Ampere architecture FlashMLA bring-up

[pzhao-eng]

Based on the Hopper architecture FlashMLA and FlashAttentionV2, the FlashMLA based on the Ampere architecture is implemented.
Without using double buffer optimization, the performance can achiev up to 300 GB/s in and 125 TFLOPS on A100.
This is the code repository:
https://github.com/pzhao-eng/FlashMLA

[pzhao-eng]

Update: supports double buffer, performance can achieve up to 360 GB/s in and 146 TFLOPS on A100