3 ms·
It's only ~1000 LoC, seems like a pretty good case study to port over to other runtimes and show they can stand up to CUDA.
by jsheard 3y ago
It's only ~1000 LoC, seems like a pretty good case study to port over to other runtimes and show they can stand up to CUDA.