From eca64989d6a23098b0747016a0ae02f399d74ae1 Mon Sep 17 00:00:00 2001 From: Rui Zhang Date: Mon, 13 Jul 2026 16:46:52 -0700 Subject: [PATCH] add vllm + lmcache on amd mi355x Signed-off-by: Rui Zhang --- benchmarks/single_node/agentic/dsv4_fp4_mi355x_vllm.sh | 10 ++-------- perf-changelog.yaml | 7 +++++++ utils/aiperf | 2 +- 3 files changed, 10 insertions(+), 9 deletions(-) diff --git a/benchmarks/single_node/agentic/dsv4_fp4_mi355x_vllm.sh b/benchmarks/single_node/agentic/dsv4_fp4_mi355x_vllm.sh index e6cc14cb7..467240482 100644 --- a/benchmarks/single_node/agentic/dsv4_fp4_mi355x_vllm.sh +++ b/benchmarks/single_node/agentic/dsv4_fp4_mi355x_vllm.sh @@ -274,14 +274,7 @@ EOF { set +x; } 2>/dev/null unset VLLM_USE_SIMPLE_KV_OFFLOAD - git clone https://github.com/LMCache/LMCache.git - cd LMCache - # https://github.com/LMCache/LMCache/pull/3853 - git checkout 9229067cec0b3a63bb8a39368d101db7ac0bc3c1 - pip install -r requirements/build.txt - pip install grpcio==1.78.0 - CXX=hipcc BUILD_WITH_HIP=1 pip install -e . --no-build-isolation - cd .. + pip install lmcache==0.5.1 python3 -c "import lmcache.integration.vllm.lmcache_mp_connector" >/dev/null @@ -330,6 +323,7 @@ EOF --max-workers "$LMCACHE_MAX_WORKERS" --eviction-policy LRU --supported-transfer-mode "$LMCACHE_TX_MODE" + --no-separate-object-group ) printf '%q ' "${LMCACHE_CMD[@]}" > "$RESULT_DIR/lmcache_command.txt" printf '\n' >> "$RESULT_DIR/lmcache_command.txt" diff --git a/perf-changelog.yaml b/perf-changelog.yaml index d77c35f1f..a7b84eb0e 100644 --- a/perf-changelog.yaml +++ b/perf-changelog.yaml @@ -4781,3 +4781,10 @@ - "Bump image to lmsysorg/sglang-rocm:v0.5.14-rocm720-mi35x-20260708" - "Clean the export envs" pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2198 + +- config-keys: + - dsv4-fp4-mi355x-vllm-agentic + description: + - "Replace pinned-commit LMCache source build (CXX=hipcc BUILD_WITH_HIP=1) with pip install lmcache==0.5.1" + - "Add --no-separate-object-group to the LMCache MP server launch command" + pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2215 diff --git a/utils/aiperf b/utils/aiperf index 0d2aa0572..8473e1545 160000 --- a/utils/aiperf +++ b/utils/aiperf @@ -1 +1 @@ -Subproject commit 0d2aa0572ac685943d38c580675c4a61023581d3 +Subproject commit 8473e1545476c1d91932aa2402b642b416a23df6