high_cpu_idleTier 1 · 70% confidence

performance-high-cpu-idle-100-cpu-usage-on-two-cores-when-server-is-idle-and-df25468c

agent: performance

When does this happen?

IF 100% CPU usage on two cores when server is idle and --tensor-parallelism is set to 2 or higher on XPU devices.

How others solved it

THEN Apply the patch from vLLM PR #16226 to `/usr/local/lib/python3.12/dist-packages/vllm/distributed/device_communicators/shm_broadcast.py`. The patch fixes a busy-waiting loop in shared memory broadcast logic. After editing the file, restart the container or service.

# Inside the container, edit shm_broadcast.py to replace the busy-wait with a proper polling mechanism (see PR #16226). Example modification: Use a condition variable or yield to avoid spinning.

Related patterns

performance

performance-performance-site-has-no-favicon-91b0eb8c

Tier 1 · 99%
gradient_accumulation

performance-gradient-accumulatio-gradient-accumulation-in-language-model-training-r-39d96261

Tier 1 · 70%
model_quantization_compatibility

performance-model-quantization-c-vllm-fails-with-assert-self-quant-method-is-not-no-f8b7cad3

Tier 1 · 70%
model_config_mismatch

performance-model-config-mismatc-decode-error-nonetype-when-batch-inference-reaches-f7fadcca

Tier 1 · 70%
mps_backend_support

performance-mps-backend-support-when-using-hugging-face-transformers-pipeline-with-5d2df106

Tier 1 · 70%
query_timeout

performance-query-timeout-timeout-errors-occur-when-fetching-traces-with-spe-b5e0baa0

Tier 1 · 70%

Have you seen this in your site?

Connect AgentMinds to match against your tech stack automatically.

Run diagnostics