Skip to content

[None][feat] Staircase: Rubin GLM Optimized Low Latency (Conc up to 56) #21380

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
dongfengy wants to merge 1 commit into
base: main
Choose a base branch
from
+14,563 −2,924
Open

[None][feat] Staircase: Rubin GLM Optimized Low Latency (Conc up to 56) #21380

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 4 additions & 4 deletions .github/CODEOWNERS
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -522,10 +522,10 @@
/tests/unittest/_torch/modeling_v2/models/glm_moe_dsa/glm_5_2_nvfp4__sm_103__tep4 @erictsai-nv

# GLM SM107 target and its native kernels/bindings
/tensorrt_llm/_torch/_experimental/modeling_v2/models/glm_moe_dsa/glm_5_2_nvfp4__sm_107__tp16 @dongfengy @Tabrizian @xwang233
/tests/unittest/_torch/modeling_v2/models/glm_moe_dsa/glm_5_2_nvfp4__sm_107__tp16 @dongfengy @Tabrizian @xwang233
/cpp/tensorrt_llm/kernels/glmMoeDsa @dongfengy @Tabrizian @xwang233
/cpp/tensorrt_llm/thop/glmMoeDsa/ @dongfengy @Tabrizian @xwang233
/tensorrt_llm/_torch/_experimental/modeling_v2/models/glm_moe_dsa/glm_5_2_nvfp4__sm_107__low_latency @dongfengy @Tabrizian @xwang233
/tests/unittest/_torch/modeling_v2/models/glm_moe_dsa/glm_5_2_nvfp4__sm_107__low_latency @dongfengy @Tabrizian @xwang233
/cpp/tensorrt_llm/kernels/glmMoeDsaRubin @dongfengy @Tabrizian @xwang233
/cpp/tensorrt_llm/thop/glmMoeDsaRubin/ @dongfengy @Tabrizian @xwang233

# Uncomment only on release branches (e.g. release/0.19):
# * @NVIDIA/trt-llm-release-branch-approval
838 changes: 0 additions & 838 deletions cpp/tensorrt_llm/kernels/glmMoeDsa/glmLowMBf16Moe.cu
View file
Open in desktop

This file was deleted.

574 changes: 0 additions & 574 deletions cpp/tensorrt_llm/kernels/glmMoeDsa/glmMoeFusedRouter.cu
View file
Open in desktop

This file was deleted.

244 changes: 0 additions & 244 deletions cpp/tensorrt_llm/kernels/glmMoeDsa/glmRouterGemm.cu
View file
Open in desktop

This file was deleted.

2 changes: 1 addition & 1 deletion ...lm/kernels/glmMoeDsa/glmDenseMlpSmallM.cu → ...rnels/glmMoeDsaRubin/glmDenseMlpSmallM.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,7 +35,7 @@
// Programmatic dependent launch: activations are read only after griddepcontrol.wait; the dependents are released at
// entry or after the weight copies are issued (Trigger). Each grid fits in one wave of resident CTAs.

#include "tensorrt_llm/kernels/glmMoeDsa/glmDenseMlpSmallM.h"
#include "tensorrt_llm/kernels/glmMoeDsaRubin/glmDenseMlpSmallM.h"

#include <cstdint>
#include <tuple>
Expand Down
0 ...llm/kernels/glmMoeDsa/glmDenseMlpSmallM.h → ...ernels/glmMoeDsaRubin/glmDenseMlpSmallM.h
View file
Open in desktop
File renamed without changes.
Loading
Loading