Deep Learning Profiler 19.08 Release Notes
Deep Learning Profiler (DLProf) is a tool for profiling deep learning models to help data scientists understand and improve performance of their models visually via Tensorboard or by analyzing text reports. It also helps understand resource usage when models are trained.
- Enabling faster generation of Tensorboard event files: Size of the protobuf file used for data collection is smaller so that more profiling data points can be collected and loading Tensorboard event files is faster.
- Kernel report showing usage per kernel: DLProf has added a new report showing CUDA Kernel usage for the benefit of advanced researchers trying to understand which kernels were run when model was trained.
- This software is only accessible in the NGC TensorFlow container.