141 lines (122 loc) · 6.06 KB

GPU Programming Languages

Page maintainer: Alessio Sclocco @isazi

Learning Resources

Carpentries GPU Programming course
- Lesson material
Introduction to CUDA C
- Slides
Introduction to OpenACC
- Slides
Introduction to HIP Programming
- Video
SYCL Introduction and Best Practices
- Video
CSCS GPU Programming with Julia
- Course recordings

Documentation

CUDA
HIP
- Kernel language syntax
- Runtime API
SYCL
- Specification
- Reference guide
OpenCL
OpenACC
- Programming guide
- Reference guide
OpenMP
- Reference guide

Overview of Libraries

CUDA
- cuBLAS
- NVBLAS
- cuFFT
- cuGRAPH
- cuRAND
- cuSPARSE
HIP
- hipBLAS
- hipFFT
- hipRAND
- hipSPARSE
SYCL
OpenCL
- CLBlast
- clFFT

Source-to-source Translation

CUDA to HIP
- hipify
CUDA to SYCL
- SYCLomatic
CUDA to OpenCL
- cutocl

Foreign Function Interfaces

C++
- CUDA
  - cudawrappers
- OpenCL
  - CLHPP
Python
- CUDA
  - PyCuda
  - CuPy
  - cuda-python
- HIP
  - PyHIP
- SYCL
  - dpctl
- OpenCL
  - PyOpenCL
Julia
- CUDA
  - CUDA.jl
- HIP
  - AMDGPU.jl
- SYCL
  - oneAPI.jl
Java
- CUDA
  - JCuda
- OpenCL
  - JOCL

High-Level Abstractions

C++
- Kokkos
- Raja
Python
- Numba
- pykokkos

Debugging and Profiling Tools

Performance Optimization

Auto-tuning

Kernel Tuner

⚡