Radeon ROCm 3.8 Released With Hipfort For Fortran On GPUs, Data Center Tool
Version 3.8 of ROCm, the Radeon Open eCosystem, is now available. This release continues making more progress on preparing the ROCm graphics compute stack for the upcoming large AMD supercomputer deployments and other data center usage.
ROCm 3.8 introduces the Hipfort interface library for accessing GPU kernels, which allows for interfacing via the Fortran programming language with Radeon GPUs. This Hipfort Fortran implementation is currently catering to GCC's Gfortran front-end.
ROCm 3.8 also introduces the ROCm Data Center Tool as a means for accessing GPU telemetry, GPU job statistics, third-party tooling integration, and more. ROCm Data Center Tool is open-source like the rest of the Radeon Open eCosystem stack.
ROCm 3.8 also introduces the ability to build static ROCm libraries and link to applications statically, fixes issues with P2P transfers, and a number of other fixes.
ROCm 3.8 continues to list GFX7/GFX8/GFX9 GPUs as officially supported with no mention yet of GFX10/Navi. I had tried with ROCm 3.7 getting it working well with Radeon RX 5700 XT but during those experiments didn't have much of a stable compute experience.
Downloads and more details on ROCm 3.8 via GitHub.
ROCm 3.8 introduces the Hipfort interface library for accessing GPU kernels, which allows for interfacing via the Fortran programming language with Radeon GPUs. This Hipfort Fortran implementation is currently catering to GCC's Gfortran front-end.
ROCm 3.8 also introduces the ROCm Data Center Tool as a means for accessing GPU telemetry, GPU job statistics, third-party tooling integration, and more. ROCm Data Center Tool is open-source like the rest of the Radeon Open eCosystem stack.
ROCm 3.8 also introduces the ability to build static ROCm libraries and link to applications statically, fixes issues with P2P transfers, and a number of other fixes.
ROCm 3.8 continues to list GFX7/GFX8/GFX9 GPUs as officially supported with no mention yet of GFX10/Navi. I had tried with ROCm 3.7 getting it working well with Radeon RX 5700 XT but during those experiments didn't have much of a stable compute experience.
Downloads and more details on ROCm 3.8 via GitHub.
15 Comments