POCL 1.6-RC1 Released With Better CUDA Performance

POCL as the "Portable Computing Language" that implements OpenCL and allows it to function atop CPUs as well as CUDA-enabled NVIDIA GPUs, HSA-supported AMD GPUs, and other possible back-ends, is preparing for a new feature release.

On Wednesday marked the release of POCL 1.6-RC1 as the test release for the next update to the Portable Computing Language.

POCL continues leveraging LLVM/Clang for doing much of the heavy lifting and supporting a diverse variety of back-ends for CPU execution and targeting other devices/accelerators. With POCL 1.6 there is support for building against the LLVM/Clang 11.0 code-base.

Also new to POCL 1.6 is improved CUDA performance and functionality when using that NVIDIA LLVM back-end for targeting. The other area focused on for POCL 1.6 development has improving the OpenCL debugging experience.

The release candidate of POCL 1.6 can be downloaded from GitHub. Those wanting to learn more about the Portable Computing Language can visit PortableCL.org.