Skip to content

Conversation

@AlexeySachkov
Copy link
Contributor

Removed SYCL_SIMPLE_SWIZZLES macro definition, because those swizzles are not used by the test.

Defining SYCL_SIMPLE_SWIZZLES has significant impact on compilation time: the test compiles about twice faster without it on my local machine.

Also fixed warnings about unused variable.

Removed `SYCL_SIMPLE_SWIZZLES` macro definition, because those swizzles
are not used by the test.

Defining `SYCL_SIMPLE_SWIZZLES` has significant impact on compilation
time: the test compiles about twice faster without it on my local
machine.

Also fixed warnings about unused variable.
@AlexeySachkov AlexeySachkov requested a review from a team as a code owner December 5, 2023 13:26
Copy link
Contributor

@maarquitos14 maarquitos14 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM. Just a couple of non-blocking questions:

  • Was the macro used before and at some point it stopped being necessary? Or it was never necessary?
  • Is this something that might be happening in other tests? 2x speedup is a very significant speedup, so if this is indeed happening in other tests, we should replicate the solution there.

@AlexeySachkov
Copy link
Contributor Author

  • Was the macro used before and at some point it stopped being necessary? Or it was never necessary?

I think that in this particular test it was introduced with the initial commit, but haven't been really used.

  • Is this something that might be happening in other tests? 2x speedup is a very significant speedup, so if this is indeed happening in other tests, we should replicate the solution there.

It could, but grep shows that there are only two other E2E tests which use the macro: swizzle_op and vector_operators. I think that both of them can be refactored to still use swizzles (methods like .lo()/.hi(), .odd()/.even() are always available) but avoid defining the macro.
I also did some cleanup related to the macro before in #10137 (and maybe in other PRs which I don't remember).

@steffenlarsen steffenlarsen merged commit a88191c into intel:sycl Dec 6, 2023
@AlexeySachkov AlexeySachkov deleted the private/asachkov/optimize-vector-byte-e2e-test branch May 22, 2024 09:41
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants