New Release: Ultralytics v8.4.152

Discussion
, ,
1

:rocket: Ultralytics v8.4.152 Released

:glowing_star: Quick Overview

Ultralytics v8.4.152 improves system monitoring by adding cached NVIDIA driver and driver-supported CUDA version information to SystemLogger. This makes GPU environment details easier to collect while keeping repeated metric collection efficient and reliable.

:sparkles: New Features

  • :desktop_computer: Added top-level driver_version and cuda_version fields to system metrics on NVIDIA-enabled systems.
  • :magnifying_glass_tilted_left: Monitoring systems can now receive NVIDIA driver and CUDA compatibility details directly from SystemLogger.
  • :pushpin: cuda_version reports the CUDA version supported by the installed NVIDIA driver, matching nvidia-smi. It does not represent the installed CUDA toolkit or PyTorch build version.

:high_voltage: Improvements

  • NVIDIA driver and CUDA version information is queried once when SystemLogger starts.
  • Cached values avoid repeated NVML calls during metric collection.
  • Validation confirms only one query per version for each logger instance.
  • Testing on an eight-GPU server confirmed accurate version reporting, reliable cumulative and rate-mode collection, and strong performance.

:shield: Compatibility and Reliability

  • GPU utilization monitoring continues working when version information is unavailable.
  • Missing NVIDIA version fields are simply omitted rather than interrupting metric collection.
  • Non-NVIDIA and macOS environments remain compatible.
  • The package version has been updated from 8.4.151 to 8.4.152.

:hammer_and_wrench: What Changed

Review every update in the full v8.4.151 to v8.4.152 changelog.

:package: Try It Today

Upgrade to the latest release:

pip install -U ultralytics

Explore the Ultralytics v8.4.152 release, try the enhanced system metrics in your monitoring workflows, and share your feedback with the Ultralytics YOLO community. :raising_hands: