Radeon ROCm 2.5 Released With rocThrust, AMD Infinity Fabric Link Support

Written by Michael Larabel in Radeon on 8 June 2019 at 06:30 AM EDT. 23 Comments
RADEON
Version 2.5 of the Radeon Compute Stack (ROCm) was released on Friday as the newest feature release to this open-source HPC / GPU computing stack for AMD graphics hardware.

New to ROCm 2.5 are quite a few additions:

- UCX 1.6 support as the Unified Communication X interface as a performant layer for MPI and OpenSHMEM libraries.

- BFloat16 was implemented in software for Radeon Instinct MI50/MI60 hardware for use with rocBLAS/Tensile.

- The rocm-smi utility now has various information additions around memory size, driver version, and firmware version being queried.

- Multi-GPU support for PyTorch.

- Also on the Radeon Instinct MI50/MI60 front are rocSPARSE library optimizations.

- rocThrust as a port of the Thrust parallel algorithm library atop HIP/ROCm.

- Support for independent kernels to run in parallel in the same stream.

- Initial AMD Infinity Fabric Link enablement for their new GPU interconnect technology.

Expect more featureful ROCm updates still in the pipe with AMD continuing to pickup new supercomputing wins.

More details on ROCm 2.5 and information on its architecture are available from GitHub.
Related News
About The Author
Michael Larabel

Michael Larabel is the principal author of Phoronix.com and founded the site in 2004 with a focus on enriching the Linux hardware experience. Michael has written more than 20,000 articles covering the state of Linux hardware support, Linux performance, graphics drivers, and other topics. Michael is also the lead developer of the Phoronix Test Suite, Phoromatic, and OpenBenchmarking.org automated benchmarking software. He can be followed via Twitter, LinkedIn, or contacted via MichaelLarabel.com.

Popular News This Week