Support releasing CUDA graph memory when paused (#7873)

Co-authored-by: ryang-max <y1cunhui.yang@gmail.com>
Co-authored-by: ryang <38470282+ryang-max@users.noreply.github.com>
This commit is contained in:
fzyzcjy
2025-10-28 14:40:50 +08:00
committed by GitHub
co-authored by ryang-max ryang
parent d2b8c4123e
commit 691c8534cf
7 changed files with 94 additions and 15 deletions
+7
View File
@@ -1,3 +1,10 @@
# GPU Memory Types
GPU_MEMORY_TYPE_KV_CACHE = "kv_cache"
GPU_MEMORY_TYPE_WEIGHTS = "weights"
GPU_MEMORY_TYPE_CUDA_GRAPH = "cuda_graph"
GPU_MEMORY_ALL_TYPES = [
GPU_MEMORY_TYPE_KV_CACHE,
GPU_MEMORY_TYPE_WEIGHTS,
GPU_MEMORY_TYPE_CUDA_GRAPH,
]