AccelerateHS/accelerate

Health v2 · maintenance only

Flags: no_license

Adoption not part of the score

1018 stars · 135 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Accelerate is an embedded domain-specific language in Haskell for high-performance, multi-dimensional array computations. Programs written as collective operations are online-compiled via LLVM or CUDA backends to run on multicore CPUs and GPUs.

Use cases

• run array computations on GPUs from Haskell
• write high-performance parallel numeric code
• accelerate compute-intensive kernels like dot products or stencil operations
• compile Haskell array programs to LLVM or CUDA
• prototype GPGPU programs in a high-level functional style
• offload dense linear algebra or image processing workloads

When to choose

• you program in Haskell and need GPU or multicore acceleration for dense array math
• you want type-safe, declarative parallel code without writing CUDA by hand
• you need portable performance across CPU and GPU backends

When to avoid

• you need irregular or sparse data structures rather than dense regular arrays
• your project is not in Haskell
• you need fine-grained manual control over GPU kernels or memory layout
• you need a mature ecosystem of ML libraries on GPU

Facets

library · maturity active

compiler gpu-computing math concurrency gpu-computing programming-languages compilers performance cpp cross-platform embedded-language array-computing llvm cuda parallel-computing haskell gpgpu high-performance-computing algorithms gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="AccelerateHS/accelerate")