[vectorIntrinsics] RFR: 8343689: AArch64: Optimize MulReduction implementation

Paul Sandoz psandoz at openjdk.org
Thu Jan 16 16:53:50 UTC 2025


On Thu, 16 Jan 2025 16:38:15 GMT, Andrew Haley <aph at openjdk.org> wrote:

> > I think it is OK to refer to the benchmarks in panama-vector (IIRC we may have done that before for other PRs to jdk).
> 
> Oh, please let's have them in mainline. It's a PITA for reviewers not to be able to see them. What's the downside?

In the past we have been reluctant to add the generated benchmarks ([here](https://github.com/openjdk/panama-vector/tree/vectorIntrinsics/test/micro/org/openjdk/bench/jdk/incubator/vector/operation)) due to the quantity of them (the scripts would also need to be included) and whether this was the right approach.

We did this before the HotSpot IR test framework was added. I think a better approach would be to use the IR test framework to assert that the right C2 nodes have been produced (recently there has been some enhancements on templating which might help). That does not obviate the need for performance tests, but it does reduce the need for them especially when testing individual operations.

-------------

PR Comment: https://git.openjdk.org/panama-vector/pull/225#issuecomment-2596228147


More information about the panama-dev mailing list