[cherry-pick][ARITH] Enhance CanonicalSimplify to Simplify ProdDiv and [ci] disable merge

[driazati]

This disables the merge to main behavior for this branch and includes the code #14725 to get this to pass CI. This also disables the docker GPU build which is currently broken on this branch and has no bearing on the release (and is being built erroneously anyways since this PR has no docker changes).

[tvm-bot]

Thanks for contributing to TVM! Please refer to the contributing guidelines https://tvm.apache.org/docs/contribute/ for useful information and tips. Please request code reviews from Reviewers by @-ing them in a comment.

• cc @Mousius, @areusch, @gigiblender, @junrushao, @leandron _{See #10317 for details}

_{Generated by tvm-bot}

[ysh329]

I found there's a test failed:

[driazati]

@tvm-bot rerun

[ysh329]

Hi @driazati , I tried this test locally on commit 6af881c and failed log as below:

$ # keep `test_proddiv_simplify` only
$ python3 ./tests/python/unittest/test_arith_canonical_simplify.py
Traceback (most recent call last):
  File "./tests/python/unittest/test_arith_canonical_simplify.py", line 427, in <module>
    test_proddiv_simplify()
  File "./tests/python/unittest/test_arith_canonical_simplify.py", line 398, in test_proddiv_simplify
    ck.verify(flm(x * 32 * x, x), 0)
  File "./tests/python/unittest/test_arith_canonical_simplify.py", line 29, in verify
    assert tvm.ir.structural_equal(res, expected), "\ndata={}\nres={}\nexpected={}".format(
AssertionError:
data=x * 32 * x % x
res=x * 32 * x % x
expected=0

Can you reproduce this failed case, either?

[driazati]

@ysh329 I believe this PR is exposing a latent bug that was being hidden by the merge before, I think we can fix it by cherry picking #14538, that works locally for me at least

[ysh329]

@ysh329 I believe this PR is exposing a latent bug that was being hidden by the merge before, I think we can fix it by cherry picking #14538, that works locally for me at least

Fixed with cherry-picked PR #14725. It's a bit strange why the previous PR CI test did not reflect this failed test.

[driazati]

@ysh329 I believe this PR is exposing a latent bug that was being hidden by the merge before, I think we can fix it by cherry picking #14538, that works locally for me at least

Fixed with cherry-picked PR #14725. It's a bit strange why the previous PR CI test did not reflect this failed test.

I think because before it was merging the PR from main, so the fix was being applied even though it hadn't been cherry-picked. To avoid the failure in your cherry-pick PR I just added both changes to this one

[ysh329]

@ysh329 I believe this PR is exposing a latent bug that was being hidden by the merge before, I think we can fix it by cherry picking #14538, that works locally for me at least

Fixed with cherry-picked PR #14725. It's a bit strange why the previous PR CI test did not reflect this failed test.

I think because before it was merging the PR from main, so the fix was being applied even though it hadn't been cherry-picked. To avoid the failure in your cherry-pick PR I just added both changes to this one

Great! Thanks for your work. 😆