A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 3, 19:00 UTC
[PD] Bump NIXL connector dependency to 1.x (#42364)
Signed-off-by: Alec Flowers <aflowers@nvidia.com>
Alec committed
May 12, 2026 at 18:05 UTC
07534b8782757b069652f56d2d32f8fb8256adfc
2 files changed
+2
-3
.buildkite/ci_config.yaml
+1
@@ -8,6 +8,7 @@ run_all_patterns:
8
- "CMakeLists.txt"
9
- "requirements/common.txt"
10
- "requirements/cuda.txt"
11
+ - "requirements/kv_connectors.txt"
12
- "requirements/build/cuda.txt"
13
- "requirements/test/cuda.txt"
14
- "setup.py"
requirements/kv_connectors.txt
+1
-3
@@ -1,5 +1,3 @@
1
lmcache >= 0.3.9
2
-nixl[cu13] >= 0.7.1, <= 0.10.1 # Required for disaggregated prefill
3
-nixl-cu12 >= 0.7.1, <= 0.10.1
4
-nixl-cu13 >= 0.7.1, <= 0.10.1
2
+nixl >= 1.1.0 # Required for disaggregated prefill
3
mooncake-transfer-engine >= 0.3.8