On Tue, 27 Feb 2024 11:20:11 GMT, Raffaello Giulietti <rgiulie...@openjdk.org> wrote:
> There are no libraries that have worse errors than OpenLibm, so I'm wondering > what these values are good for? When I was working on updating the worst-case tests for Math, I would check the input values in Math.foo() and StrictMath.foo() and if they differed pick the smaller one as the reference value. The those inputs that differ, the input is a fingerprint marker for FDLIBM and I added cases to the corresponding StrictMath test. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/15879#discussion_r1504998720
