KernelAbstractions.jl

KernelAbstractions.jl copied to clipboard

add N for hardware indices

1

vchuravy

Use MultiplicativeInverse to speedup Linear to Cartesian indexing operations

5

Related to https://github.com/JuliaGPU/Metal.jl/issues/101 Using the idea from @N5N3 https://github.com/JuliaGPU/Metal.jl/issues/101#issuecomment-1447420196

vchuravy

Multi-GPU backend

5

Hi! I am trying to use KernelAbstractions.jl to work on a FDTD (Finite-Difference) simulation package. Overall, from what I understand, everything is fine except for multi-GPU. I am having trouble...

lucasgrjn

Allow return statements for GPU-only kernels

2

It's a bit annoying to not be able to use return statements. Currently you have to do something like this, but it looks weird: ```julia @kernel function ker!(args...) i =...

pxl-th

The kind of kernels that are expressible in KernelAbstractions are currently limited by the CPU support having to opperate on a macro level. With the CPU support being implemented in...

vchuravy

1.11: reduce_private test generates gcframe under --check-bounds=yes

1

MWE, reduced from the tests: ```julia using KernelAbstractions @kernel function reduce_private(out, A) I = @index(Global, NTuple) i = @index(Local) priv = @private eltype(A) (1,) @inbounds begin priv[1] = zero(eltype(A)) for...

maleadt

amd gpu give different results when nested loop is used

18

Hi, I noticed that the following script produces different results depending on the backend. On my machine, the output is: ```julia cpu: [18.0; 18.0; 18.0; 18.0; 18.0; 18.0; 18.0; 18.0;...

ww1g11

vchuravy

As the title suggests, there is a behaviour discrepancy between `import KernelAbstractions` and `using KernelAbstractions` in relation to the `@Const` macro. The following minimum example demonstrates the issue. ```julia import...

Hamiltonian-Action

Use Printf to report errors from POCL

3

- **enable printf in errors again** - **turn off 1.10 testing**

vchuravy