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
{{ message }}
Repository navigation
How does OEq compare to FlashTP for higher l operations? #171
FlashTP seems to outperform CuEq by a lot for higher angular momentum ops. They sadly do not benchmark against OEq this is just a general inquiry as to whether this is on the mantainers radar.
Hmm- we did some benchmarks on Nequip / MACE-style configs and for inference we outperform, training we are competitive against FlashTP.
Worth noting: their paper compares against an older version of CuEq. CuEq is now much more competitive for higher irreps (see our roofline plot in the paper, where we go up to L=7.
Can you define "higher angular momentum"? Some sample input representations would be helpful.
Below: some benchmarks for flashTP, forward and backward, on what I believe are the irreps used in their experiments. Tested fused convolutions on 3 different graphs.
Takeaways:
The latest version of cuE (which was released in March) is much more competitive than the one benchmarked in the FlashTP paper (v0.2.0), and generally has the fastest backward pass.
We are very competitive on the backward pass and often 2-3x faster than flashTP.
Not shown are double-backward kernels for training. Haven't benchmarked, but given recent improvements to our kernel fusion strategy, I'd wager we'd beat flashTP and hold our own (although probably worse) against cuE.
Activity
vbharadwaj-bk commentedon Jan 12, 2026
Hmm- we did some benchmarks on Nequip / MACE-style configs and for inference we outperform, training we are competitive against FlashTP.
Worth noting: their paper compares against an older version of CuEq. CuEq is now much more competitive for higher irreps (see our roofline plot in the paper, where we go up to L=7.
Can you define "higher angular momentum"? Some sample input representations would be helpful.
CompRhys commentedon Jan 12, 2026
by higher angular momentum I meant l>2 from the table 2 in the paper.
vbharadwaj-bk commentedon Jan 12, 2026
Sure - we can add to our radar to extract / benchmark some of those configs.
vbharadwaj-bk commentedon Jan 20, 2026
Below: some benchmarks for flashTP, forward and backward, on what I believe are the irreps used in their experiments. Tested fused convolutions on 3 different graphs.
Takeaways:
vbharadwaj-bk commentedon Jan 20, 2026
Let's convert this to a discussion now that the main issue has been answered. Can follow up there if needed.