Tom Vander Aa

Hi Guys, I have the same issue in some of my own codes. How do I at least get a more meaningful error message, such that I know that the...

Hey, does this mean that tracing is currently broken? It seems tracing expect newpc with a _normal_ pointer and not a `PlainMetaAddress`: https://github.com/revng/revng/blob/9573946a92d7fbb10b4f709de8a35505dd8baa83/runtime/support.c#L335-L339 While the call to `newpc` puts a...

Thanks for the suggestion. I did not try this yet, but the following (also?) works: ``` void newpc( uint32_t Epoch, uint16_t address_space, uint16_t type, uint64_t address, uint64_t instruction_size, uint32_t is_first,...

The following patch seems to fix this: ``` commit dd0fe0e8a43fdef6069419fe75d315cc39aefb6d (HEAD -> gcc10, imec/gcc10) Author: Tom Vander Aa Date: Tue Dec 14 15:13:53 2021 +0100 WIP: consistent ordering of ZeroSizedLabels...

NP, next time you’ll get a PR

Working on this, started with a CPU-only conda version for arrayfire 3.7.1 -- https://github.com/tvandera/arrayfire/tree/master/conda-recipe/arrayfire-cpu-3.7.1

There now is also a CUDA enabled arrayfire-gpu: - Recipe at: https://github.com/tvandera/arrayfire/tree/master/conda-recipe/arrayfire-gpu-3.7.3 - Binary packages are available from https://anaconda.org/vanderaa/repo To build or install the package you need CUDA 11.0 from...

Okay, slowdown was due to an inconsistency in AF_CUDA_MAX_JIT_LEN. This has been fixed now. Anyone from arrayfire.com interested in conda packages?

Working on it: https://github.com/tvandera/staged-recipes/tree/arrayfire

Hi, I'm currently building - Linux + CUDA 10.2 - Linux + MKL (CPU) + Intel OpenMP - Windows + MKL (CPU) + TBB - OSX + MKL (CPU) +...