xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-02 05:23:30 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Bugfix] Optimize CpuGpuBuffer initialization (#25447)

Browse Source 
Signed-off-by: Naman Lalit <nl2688@nyu.edu>
This commit is contained in:
Naman Lalit 2025-09-26 17:07:36 -07:00 committed by GitHub
parent dc48ba0c75
commit 4e33a7ea85
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/v1/utils.py
View File

@ -117,7 +117,7 @@ class CpuGpuBuffer:
dtype=dtype, dtype=dtype,
device="cpu", device="cpu",
pin_memory=pin_memory) pin_memory=pin_memory)
self.gpu = self.cpu.to(device) self.gpu = torch.zeros_like(self.cpu, device=device)
self.np: np.ndarray self.np: np.ndarray
# To keep type hints simple (avoiding generics and subclasses), we # To keep type hints simple (avoiding generics and subclasses), we
# only conditionally create the numpy array attribute. This can cause # only conditionally create the numpy array attribute. This can cause