Siu Chi Chan

I was able to reproduce the memory access error in `void mxnet::op::col2im_gpu_kernel` on a fiji GPU. As I mentioned earlier, this type of error usually suggests a out-of-bound access. There...

We don't generate debug information in the current version of hcc and also I don't think a debugger would help you much for this. For simplicity, can you pass in...

We currently don't support this due to a limitation in hcc on how we support multiple architecture in a single binary file. Our frontend currently only parses the source code...

clang has a compiler switch called --gcc-toolchain that seems to support this kind of usage, do you mind giving that a try? --gcc-toolchain=, -gcc-toolchain Use the gcc toolchain at the...

Could you try setting the env var `CUDA_PATH=/opt/nvidia/hpc_sdk/Linux_x86_64/23.9/cuda/12.2` and see if that works around the non-standard location?