NVIDIA Optimized Frameworks
NVIDIA Optimized Frameworks (Latest Release) Download PDF

Deep Learning Profiler 21.07 Release Notes

Description

DLProf Release for 21.07, available in the NVIDIA TensorFlow 1.x, TensorFlow 2.x, and PyTorch NGC containers, and as a Python Wheel on the NVIDIA PY Index.

Driver Requirements

Release 21.07 is based on NVIDIA CUDA 11.4.0, which requires NVIDIA Driver release 470 or later. However, if you are running on Tesla (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), or 460.27 (or later R460). The CUDA driver's compatibility package only supports particular drivers. For a complete list of supported drivers, see the CUDA Application Compatibility topic. For more information, see CUDA Compatibility and Upgrades.

New Features

The key features of DLProf v1.3.0 / r21.07 are:

  • Released in the TensorFlow 1.x 21.07, TensorFlow 2.x 21.07 and PyTorch 21.07 NGC container.
  • Latest DLProf build is based on TensorFlow 1.15.5, TensorFlow 2.5.0, PyTorch 1.10.0, TensorRT 8.0, and Nsight Systems 2021.2.4.12.
  • Added database versioning and backwards compatibility starting with v1.2 databases.
  • Added underutilized GPU detector.

Known Issues

  • This software is only supported for TensorFlow 1.15, TensorFlow 2.5.0, PyTorch 1.10.0, and TensorRT 8.0.

Resolved Issues

  • Fixed issue where memory could get corrupted if not all GPUs were used
© Copyright 2024, NVIDIA. Last updated on Jul 26, 2024.