ICLDisco / dplasma

DPLASMA is a highly optimized, accelerator-aware, implementation of a dense linear algebra package for distributed heterogeneous systems. It is designed to deliver sustained performance for distributed systems where each node featuring multiple sockets of multicore processors, and if available, accelerators, using the PaRSEC runtime as a backend.

Other

11 stars 9 forks source link

Update the GH runner same way we did for PaRSEC #75

Closed abouteiller closed 1 year ago

abouteiller commented 1 year ago

The runner now works correctly, looks like we have a guenuine problem with gemm_lowmem_gpu

abouteiller commented 1 year ago

77 fixes divide by zero, but not the incorrect results with NT, TN, TT

abouteiller commented 1 year ago

I disabled the lowmen tests (from the CI, they still run manually)