Regarding sparse kernel benchmarks in MLIR

Hey Aart and everyone,

I am not sure if this is the right mailing list for discussing MLIR things. Please let me know if there’s a better mailing list for that. I am posting here because the bugzilla is currently non editable, likely because of the migration.

I am interested in working on sparse_tensor dialect and started working on this bug. In my last comment there, I listed some ideas about implementing benchmarks for sparse tensors. I would like to hear your thoughts on whether this is enough to get started or does it need more investigation.

Many thanks,

Hi Saurabh,

You’ll find that most of the discussion about MLIR happens on Discourse, in particular for the Sparse starter projects:


Hi Saurabh,
Thanks for your continuing interest in contributing to this project. I
look forward to your "sparse" contributions :slight_smile:
As Mehdi already replied, most MLIR related (and afaik, *all* sparse)
discussions happen on Discourse.
So we hope to meet you there to hear your ideas for the next steps!
(your last comment on
looks good to me, but I would suggest
copy and pasting this into discourse so we can have a broader
discussion there; as you can tell, I check bugzilla
less frequently than discourse :wink:

Thank you Mehdi and Aart for redirecting me to discourse. I will continue the discussion there.