On Fri, 17 Oct 2025 22:47:21 GMT, Sergey Kuksenko <[email protected]> wrote:
>> Add StrictMath microbenchmarks to cover FDLIBM algorithms
>>
>> The methods that require fdlibm semantics are sin, cos, tan, asin, acos,
>> atan, exp, log, log10, cbrt, atan2, pow, sinh, cosh, tanh, hypot, expm1, and
>> log1p.
>
> Sergey Kuksenko has updated the pull request incrementally with one
> additional commit since the last revision:
>
> Update StrictMathExtraBench.java
>
> Fix copyright format
test/micro/org/openjdk/bench/java/lang/StrictMathExtraBench.java line 51:
> 49: this.min = min;
> 50: this.max = max;
> 51: this.signed = signed;
This field is never read.
I guess you want to change the `setup()` method to use this field rather than
rely on `nextBoolean()`?
test/micro/org/openjdk/bench/java/lang/StrictMathExtraBench.java line 626:
> 624: public static class RangeHalfLn2To56Ln2 extends RangeState {
> 625: public RangeHalfLn2To56Ln2() {
> 626: super(0.34657359, 38.8162421);
Please consider declaring "magic" values related to ln(2) as `private static
final double`s
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/27881#discussion_r2445241030
PR Review Comment: https://git.openjdk.org/jdk/pull/27881#discussion_r2445271711
