You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hey guys, we are currently using rooflines to identify where the bottlenecks in inference performance lie.
However, when using xprof or similar tools to obtain actual arithmetic intensity data for the Pallas kernel, the results consistently show as 0.00 or Unknown. Are there any better methods for acquiring this data?
(The RPA-related data in the figure refers to the ragged_paged_attention kernel.)
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Hey guys, we are currently using rooflines to identify where the bottlenecks in inference performance lie.

However, when using xprof or similar tools to obtain actual arithmetic intensity data for the Pallas kernel, the results consistently show as 0.00 or Unknown. Are there any better methods for acquiring this data?
(The RPA-related data in the figure refers to the ragged_paged_attention kernel.)
Beta Was this translation helpful? Give feedback.
All reactions