This is an archive of the discontinued LLVM Phabricator instance.

[mlir][sparse] fully implement sparse tensor to sparse tensor conversions
ClosedPublic

Authored by aartbik on Aug 25 2021, 11:54 AM.

Diff Detail

Event Timeline

aartbik created this revision.Aug 25 2021, 11:54 AM
aartbik requested review of this revision.Aug 25 2021, 11:54 AM
aartbik updated this revision to Diff 368706.Aug 25 2021, 12:42 PM

lint issue

aartbik updated this revision to Diff 368708.Aug 25 2021, 12:46 PM

rebase with main

bixia added inline comments.Aug 25 2021, 1:30 PM
mlir/test/Integration/Dialect/SparseTensor/CPU/sparse_conversion.mlir
118–168

I wonder where there is a better way to assert %10 == %20, %11 == %21, so that we don't have to repeat the dump output in this block, which is the same as line 110-115.
For example, we can just print a boolean (%10 == %20) the FileCheck the boolean, if we can't think of a better way?

aartbik added inline comments.Aug 25 2021, 2:34 PM
mlir/test/Integration/Dialect/SparseTensor/CPU/sparse_conversion.mlir
118–168

Yeah, fair enough. After working with FileCheck for close to two years now, I have a bit of the "when you have a hammer everything looks like a nail" approach to checking with output. I will rewrite this into a form that uses more intuitive comparisons, so this can become a simple pass/fail check. Thanks for the suggestion!

aartbik updated this revision to Diff 368806.Aug 25 2021, 10:30 PM

exhaustive but silent checks

bixia accepted this revision.Aug 27 2021, 2:30 PM
This revision is now accepted and ready to land.Aug 27 2021, 2:30 PM