chenchenghao/DeepGEMM
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: implement sm90 megamoe phase4 l1 epilogue

Browse Source
This commit is contained in:
Xinyi Liu
2026-06-18 01:09:45 +08:00
parent 2bb1756787
commit f3553f976c
6 changed files with 330 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
deep_gemm/mega/__init__.py
View File

@@ -68,6 +68,7 @@ class SymmBuffer:
self.l2_acts, self.l2_acts_sf,
self.expert_recv_count_sum,
self.l1_arrival_count,
self.l2_arrival_mask,
self.token_src_metadata,
self.l1_accum_debug) = buffer_views
else:
@@ -78,6 +79,7 @@ class SymmBuffer:
self.l1_topk_weights = None
self.expert_recv_count_sum = None
self.l1_arrival_count = None
self.l2_arrival_mask = None
self.token_src_metadata = None
self.l1_accum_debug = None
@@ -96,6 +98,7 @@ class SymmBuffer:
self.l2_acts_sf = None
self.expert_recv_count_sum = None
self.l1_arrival_count = None
self.l2_arrival_mask = None
self.token_src_metadata = None
self.l1_accum_debug = None