Newest benchmark of #dyGiLa from LUMI-G Mi250x nodes @lumi-supercomputer.eu! Strong scaling, weak scaling, checkpoints are all tested. Beside #dyGiLa's perfect strong scaling, the weak scaling is also super impressive. 😃 #CFTHILA Full report @ dygila.github.io/benchmark

dyGiLa, our #CFTHILA based LFT p-wave superfluid simulation project supported by #QUESTDMC, has perfect strong scaling on @AMDInstinct
Mi250x GPUs! Here is fresh benchmark data done on LUMI @lumi-supercomputer.eu , thanks!😃 Read details in homepage: dygila.github.io/benchmark

Thanks to @csc.fi for the great support and efforts from Aaron, Kari, Jaakko—our pilot project "Evolving HILA" has been a great success. #CFTHILA has achieved a substantial performance boost on GH200 clusters, along with significant communication improvements through #nvshmem.

Is #dyGiLa ready for the NVIDIA #GH200 cluster? Absolutely, yes!
Thanks to the effort of #HILA development team and hardware support from nvidia 🙌, #dyGiLa 's strong scaling benchmark on NVIDIA's internal GH200 cluster with #CUDA 13.0 is here:👇👉 dygila.github.io/benchmark
@csc.fi

#CFTHILA