And on the Nvidia H100, FP32 runs at 67 teraFLOPS....and BFLOAT16/FP16 runs at *...

		phonon on Jan 16, 2023 \| parent \| context \| favorite \| on: Supporting half-precision floats is really annoyin... And on the Nvidia H100, FP32 runs at 67 teraFLOPS....and BFLOAT16/FP16 runs at 1,979 teraFLOPS. https://www.nvidia.com/en-us/data-center/h100/