Ayush Baiswar

Weight Decompression and BF16 performance comparison

2

Executed BenchDNN (main) for _MatMul weight decompression (src:bf16, wei:s4)_ and compared with _BF16 (src:bf16, wei:bf16) MatMul_ with different **tags (ab, any)** for some M,N,K dimensions and observed the following behavior...