nvvm
FFI bindings to NVVM
https://github.com/tmcdonell/nvvm
| Version on this page: | 0.8.0.1 |
| LTS Haskell 23.21: | 0.10.0.1 |
| Stackage Nightly 2024-12-09: | 0.10.0.1 |
| Latest on Hackage: | 0.10.0.1 |
nvvm-0.8.0.1@sha256:7875cf904b864f208f143e5956a21abcb620255a60ea75c9d83b490f5baaed66,2349Module documentation for 0.8.0.1
Haskell FFI Bindings to NVVM
The NVVM library compiles NVVM IR (a subset of LLVM IR) into PTX code which can then be executed on NVIDIA GPUs.
In contrast to the standard NVPTX target built in to the LLVM
toolchain, NVVM includes a set of proprietary optimisations which are otherwise
only available by compiling CUDA code with the nvcc compiler. On the other
hand, the version of LLVM that NVVM is internally based on typically lags the
public release by several generations (years), so these secret optimisations may
or may not be worthwhile to your application.
The resulting PTX code can be loaded onto the GPU and executed using the cuda package.
The NVVM library is a compiler component available a part of the CUDA toolkit:
https://developer.nvidia.com/cuda-toolkit
The configure step will look for your CUDA installation in the standard places,
and if the nvcc compiler is found in your PATH, relative to that.