DeepSpeed
24a41afa - Add fused triton kernel for swiglu (#8244)

Commit
5 days ago
Add fused triton kernel for swiglu (#8244) **Performance**: SwiGLU Benchmark **Configuration:** `hidden=1024`, `dtype=bf16` | Tokens | Variant | Forward (µs) | Forward + Backward (µs) | Backward (µs) | |-------:|:--------|-------------:|------------------------:|--------------:| | 8,192 | `triton_fused` | 13.05 | 36.29 | 23.24 | | 8,192 | `eager` | 22.53 | 61.47 | 38.94 | | 16,384 | `triton_fused` | 26.25 | 67.65 | 41.40 | | 16,384 | `eager` | 44.97 | 120.95 | 75.98 | | 65,536 | `triton_fused` | 94.71 | 251.60 | 156.89 | | 65,536 | `eager` | 163.12 | 444.64 | 281.52 | --------- Signed-off-by: Hongwei Chen <hongweichen@microsoft.com>
Author
Parents
Loading