Ian Paul

> bounty locked, good luck! Thanks! > Your test output is correct but the implementation can be simpler, Can you rewrite all gated stores with IF? There shouldn't be two...

> I think if you do the graph rewriting right you don't need to worry about the queue. Ok, I'll give it a closer look and figure it out. Thanks!

W.r.t. point 1: Yes I saw that too after rebasing and having those tests now uncommented. I think it should be fixed now by joining the IFs in `delete_redundant_gates`. LMK...

I'm not sure about the norms in this repo, but would you like me to clean up the commit history into only a few isolated commits, or do you normally...

I just tried now, seems to be only marginally slower. Not sure how much of a diff is acceptable. (edit: also, is there a more scientific way you were looking...

More than happy to split into two separate PRs, but `Rewrite all gated stores to UOps.IF` PR won't pass tests until `Add merge_gates` is included. Are you ok with me...

It doesn't pass because of the issue when a STORE is dependent on a variable within a RANGE but the IF isn't, so the IF gets written first but the...

Huh interesting. I will take a closer look, but any tips on debugging either of those tests without an AMD machine? `TestLinearizerFailures.test_failure_43` passes on my local METAL mac and on...

I think the local reduce gating should be fixed and process replay should now be working. Thanks for the pointers on debugging. It took me a while to wrap my...

I found that the cleanest way to have the IF to correctly reference the STORE's 3rd src was to have it happen after the expand. When I tried to add...