[Elastic EP] Reuse CUDA graphs across reconfiguration (#54985)

This commit is contained in:
Itay Alroy
2026-09-12 10:26:39 -04:00
committed by GitHub
parent 13e221f830
commit 658c8131c7
22 changed files with 372 additions and 270 deletions
+1 -1
View File
@@ -2,7 +2,7 @@ lmcache >= 0.3.9
# CuPy 14.1.0 imports pytest from cupy.testing._random, which breaks runtime
# images. 14.1.1 fixes the root cause, so only 14.1.0 is excluded.
cupy-cuda13x != 14.1.0
nixl == 1.3.2
nixl == 1.4.1
# CUDA 12 build. On the CUDA 13 image install-kv-connectors.sh swaps this for
# the mooncake-transfer-engine-cuda13 variant of the same version.
mooncake-transfer-engine >= 0.3.12