Skip to content

Fix workspace allocation for f8f8bf16_rowwise_batched #5098

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

Closed
wants to merge 1 commit into from
Closed

Fix workspace allocation for f8f8bf16_rowwise_batched #5098

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
5 changes: 3 additions & 2 deletions .../quantize/cutlass_extensions/f8f8bf16_rowwise_batched/f8f8bf16_rowwise_batched_common.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -274,7 +274,8 @@ at::Tensor f8f8bf16_rowwise_batched_impl(
size_t workspace_size = Gemm::get_workspace_size(arguments);

// Allocate workspace memory
cutlass::device_memory::allocation<uint8_t> workspace(workspace_size);
at::Tensor workspace =
at::empty(workspace_size, XQ.options().dtype(at::kByte));

// Check the problem size is supported or not
cutlass::Status status = gemm.can_implement(arguments);
Expand All @@ -283,7 +284,7 @@ at::Tensor f8f8bf16_rowwise_batched_impl(
}

// Initialize CUTLASS kernel with arguments and workspace pointer
status = gemm.initialize(arguments, workspace.get());
status = gemm.initialize(arguments, workspace.data_ptr());
if (status != cutlass::Status::kSuccess) {
throw std::runtime_error("cutlass cannot initialize");
}
Expand Down
Loading