Skip to content

[CK] Fix MoE 2-stage dispatch for non-128-divisible inter_dim #3973

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
jonahbernard wants to merge 1 commit into
base: main
Choose a base branch
from
+6 −6
Open

[CK] Fix MoE 2-stage dispatch for non-128-divisible inter_dim #3973

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
12 changes: 6 additions & 6 deletions csrc/ck_gemm_moe_2stages_codegen/gen_instances.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -84,7 +84,7 @@
}}
else if (block_m == 128)
{{
if (inter_dim <= 192)
if (inter_dim <= 192 || inter_dim % 128 != 0)
{{
return ck_moe_stage1_gemm<{A0DataType}, {B0DataType}, {AccDataType}, {EDataType}, {CDEElementOp}, V1, 256, 128, 64, 128/sizeof({A0DataType}), 1, 4, {Nswizzle}, {Quant} == static_cast<int>(QuantType::per_Tensor), {MulRoutedWeight}, {ActOP}>;
}}
Expand All @@ -95,7 +95,7 @@
}}
else if (block_m == 256)
{{
if (inter_dim <= 192)
if (inter_dim <= 192 || inter_dim % 128 != 0)
{{
return ck_moe_stage1_gemm<{A0DataType}, {B0DataType}, {AccDataType}, {EDataType}, {CDEElementOp}, V1, 256, 256, 64, 128/sizeof({A0DataType}), 1, 4, {Nswizzle}, {Quant} == static_cast<int>(QuantType::per_Tensor), {MulRoutedWeight}, {ActOP}>;
}}
Expand Down Expand Up @@ -370,7 +370,7 @@
{{
if (block_m == 32)
{{
if (inter_dim <= 192)
if (inter_dim <= 192 || inter_dim % 128 != 0)
{{
return ck_moe_stage2_gemm<{A0DataType}, {B0DataType}, {AccDataType}, {EDataType}, {CDEElementOp}, V1, 256, 32, 64, 64, 1, 4, {Nswizzle}, {Quant} == static_cast<int>(QuantType::per_Tensor), {MulRoutedWeight}, {ActOP}>;
}}
Expand All @@ -381,7 +381,7 @@
}}
else if (block_m == 64)
{{
if (inter_dim <= 192)
if (inter_dim <= 192 || inter_dim % 128 != 0)
{{
return ck_moe_stage2_gemm<{A0DataType}, {B0DataType}, {AccDataType}, {EDataType}, {CDEElementOp}, V1, 256, 64, 128, 64, 1, 4, {Nswizzle}, {Quant} == static_cast<int>(QuantType::per_Tensor), {MulRoutedWeight}, {ActOP}>;
}}
Expand All @@ -392,7 +392,7 @@
}}
else if (block_m == 128)
{{
if (inter_dim <= 192)
if (inter_dim <= 192 || inter_dim % 128 != 0)
{{
return ck_moe_stage2_gemm<{A0DataType}, {B0DataType}, {AccDataType}, {EDataType}, {CDEElementOp}, V3, 256, 128, 64, 64, 1, 4, {Nswizzle}, {Quant} == static_cast<int>(QuantType::per_Tensor), {MulRoutedWeight}, {ActOP}>;
}}
Expand All @@ -403,7 +403,7 @@
}}
else if (block_m == 256)
{{
if (inter_dim <= 192)
if (inter_dim <= 192 || inter_dim % 128 != 0)
{{
return ck_moe_stage2_gemm<{A0DataType}, {B0DataType}, {AccDataType}, {EDataType}, {CDEElementOp}, V3, 256, 256, 128, 64, 1, 4, {Nswizzle}, {Quant} == static_cast<int>(QuantType::per_Tensor), {MulRoutedWeight}, {ActOP}>;
}}
Expand Down
Loading