Aurora Perego
Aurora Perego
> EDIT: Missed your code snippet. Using `printf` inside kernels is illegal in SYCL. Thanks, without the `printf` everything works apart from the host device which continues to fail. There...
> Do you know why you removed the additional `extern` declaration of the variable? With the new implementation the `extern` declaration was not working, I used the `inline` attribute instead
yes, we finished on the same page :) I'll look into it tomorrow with Mehmet
I confirm that this is version-specific, I can reproduce it with CUDA 12.4 and the flag `-rdc=false` but not with CUDA 12.{0,1,2}. It compiles with `static` instead of `inline` (see...
The `DevGlobal` internally is already templated on the `Tag` and there is a map from `Acc` to `Tag` [here](https://github.com/alpaka-group/alpaka/blob/develop/include/alpaka/mem/global/Traits.hpp#L44). I don't remember why we are passing the `Acc` and not...
I've no idea why the tests failed with rocm 5.7 and passed with 5.1 to 5.6 and 6.0 :( The one that fails is `mathTest`, with `Idx i: 634 computed...
The error in the CI is in the build of CUDA 12.4 with GCC 13.1 (supported according to [the documentation](https://docs.nvidia.com/cuda/archive/12.4.0/cuda-installation-guide-linux/index.html#host-compiler-support-policy)): ```bash $ cudafe++ --c++17 --gnu_version=130100 --display_error_number --orig_src_file_name "CMakeCUDACompilerId.cu" \ --orig_src_path_name...
with the following configuration: ```shell $ gcc -v Using built-in specs. COLLECT_GCC=gcc COLLECT_LTO_WRAPPER=/cvmfs/sft.cern.ch/lcg/releases/gcc/13.1.0-b3d18/x86_64-centos7/bin/../libexec/gcc/x86_64-pc-linux-gnu/13.1.0/lto-wrapper Target: x86_64-pc-linux-gnu Configured with: /workspace/build/contrib/gcc-13.1.0/src/gcc/13.1.0/configure --prefix=/workspace/lcgcmake-install/gcc/13.1.0/x86_64-centos7 --disable-nls --with-system-zlib --disable-multilib --enable-languages=c,c++,fortran,lto,go --with-gnu-ld --with-gnu-as --enable-__cxa_atexit --disable-libunwind-exceptions --enable-gnu-unique-object --enable-plugin --with-linker-hash-style=gnu...
> We have an error with CUDA 12.4 and GCC 13.1, that seems unrelated to this PR: > > ``` > /usr/include/x86_64-linux-gnu/bits/floatn-common.h(214): error: invalid > combination of type specifiers >...
> The path looks like it use the standard library of the host compiler which is GCC 9 in a Ubuntu 20.04 container. Can you please check with this application...