# AdaptiveCpp/AdaptiveCpp

Compiler for multiple programming models (SYCL, C++ standard parallelism, HIP/CUDA) for CPUs and GPUs from all vendors: The independent, community-driven compiler for C++-based heterogeneous programming models. Lets applications adapt themselves to all the hardware in the system - even at runtime!

Repository: https://github.com/AdaptiveCpp/AdaptiveCpp
Canonical: https://ross.abutalabs.com/products/adaptivecpp
Homepage: https://adaptivecpp.github.io/
Language: C++
License: BSD-2-Clause
License Family: permissive
Topics: sycl, gpgpu, high-performance, gpu-computing, high-performance-computing, hipsycl, opensycl, stdpar, adaptivecpp, compiler, hpc
Last push: 2026-08-26T19:03:42+00:00

## Health v2 (maintenance only)
Score: 75/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 31, longevity 100
- inputs: {"age_days": 2972, "days_push": 7, "days_rel": 301, "gap_med": 159.0, "n_releases_24m": 3}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1928, forks 225 (observed 2026-08-28T04:05:55.827172+00:00)

## What it is
AdaptiveCpp (formerly hipSYCL/Open SYCL) is an independent, community-driven C++ compiler platform for heterogeneous programming models including SYCL, C++ standard parallelism (stdpar), and portable CUDA/HIP. It targets CPUs and GPUs from all major vendors (NVIDIA, AMD, Intel, Apple) and can produce a single binary that adapts to all hardware in the system, even at runtime.

## Use cases
- compile SYCL code for NVIDIA, AMD, Intel, and Apple GPUs
- offload C++ standard parallelism (parallel STL) to GPUs from any vendor
- compile CUDA or HIP code into a single portable binary that also runs on CPUs
- run one binary across heterogeneous hardware from multiple vendors simultaneously
- use vendor profilers and debuggers with portable GPU code
- avoid vendor lock-in for GPU-accelerated HPC applications

## When to choose
- you need a vendor-independent SYCL implementation
- you want C++ stdpar offloading across NVIDIA, AMD, and Intel GPUs
- you need a single binary targeting CPUs and GPUs from multiple vendors
- you want to mix CUDA and SYCL in the same source file
- you need portable CUDA/HIP code that also runs on CPUs

## When to avoid
- you only target a single vendor and prefer the vendor's fully supported toolchain (CUDA, ROCm, oneAPI)
- you need guaranteed vendor-level support and certification
- you rely on experimental features like PCUDA or stdpar offloading requiring production stability

## Facets
- artifact type: cli-tool
- maturity: active
- function: compiler, gpu-computing, concurrency
- domain: gpu-computing, compilers, cross-platform
- platform: windows, cpp
- tags: sycl, stdpar, cuda, hip, heterogeneous-computing, hpc, llvm, jit-compiler, gpgpu, compiler, high-performance-computing, linux, macos, gpu

## Member repositories
- AdaptiveCpp/AdaptiveCpp (main) score 75

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:55.827172+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T08:32:12.827950+00:00, confidence not recorded.
  - readme: https://github.com/AdaptiveCpp/AdaptiveCpp (fetched 2026-08-28T04:05:55.827172+00:00, sha 0f967759b894)
  - homepage: https://adaptivecpp.github.io/ (fetched 2026-08-29T10:48:26.511998+00:00, sha bd1febc6de16)
- Data as of 2026-08-30T08:39:29.467469+00:00.