Tensor FP16/BF16 performance
PFLOP/sPeak tensor 16-bit throughput per accelerator.
↑ higher is betterEpoch AI
- #01AMD Instinct MI355X2.52
- #02NVIDIA GB300 (Blackwell Ultra)2.50
- #03NVIDIA GB2002.50
- #04Google TPU v7 Ironwood2.31
- #05AMD Instinct MI350X2.31
- #06NVIDIA B300 (Blackwell Ultra)2.25
- #07NVIDIA B2002.25
- #08Intel Habana Gaudi31.68
- #09AMD Instinct MI325X1.31
- #10AMD Instinct MI300X1.31
- #11Microsoft Maia 2001.27
- #12Biren BR1001.02