This document is relevant for: Inf1, Inf2, Trn1, Trn1n

Neuron Plugin for TensorBoard Release Notes#

Known Issues and Limitations - Updated 11/29/2022#

The following are not limitations in the Neuron plugin, but may affect your ability to use TensorBoard.

  • The Neuron plugin for Trn1 (tensorboard-plugin-neuronx) is not compatible with the Neuron plugin for Inf1 (tensorboard-plugin-neuron). Please ensure you only have only the correct package installed.

Neuron Plugin for TensorBoard release [2.6.1.0]#

Date: 12/21/2023

Summary#

  • Now uses local third-party dependencies instead of relying on a CDN.

Neuron Plugin for TensorBoard release [2.5.39.0]#

Date: 7/19/2023

Summary#

  • Minor updates.

Neuron Plugin for TensorBoard release [2.5.37.0]#

Date: 6/14/2023

Summary#

  • Minor updates.

Neuron Plugin for TensorBoard release [2.5.26.0]#

Date: 05/01/2023

Summary#

  • Neuron operator timeline view now includes Neuron Runtime setup/teardown time and a collapsed execution of NC engines and DMA - see Tensorboard tutorial for updated views.

  • Improved execution categorization to include “control” instructions

Neuron Plugin for TensorBoard release [2.5.25.0]#

Date: 03/28/2023

Summary#

  • Supports INF2 and TRN1.

Neuron Plugin for TensorBoard release [2.5.0.0]#

Date: 12/09/2022

Summary#

  • Added support for PyTorch Neuron on Trn1 (torch-neuronx) with new views! Includes a trace view, an operator view, and an operator timeline view. For more info, check out the documentation Neuron Plugin for TensorBoard (Trn1).

    Important

    • You must update to the latest Neuron Tools (aws-neuronx-tools version 2.6 or newer) and install tensorboard-plugin-neuronx for proper functionality of the Neuron plugin on Trn1.

    • For Inf1, please continue to use tensorboard-plugin-neuron. Refer to the getting started guide on Inf1 Neuron Plugin for TensorBoard (Inf1).

Neuron Plugin for TensorBoard release [2.4.0.0]#

Date: 04/29/2022

Summary#

  • Minor updates.

Neuron Plugin for TensorBoard release [2.3.0.0]#

Date: 03/25/2022

Summary#

  • Minor updates.

Neuron Plugin for TensorBoard release [2.2.0.0]#

Date: 10/27/2021

New in this release#

[2.1.2.0]#

Date: 8/12/2021

Summary#

  • Adds support for Neuron Tensorflow 2.5+

[2.1.0.0]#

Date: 5/28/2021

Summary#

  • No major changes or fixes. Released with other Neuron packages.

[2.0.29.0]#

Date: 4/30/2021

Summary#

  • First release Neuron plugin for TensorBoard. Check out it out here: Neuron Plugin for TensorBoard (Inf1).

    • The Neuron plugin is now compatible with TensorBoard 2.0 and higher, in addition to TensorBoard 1.15

    • Provides a centralized place to better understand execution using Neuron SDK.

    • Continues support visualization for TensorFlow graphs, with support for PyTorch and MXNet coming in future releases.

  • Neuron plugin for TensorBoard is supported for Neuron tools >= 1.5, which is first introduced in Neuron v1.13.0 release

  • TensorBoard-Neuron is deprecated, and only supported for Neuron tools <= 1.4.12.0. The final version, 1.4.12.0 is part of Neuron v1.12.2 release.

[1.15.0.1.2.6.0]#

Date: 2/24/2021

Summary#

  • Fix for CVE-2021-3177.

[1.15.0.1.1.1.0]#

Date: 12/23/2020

Summary#

  • Minor internal improvements.

[1.15.0.1.0.615.0]#

Date: 11/17/2020

Summary#

  • Fix issue with viewing chrome trace in Neuron profile plugin in Chrome 80+.

Resolved Issues#

  • Updated dependencies to polyfill missing APIs used by chrome trace in newer browser versions.

[1.15.0.1.0.600.0]#

Date: 09/22/2020

Summary#

  • Minor internal improvements.

[1.15.0.1.0.570.0]#

Date: 08/08/2020

Summary#

  • Minor internal improvements.

[1.15.0.1.0.513.0]#

Date: 07/16/2020

Summary#

  • Minor internal improvements.

[1.15.0.1.0.491.0]#

Date 6/11/2020

Summary#

Fix issue where utilization was missing in the op-profile view.

Resolved Issues#

  • The op-profile view in the Neuron Profile plugin now correctly shows the overall NeuronCore utilization.

[1.15.0.1.0.466.0]#

Date 5/11/2020

Summary#

Fix potential installation issue when installing both tensorboard and tensorboard-neuron.

Resolved Issues#

  • Added tensorboard as a dependency in tensorboard-neuron. This prevents the issue of overwriting tensorboard-neuron features when tensorboard is installed after tensorboard-neuron.

Other Notes#

[1.15.0.1.0.392.0]#

Date 3/26/2020

Summary#

Added ability to view CPU node latency in the Graphs plugin and the Neuron Profile plugins.

Major New Features#

  • Added an aggregate view in addition to the current Neuron subgraph view for both the Graphs plugin and the Neuron Profile plugin.

  • When visualizing a graph executed on a Neuron device, CPU node latencies are available when coloring the graph by “Compute time” using the “neuron_profile” tag.

  • The Neuron Profile plugin now has an overview page to compare time spent on Neuron device versus on CPU.

Other Notes#

  • Requires Neuron-RTD config option “enable_node_profiling” to be set to “true”

[1.15.0.1.0.366.0]#

Date 02/27/2020

Summary#

Reduced load times and fixed crashes when loading large models for visualization.

Resolved Issues#

  • Enable large attribute filtering by default

  • Reduced load time for graphs with attributes larger than 1 KB

  • Fixed a fail to load graphs with many large attributes totaling more than 1 GB in size

[1.15.0.1.0.315.0]#

Date 12/20/2019

Summary#

No major chages or fixes. Released with other Neuron packages.

[1.15.0.1.0.306.0]#

Date 12/1/2019

Summary#

Major New Features#

Resolved Issues#

Known Issues & Limits#

Same as prior release

Other Notes#

[1.15.0.1.0.280.0]#

Date 11/29/2019

Summary#

Initial release packaged with DLAMI.

Resolved Issues#

N/A - first release

Known Issues & Limits#

  • Must install TensorBoard-Neuron by itself, or after regular TensorBoard is installed. If regular Tensorboard is installed after TensorBoard-Neuron, it may overwrite some needed files.

  • Utilization missing in Op Profile due to missing FLOPs calculation (see overview page instead)

  • Neuron Profile plugin may not immediately show up on launch (try reloading the page)

  • Graphs with NeuronOps may take a long time to load due to attribute size

  • Instructions that cannot be matched to a framework layer/operator name show as “” (blank)

  • CPU Usage section in chrome-trace is not applicable

  • Debugger currently supports TensorFlow only

  • Visualization requires a TensorFlow-compatible graph

Other Notes#

This document is relevant for: Inf1, Inf2, Trn1, Trn1n