Hello, I was wondering if Nvidia is planning on releasing CUDA enabled HPL (High-Performance Linpack) code. I have seen a few discussions implying that Nvidia is already working on this.
Is there any even sub-optimal code which can be shared (even under NDA) to be used as a starting point ? Starting from scratch (HPL_+ CUDA + CUBlas) is very lengthy process and involved process and it would be great if there could be some help from Nvidia on this.
As a background, the HPL+CUDA code will target a high-end HPC cluster.
best … dT