Valentin Churavy
Valentin Churavy
You still have the same issue that if Torch.jl includes `Torch_jll, TorchCUDA_jll,...` all dependencies must be fulfilled. Better to use a platform tag and platform specific dependencies.
That is great to hear!
No, the Travis setup is a bit flaky though, so hopefully this retriggered it
I dream of using clang everywhere (if only flang was production ready)
What would also be somewhat cool is to record the the flags to cmake/configure.
I am encountering this in PMIx_jll and pprte_jll which both copy OpenMPI MCA approach. It is again an environment variable relative to the artifact_dir.
#42 allows for user to call certain functions only if they is a platform available that should support those api calls. But it is still lacking a mechanism for the...
I would say that the breakage currently would be acceptable, especially since we could add ``` julia import Api const api = Api ``` into the OpenCL Module to minimize...
On Travis we are currently using `AMD-APP-SDK-linux-v2.9-1.599.381-GA-x64.tar.bz2` If I remember correctly because of bugs like these...
Just did and I am seeing the same behaviour. The seqfault happens when setting the first argument of the kernel https://github.com/JuliaGPU/OpenCL.jl/blob/409cb88e0921ffa5c12e79525fb16b2ab89025da/src/kernel.jl#L76, so for `buffer(B)`. Does anybody have an AMD gpu...