ArrayFire v2.0 is now available for download. The second iteration of our free, fast, and simple GPU library now supports both CUDA and OpenCL devices.
Major Updates
- ArrayFire now works on OpenCL enabled devices
- New and improved documentation
- Optimized for new GPUs–NVIDIA Kepler (K20) and AMD Tahiti (7970)
New in ArrayFire OpenCL
- Same APIs as ArrayFire CUDA version
- Supports both Linux and Windows
- Just In Time Compilation (JIT) of kernels
- Parallel for: gfor
- Accelerated algorithms in the following domains
New in ArrayFire CUDA
- New Signal and Image processing functions
- Faster transpose and matrix multiplication
- Better debugging support for GDB and Visual Studio
- Bug fixes to make overall experience better
For a more complete list of the many ArrayFire v2.0 enhancements available, check out our release notes.
To learn more about ArrayFire licensing, visit the ArrayFire licensing page.
We leverage our deep GPU expertise and vast code base for GPU consulting services. Set up a free GPU consultation today by emailing us at sales@accelereyes.com.