Valentin Churavy
Valentin Churavy
This is due to https://github.com/JuliaGPU/KernelAbstractions.jl/blob/1f84b17414bc325fce199d51e97f75a408aa1e63/src/macros.jl#L237
I am pretty sure that this is the same error as the one from https://github.com/EnzymeAD/Enzyme.jl/pull/2751#issuecomment-3513929585
Huh? That should be fixed post #615 and JuliaGPU/OpenCL.jl#325 each package should use it's own `method_table` and then stack that on top of the SPIRV method table.
@simeonschaub could I bother you to take a look at what it would take to implement device side rand for pocl?
That should throw a better error....
This is a common issue with low-level math functionality. You will want to either call an llvm intrinsic or define a custom rule for that math operator?
In order for code to relocatable you may not observe world as a compile time constant. You may look at the tls_world_age, but you may not cache that or put...
Might also make sense to add `versioninfo(backend)`.
KA is split across multiple packages, there are the backends in CUDA/AMDGPU/oneAPI/OpenCL... and then of course the core here. One question for now would be what does `similar` do with...