[CPU] Update torch 2.9.1 for CPU backend (#29664)

Signed-off-by: jiang1.li <jiang1.li@intel.com>

.buildkite/scripts/hardware_ci/run-cpu-test.sh

@@ -21,8 +21,8 @@ trap remove_docker_container EXIT
 remove_docker_container
 
 # Try building the docker image
-numactl -C "$CORE_RANGE" -N "$NUMA_NODE" docker build --tag cpu-test-"$NUMA_NODE" --target vllm-test -f docker/Dockerfile.cpu .
+numactl -C "$CORE_RANGE" -N "$NUMA_NODE" docker build --progress plain --tag cpu-test-"$NUMA_NODE" --target vllm-test -f docker/Dockerfile.cpu .
-numactl -C "$CORE_RANGE" -N "$NUMA_NODE" docker build --build-arg VLLM_CPU_DISABLE_AVX512="true" --tag cpu-test-"$NUMA_NODE"-avx2 --target vllm-test -f docker/Dockerfile.cpu .
+numactl -C "$CORE_RANGE" -N "$NUMA_NODE" docker build --progress plain --build-arg VLLM_CPU_DISABLE_AVX512="true" --tag cpu-test-"$NUMA_NODE"-avx2 --target vllm-test -f docker/Dockerfile.cpu .
 
 # Run the image, setting --shm-size=4g for tensor parallel.
 docker run -itd --cpuset-cpus="$CORE_RANGE" --cpuset-mems="$NUMA_NODE" --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --privileged=true -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=16 --env VLLM_CPU_CI_ENV=1 -e E2E_OMP_THREADS="$OMP_CORE_RANGE" --shm-size=4g --name cpu-test-"$NUMA_NODE" cpu-test-"$NUMA_NODE"

csrc/cpu/utils.cpp

@@ -51,12 +51,13 @@ std::string init_cpu_threads_env(const std::string& cpu_ids) {
       if (node_id != -1) {
         node_ids.insert(node_id);
       }
-      TORCH_WARN(node_id == mem_node_id, "CPU ", cpu_id, " is on NUMA node ",
+      if (node_id != mem_node_id) {
-                 node_id, ", but CPU ", omp_cpu_ids.front(),
+        TORCH_WARN("CPU ", cpu_id, " is on NUMA node ", node_id, ", but CPU ",
-                 " is on NUMA node ", mem_node_id,
+                   omp_cpu_ids.front(), " is on NUMA node ", mem_node_id,
-                 ". All CPUs should be on the same NUMA node for optimal "
+                   ". All CPUs should be on the same NUMA node for optimal "
-                 "performance. Memory will be bound to NUMA node ",
+                   "performance. Memory will be bound to NUMA node ",
-                 mem_node_id, ".");
+                   mem_node_id, ".");
+      }
     }
     // Concatenate all node_ids into a single comma-separated string
     if (!node_ids.empty()) {

docker/Dockerfile.cpu

@@ -132,7 +132,7 @@ RUN --mount=type=bind,src=requirements/test.in,target=requirements/test.in \
       esac; \
     }; \
     remove_packages_not_supported_on_aarch64 && \
-    sed -i 's/^torch==.*/torch==2.8.0/g' requirements/cpu-test.in && \
+    sed -i 's/^torch==.*/torch==2.9.1/g' requirements/cpu-test.in && \
     sed -i 's/torchaudio.*/torchaudio/g' requirements/cpu-test.in && \
     sed -i 's/torchvision.*/torchvision/g' requirements/cpu-test.in && \
     uv pip compile requirements/cpu-test.in -o requirements/cpu-test.txt --index-strategy unsafe-best-match --torch-backend cpu

requirements/cpu-build.txt

@@ -4,9 +4,8 @@ packaging>=24.2
 setuptools>=77.0.3,<81.0.0
 setuptools-scm>=8
 --extra-index-url https://download.pytorch.org/whl/cpu
-torch==2.8.0+cpu; platform_machine == "x86_64" or platform_machine == "s390x"
+torch==2.9.1+cpu; platform_machine == "x86_64" or platform_machine == "s390x"
-torch==2.9.0; platform_system == "Darwin"
+torch==2.9.1; platform_system == "Darwin" or platform_machine == "ppc64le" or platform_machine == "aarch64"
-torch==2.8.0; platform_machine == "ppc64le" or platform_machine == "aarch64"
 scons; platform_machine == "aarch64"    # needed to build Arm Compute Library (ACL)
 wheel
 jinja2>=3.1.6

requirements/cpu.txt

@@ -4,25 +4,18 @@
 numba == 0.61.2; platform_machine != "s390x" # Required for N-gram speculative decoding
 
 # Dependencies for CPUs
-packaging>=24.2
-setuptools>=77.0.3,<81.0.0
 --extra-index-url https://download.pytorch.org/whl/cpu
-torch==2.8.0+cpu; platform_machine == "x86_64" or platform_machine == "s390x"
+torch==2.9.1+cpu; platform_machine == "x86_64" or platform_machine == "s390x"
-torch==2.9.0; platform_system == "Darwin"
+torch==2.9.1; platform_system == "Darwin" or platform_machine == "ppc64le" or platform_machine == "aarch64"
-torch==2.8.0; platform_machine == "ppc64le" or platform_machine == "aarch64"
 
 # required for the image processor of minicpm-o-2_6, this must be updated alongside torch
-torchaudio; platform_machine != "ppc64le" and platform_machine != "s390x"
+torchaudio; platform_machine != "s390x"
-torchaudio==2.8.0; platform_machine == "ppc64le"
 
 # required for the image processor of phi3v, this must be updated alongside torch
-torchvision; platform_machine != "ppc64le" and platform_machine != "s390x"
+torchvision; platform_machine != "s390x"
-torchvision==0.23.0; platform_machine == "ppc64le"
-datasets # for benchmark scripts
 
 # Intel Extension for PyTorch, only for x86_64 CPUs
 intel-openmp==2024.2.1; platform_machine == "x86_64"
-triton==3.2.0; platform_machine == "x86_64" # Triton is required for torch 2.6+cpu, as it is imported in torch.compile.
 
 # Use this to gather CPU info and optimize based on ARM Neoverse cores
 py-cpuinfo; platform_machine == "aarch64"

vllm/platforms/cpu.py

@@ -384,7 +384,6 @@ class CpuPlatform(Platform):
 
     @classmethod
     def is_pin_memory_available(cls) -> bool:
-        logger.warning("Pin memory is not supported on CPU.")
         return False
 
     @classmethod