Closed Yuxin-CV closed 1 month ago
Hi, I suggest we modify the FLOPs calculation in the MFU according to the FlashAttention benchmark script.
Specifically, the current calculation for the casual mask can exceed 100% MFU for seq_len = 16k (189 * 2 / 312 = 1.21), which is inaccurate. The FLOPs for the casual mask setting should be divided by 2 when using FlashAttention.
There was some past discussion on this (https://github.com/pytorch/torchtitan/pull/280).
Hi, I suggest we modify the FLOPs calculation in the MFU according to the FlashAttention benchmark script.
Specifically, the current calculation for the casual mask can exceed 100% MFU for seq_len = 16k (189 * 2 / 312 = 1.21), which is inaccurate. The FLOPs for the casual mask setting should be divided by 2 when using FlashAttention.