On Tue, 25 Aug 2026 09:26:09 GMT, Andrew Dinn <[email protected]> wrote:

>> Thank you @merykitty!  After updating the code with this solution, the 
>> intrinsics performance was inline (no pun intended) with the other 
>> intrinsics performance for multiplication and squaring.  Done!
>
> @smemery Could you also limit the test to run only on x86 and aarch64. We 
> don't really want to run it on arches that don't have an intrinsic as it 
> results will be bogus.

I'm reluctant in restricting this benchmark to just x86 and aarch64 given that 
the other half of the benchmarks, `IntegerPolynomialP256`, are not dependent 
upon intrinsics and could prove to be useful if we decide to optimize the Java 
implementation, like we have for other curves, e.g., Curve25519, and need a way 
to measure performance gains from these optimizations.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/32047#discussion_r3858656944

Reply via email to