Anton Smirnov

@jpsamaroo does `#0 0xffffffffffffffff in ?? ()` at the top mean what you thought, that it tries to execute `-1` while it should ignore it?

One situation that causes this issue and is present in the tests as well is the following: 1. We are [launching](https://github.com/JuliaGPU/AMDGPU.jl/blob/166043c4e689c330f275fb7419f65471677b8929/test/hsa/error.jl#L8) kernel and it also creates default queue which is...

Sorry for the lack of info. Here's full recipe which is used by BinaryBuilder: [link](https://github.com/JuliaPackaging/Yggdrasil/blob/master/R/rocBLAS/common.jl). I tried doing build on two machines. - One has 8 cores and 64 GB...

On the note of RAM consumption, when rocBLAS is built with `all` amdgpu targets and first call to `rocblas_sgemm` happens (for example), the amount of RAM that is used increases...

> @pxl-th does this seem right, or should we be removing the Julia compat instead? I see that it used incorrect versions of dependencies, like ROCmOpenCLRuntime 5.2.3. I'm thinking maybe...

Right now it is possible to use `DataLoader` with datasets that do not fit into memory, but feels somewhat hacky... or not. Let's say we have dataset of images, which...

Not sure what is the exact problem, but you can try using my config, which works for me in Neovim. Specify plugins (and install them): ```vim " Julia Plug 'JuliaEditorSupport/julia-vim',...

You can enable verbose logging and specify in `.vimrc` path to log file to see what's going on under the hood: ```vim let g:lsp_log_verbose = 1 let g:lsp_log_file = expand('~/vimfiles/logs/vim-lsp.log')...

Mapreduce is now defined for `AnyROCArray`, feel free to re-open if you still have issues with it.

There are NNlib and Flux extensions now