AMD Announces Radeon Open eCosystem ROCm 3.0
AMD's press release mentions ROCm 3.0 being released though as of writing it has yet to appear via the ROCm repositories on GitHub. Once the actual drop happens, I'll certainly be writing about it and digging deeper into the other changes in full.
AMD mentions the ROCm 3.0 highlights as:
- Introduction of ROCm 3.0 with new innovations to support HIP-clang – a compiler built upon LLVM, improved CUDA conversion capability with hipify-clang, library optimizations for both HPC and ML.
- ROCm upstream integration into leading TensorFlow and PyTorch machine learning frameworks for applications like reinforcement learning, autonomous driving, and image and video detection.
- Expanded acceleration support for HPC programing models and applications like OpenMP programing, LAMMPS, and NAMD.
- New support for system and workload deployment tools like Kubernetes, Singularity, SLURM, TAU and others.
The improved HIP-clang will be interesting to see for how well it's working out in porting CUDA codes to run on AMD Radeon / Radeon Pro hardware. Also expanded support for OpenMP and scientific workloads like LAMMPS and NAMD will be interesting to see and benchmark. ROCm support for Kubernetes is another interesting addition. Hopefully the actual ROCm 3.0 release notes will shed more light on these specifics and will be trying them out as time allows.
Interestingly, the press release doesn't mention if ROCm 3.0 has Navi support. Since the July launch of the Radeon RX 5700 series there hasn't been any OpenCL/ROCm support for Navi on Linux. Hopefully with ROCm 3.0 there will be Navi/GFX10 support in place but we should know soon enough once the release hits GitHub.
Besides releasing ROCm 3.0 for SC19, the rest of AMD's press release is mostly EPYC news. 2nd Gen AMD EPYC CPUs are seeing more Amazon EC2 instances roll out, EPYC CPUs appearing on the TOP500 list, and other EPYC adoption. Their SC19 press release can be read at AMD.com.