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
We should add the largest systolic array that we can push through the synthesis flow as a part of this test suite. @calebmkim might already know the answer: what is the largest systolic array that meets timing?
The text was updated successfully, but these errors were encountered:
Okay, do you have the timing reports for that somewhere? I think we should:
Pinpoint what is the bottleneck exactly
Do a short literature search and see what are the largest systolic arrays people have been able to fit onto an FPGA. Maybe the PolySA paper would be a good place to look for this
I want to propose area optimization as a key theme for 2024 (https://github.com/orgs/calyxir/discussions/1825) and so I think we should pick a particular set of targets, like being able to synthesize a 128x128 SA, and see how far we can get with it.
We should add the largest systolic array that we can push through the synthesis flow as a part of this test suite. @calebmkim might already know the answer: what is the largest systolic array that meets timing?
The text was updated successfully, but these errors were encountered: