NVIDIA C++ Standard Library Now Available Via GitHub
Introduced last year as part of CUDA 10.2 was libcu++ as the CUDA C++ standard library, which works with not only NVIDIA CUDA enabled configurations but also CPUs. The libcu++ sources are now available via GitHub.
Libcu++ as the CUDA C++ Standard Library has been taking shape over the past year and has been advancing since with more capabilities. NVIDIA's libcu++ is forked from LLVM's libc++ standard library.
NVIDIA has talked previously of upstreaming all their libcu++ changes back into LLVM libc++, but at least for now that hasn't yet happened and thus libcu++ has merit on its own right as a C++ standard library for CUDA + CPU configurations.
New as of yesterday is libcu++ is open-source on GitHub. It was destined from the beginning to be open-source albeit took longer than expected. Libcu++ is available via NVIDIA/libcudacxx. It will be interesting to see if all the work gets upstreamed back into LLVM's libc++ or if libcu++ sees much adoption in its own right outside of the specific CUDA focus.
Libcu++ as the CUDA C++ Standard Library has been taking shape over the past year and has been advancing since with more capabilities. NVIDIA's libcu++ is forked from LLVM's libc++ standard library.
NVIDIA has talked previously of upstreaming all their libcu++ changes back into LLVM libc++, but at least for now that hasn't yet happened and thus libcu++ has merit on its own right as a C++ standard library for CUDA + CPU configurations.
New as of yesterday is libcu++ is open-source on GitHub. It was destined from the beginning to be open-source albeit took longer than expected. Libcu++ is available via NVIDIA/libcudacxx. It will be interesting to see if all the work gets upstreamed back into LLVM's libc++ or if libcu++ sees much adoption in its own right outside of the specific CUDA focus.
14 Comments