feat: integrate deepgemm into EPMoE (#6821)

Co-authored-by: tianqilin.99 <tianqilin.99@bytedance.com> Co-authored-by: TianQiLin666666 <1834987979@qq.com> Co-authored-by: Cheng Wan <54331508+ch-wan@users.noreply.github.com>
2025-06-23 16:38:58 +08:00
parent 30ceccc74a
commit 506c4928f5
5 changed files with 336 additions and 3 deletions
--- a/sgl-kernel/benchmark/bench_moe_ep_post_reorder.py
+++ b/sgl-kernel/benchmark/bench_moe_ep_post_reorder.py
@@ -77,6 +77,7 @@ def benchmark(batch_size, provider):
                end_expert_id,
                topk,
                hidden_size,
+                0,
                block_size,
            )

--- a/sgl-kernel/tests/test_ep_moe_post_reorder_kernel.py
+++ b/sgl-kernel/tests/test_ep_moe_post_reorder_kernel.py
@@ -85,6 +85,7 @@ def run_triton_kernel(
        end_expert_id,
        topk,
        hidden_size,
+        0,
        block_size,
    )
    return output