Tech-X announced the latest release of the GPULib today. GPULib provides a library of mathematical functions that facilitate the use of high performance computing resources available on modern GPUs. Version 1.4 supports CUDA streams, enabling users to concurrently execute multiple kernels. It also supports asynchronous data transfers between host and GPU memory.
The folks at Tech-X will be demonstrating their new library release the NVIDIA’s GPU-Tech Conference in San Jose this week. If you’re around the conference, stop by and say hello. For more info, check out their full release here.