stotko/stdgpu

Health v2 · maintenance only

Adoption not part of the score

1270 stars · 102 forks observed · 2026-08-28

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

stdgpu is a lightweight C++17 library providing STL-like generic data structures (vector, unordered_map, unordered_set, deque, queue, stack, bitset, atomics) for GPU computing. It supports CUDA, OpenMP, and experimental HIP backends and interoperates with thrust.

Use cases

• use hash maps and sets on the GPU in CUDA kernels
• STL-like containers for GPU programming in C++
• manage dynamic data structures on GPU memory
• write portable GPU code across CUDA, HIP, and OpenMP
• concurrent insertion into GPU unordered_map from parallel kernels
• replace thrust with richer container support for GPU algorithms

When to choose

• you need associative or dynamic containers (maps, sets, deques) on the GPU rather than just contiguous arrays
• you want STL-familiar APIs and backend-agnostic code across CUDA/HIP/OpenMP
• you already use thrust and need data management beyond contiguous storage

When to avoid

• you only need simple contiguous arrays and GPU algorithms - thrust or ArrayFire may suffice
• you need a full GPU algorithm/BLAS ecosystem rather than a container library
• you require production HIP support - that backend is still experimental

Facets

library · maturity active

gpu-computing concurrency developer-tools gpu-computing developer-tools cpp windows stl-containers gpgpu hip openmp data-structures cpp17 cpp20 algorithms gpu cuda linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="stotko/stdgpu")