the C++ for OpenCL kernel language in runtime. Applications may pass -cl-std=CLC++ to clCompileProgram and clBuildProgram for programs created using clCreateProgramFromSource to request the program be built as C++ for OpenCL.
If OpenCL device allows to create the program with SPIR-V, then it is possible to compile C++ for OpenCL source into intermediate LLVM IR:
clang -c -cl-std=clc++ -target spir64 -O3 -emit-llvm -o test.ll test.clcpp
Next, LLVM IR can be translated into SPIR-V using llvm-spirv:
llvm-spirv test.ll -o test.spv
Finally, OpenCL program can be created using clCreateProgramWithIL call:
std::ifstream il_file("test.spv", std::ios::binary);
std::vector<char> il_buffer;
std::copy(std::istreambuf_iterator<char>(il_file),
std::istreambuf_iterator<char>(),
std::back_inserter(il_buffer));
cl_program program =
clCreateProgramWithIL(context, il_buffer.data(), il_buffer.size(), &ret);
For PyOpenCL:
with open('test.spv', 'rb') as il_file:
il_buffer = bytes(il_file.read())
prg = cl.Program(ctx, il_buffer)
To check that OpenCL device supports SPIR-V modules, you need to use CL_DEVICE_IL_VERSION query in OpenCL 2.1 or newer and the CL_DEVICE_ILS_WITH_VERSION query in OpenCL 3.0 or newer.
For additional information about offline compilation please see Offline Compilation of OpenCL Kernels into SPIR-V Using Open Source Tooling article.