{"full_name":"arrayfire/arrayfire","owner":"arrayfire","name":"arrayfire","shard":"1","url":"https://github.com/arrayfire/arrayfire","description":"ArrayFire: a general purpose GPU library.","homepage":"https://arrayfire.com","topics":["arrayfire","c-plus-plus","gpu","cuda","opencl","gpgpu","hpc","performance","scientific-computing","cpp","c"],"tag_pages":["c","c-plus-plus","cpp","cuda","gpgpu","gpu","hpc","opencl","performance","scientific-computing"],"language":"C++","license":"BSD-3-Clause","avatar":"https://avatars.githubusercontent.com/u/5395442?v=4","created_at":"2014-10-28T20:58:33Z","pushed_at":"2026-11-12T13:45:03Z","metrics":{"stars":4913,"forks":558,"watchers":140,"issues":1714,"prs":2029,"releases":36,"issues_open":294,"issues_closed":1420,"prs_open":44,"prs_closed":206,"prs_merged":1779,"discussions":0},"growth":{"day":{"stars":2,"forks":0,"base_day":"2026-11-15"}},"history":{"dates":["2026-11-10","2026-11-11","2026-11-12","2026-11-13","2026-11-14","2026-11-15","2026-11-16"],"stars":[4910,4911,4911,4911,4911,4911,4913],"forks":[558,558,558,558,558,558,558],"watchers":[null,140,140,140,140,140,140],"issues_open":[null,294,295,295,295,294,294],"prs_open":[null,44,45,44,44,44,44]},"boards":{},"same_owner":[],"issues":[{"number":1673,"title":"Support for fp16","comments":33,"created_at":"2016-12-14T21:09:01Z"},{"number":2518,"title":"in v3.6.4 c2r FFT changes the value of the input","comments":21,"created_at":"2019-05-21T21:20:52Z"},{"number":482,"title":"Expose array references through the API.","comments":19,"created_at":"2015-03-10T09:21:18Z"},{"number":313,"title":"Arrayfire Static Library","comments":19,"created_at":"2015-01-08T05:55:05Z"},{"number":858,"title":"Make sure all functions have asynchronous counterparts","comments":18,"created_at":"2015-06-27T11:19:26Z"},{"number":904,"title":"Add eigen value decomposition for all backends","comments":17,"created_at":"2015-07-13T15:31:28Z"},{"number":113,"title":"Implement streaming support for all backends","comments":17,"created_at":"2014-11-13T16:12:11Z"},{"number":3384,"title":"[BUG] Various tests failing in the CPU backend","comments":16,"created_at":"2023-03-26T18:37:16Z"}],"discussions":[],"latest_release":{"name":"v3.10.1","tag":"v3.10.1","published_at":"2026-11-11T12:44:17Z","body":"v3.10.1\n======\n\n## Improvements\n- Added support for CUDA 13.0 through 13.4\n- Added support for NVIDIA Rubin GPUs (compute capability 10.7) with CUDA 13.4\n- Added support for the per-toolkit layout of cuDNN 9 installers. Set\n  `cuDNN_ROOT_DIR` to the cuDNN install directory, such as\n  `C:/Program Files/NVIDIA/CUDNN/v9.27`\n- Windows and Linux installers are now published for CUDA 12.9 and CUDA 13.4,\n  and the toolkit version is part of the installer name\n- The Linux installers are built on Rocky Linux 8 with oneAPI 2026.2 and run\n  on any x86_64 distribution with glibc 2.26 or newer\n- `CUDA_architecture_build_targets=All` now targets every architecture supported\n  by the CUDA toolkit, plus PTX\n\n## Compatibility\n- The CUDA 13.4 installer requires a GPU driver from the R615 branch or newer,\n  because ArrayFire's JIT generates PTX with the bundled NVRTC\n- CUDA 13 removed Maxwell, Pascal and Volta GPUs. Use the CUDA 12.9 installer\n  for these GPUs\n- The Linux CUDA 12.9 installer does not carry SASS for the Tegra-only\n  architectures (sm_53, sm_62, sm_72, sm_87)\n## Fixes\n- Fix oneAPI errors on devices without double precision support\n- Fix fast math being enabled in the CPU and OpenCL backends when building with\n  the Intel oneAPI compiler on Windows\n- Fix the CUDA compiler being taken from the PATH instead of the selected CUDA\n  toolkit when several toolkits are installed\n- Fix broken page links in the documentation and docs generation on Windows\n- Fix the static LAPACK library not being linked on CUDA 12 and later, which\n  left `libafcuda.so` with an unresolved symbol when\n  `AF_WITH_STATIC_CUDA_NUMERIC_LIBS=ON` #3349 #3742\n- Fix the static cuFFT library leaving `libafcuda.so` with an unresolved\n  device-link symbol when `AF_WITH_STATIC_CUDA_NUMERIC_LIBS=ON` #3745\n- Fix the Linux standalone install missing the Intel compiler and SYCL\n  runtime libraries #3744\n- Add a mirror for the Matrix Market test archives #3743\n","is_prerelease":false},"readme":true,"readme_zh":false,"news":[],"stale":false,"snapshot_at":"2026-11-16T09:23:14Z","history_start":"2026-09-06"}