On Thu, 23 Feb 2023 23:28:11 GMT, Joe Darcy <da...@openjdk.org> wrote:
> The wheel of FDLIBM porting turns a notch and sqrt comes into play. > > While the sqrt operation usually has a hardware implementation that is > intrinsified, for completeness a software implementation should be available > as well. Otherwise LGTM src/java.base/share/classes/java/lang/FdLibm.java line 659: > 657: q1 += r; > 658: } > 659: ix0 += ix0 + ((ix1 & sign) >>> 31); You may want to add the traditional `// unsigned shift` comment even here, just to be consistent. test/jdk/java/lang/StrictMath/ExhaustingTests.java line 71: > 69: // Since sqrt is correctly rounded, could additionaly the > 70: // transliteration or StrictMath sqrt methods to against > 71: // Math::sqrt. I guess this comment has been messed up by some copy&paste activities? Or is it a ChatGPT output? test/jdk/java/lang/StrictMath/FdlibmTranslit.java line 653: > 651: q1 += r; > 652: } > 653: ix0 += ix0 + ((ix1&sign)>>>31); You may want to add the traditional `// unsigned shift` comment even here, just to be consistent. ------------- PR: https://git.openjdk.org/jdk/pull/12736
