Delete opt_mul_scalar_out

[swolchok]

The handwritten optimized code is similar to what we should be getting from the optimized portable op, as follows.

handwritten optimized code:

• if the input type matches the output type, then perform a vectorized loop
• otherwise, generate specific mixed-dtype kernels, which aren't vectorized.

optimized portable op:

• if the input type matches the output type, then perform a vectorized loop. (dtype_specialized_elementwise_fn_impl in elementwise_util.h)
• otherwise, generate one specific kernel per compute type. those
  kernels use non-inlined function calls to do loads and stores,
  trading off performance for a significant size reduction. (apply_elementwise_fn_generic_impl in elementwise_util.h)

Both cases in the portable op variant also use parallel_for.

I attempted to do a performance test, but I found that
torch.mul(some_tensor, 2.0) is exported as a call to mul.Tensor,
not
mul.Scalar. 41e7ffa
added the ability to pass our tests if we do emit mul.Scalar for this,
but the follow-up diff to make that happen seems not to have
landed. So, I think another reason to delete this is that (if I
understand correctly) it's not used, therefore we don't have specific
knowledge that we need it to exist and we can't just use the optimized
portable op.

[swolchok]

Stack from ghstack (oldest at bottom):

• -> Delete opt_mul_scalar_out #12145

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/12145

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

✅ You can merge normally! (2 Unrelated Failures)

As of commit 80a49db with merge base 7e28a04 ():

BROKEN TRUNK - The following jobs failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / test-models-linux (mobilebert, portable, linux.2xlarge) / linux-job (gh) (trunk failure)
  ##[error]The operation was canceled.
• pull / test-models-linux (mobilebert, xnnpack-quantization-delegation, linux.2xlarge) / linux-job (gh) (trunk failure)
  ##[error]The operation was canceled.

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[kimishpatel]

I trust that the summary holds true. I think the question of mult-threading remains in that it is not always beneficial to multithread. Our portable implementation should account for that else for some case, I wont be surprised if I saw pref regression