Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Graph processing is slow, freezes tabs #3264

Open
psybuzz opened this issue Feb 20, 2020 · 1 comment
Open

Graph processing is slow, freezes tabs #3264

psybuzz opened this issue Feb 20, 2020 · 1 comment
Assignees
Labels
core:frontend plugin:graph theme:performance Performance, scalability, large data sizes, slowness, etc.

Comments

@psybuzz
Copy link
Contributor

psybuzz commented Feb 20, 2020

The Graph v1 dashboard currently builds the layout for its graph on the main thread. While the graph is loading (progress bar shows reading pbtxt, parsing, etc) the user can switch to another dashboard and see an unresponsive UI. For example, switching to the Scalars dashboard while the Graph is loading, and being unable to click on anything.

This performance recently regressed from Chromium M79 to M80. In M79, a case that used to freeze the tab for ~4 sec now freezes the TensorBoard for ~50sec, rendering it unusable.

Sample performance trace:

image

We should invest in either

  • Moving heavy JS work from the main thread to a worker
  • Producing a small reproducible case and filing a bug against Chromium to fix the performance regression
@psybuzz psybuzz added core:frontend plugin:graph theme:performance Performance, scalability, large data sizes, slowness, etc. labels Feb 20, 2020
@mherzog01
Copy link

Not sure if it's related, but the following is a simple way to cause large performance problems in TensorBoard graph displays: https://stackoverflow.com/questions/63893714/tensorboard-hangs-when-visualizing-an-object-detection-graph.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
core:frontend plugin:graph theme:performance Performance, scalability, large data sizes, slowness, etc.
Projects
None yet
Development

No branches or pull requests

3 participants