Skip to content

Support GPU tensors in tensor_data() to enable GPU-accelerated multimodal preprocessing #30667

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
storyicon wants to merge 1 commit into from
+8 −2
Closed

Support GPU tensors in tensor_data() to enable GPU-accelerated multimodal preprocessing #30667

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
10 changes: 8 additions & 2 deletions vllm/v1/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -405,10 +405,16 @@ def tensor_data(tensor: torch.Tensor) -> memoryview:
"""Get the raw data of a tensor as a uint8 memoryview, useful for
serializing and hashing.

Note:
The `.cpu()` call is necessary to handle GPU tensors, as PyTorch's
`.numpy()` method only works on CPU tensors. This involves a
device-to-memory (device2mem) transfer for GPU tensors. For tensors
already on CPU, `.cpu()` is a no-op.

Args:
tensor: The input tensor.
tensor: The input tensor (can be on CPU or GPU).

Returns:
A memoryview of the tensor data as uint8.
"""
return tensor.flatten().contiguous().view(torch.uint8).numpy().data
return tensor.flatten().contiguous().view(torch.uint8).cpu().numpy().data