Originally published at: https://developer.nvidia.com/blog/simplifying-gpu-programming-for-hpc-with-the-nvidia-grace-hopper-superchip/
The new hardware developments in NVIDIA Grace Hopper Superchip systems enable some dramatic changes to the way developers approach GPU programming. Most notably, the bidirectional, high-bandwidth, and cache-coherent connection between CPU and GPU memory means that the user can develop their application for both processors while using a single, unified address space. Each processor retains…
Great blog, is it possible for you to share the code with changes to run LULESH and POT3D with unified memory? My understanding is for these, just recompiling with --gpu=unified flag is not enough and some code changes are needed.
If that’s not possible, can you suggest some other HPC workloads that can be just recompiled with unified flag?
Thanks