chenchenghao/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 11 Packages Projects Releases Wiki Activity

[Diffusion] [NPU] Fix CI run (#18921)

Browse Source
This commit is contained in:
Makcum888e
2026-02-17 16:54:19 +03:00
committed by GitHub
parent 8bb1037796
commit 2aa0db7d9c
4 changed files with 9 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
python/sglang/multimodal_gen/runtime/managers/gpu_worker.py
View File

@@ -154,8 +154,8 @@ class GPUWorker:
output_batch.timings.record_memory_snapshot("mem_analysis", final_snapshot)
# for details on max_memory_reserved: https://docs.pytorch.org/docs/stable/generated/torch.cuda.memory.max_memory_reserved.html
peak_reserved_bytes = torch.cuda.max_memory_reserved()
peak_allocated_bytes = torch.cuda.max_memory_allocated()
peak_reserved_bytes = torch.get_device_module().max_memory_reserved()
peak_allocated_bytes = torch.get_device_module().max_memory_allocated()
output_batch.peak_memory_mb = peak_reserved_bytes / (1024**2)
peak_reserved_gb = peak_reserved_bytes / (1024**3)