https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96958
Jonathan Wakely <redi at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Target Milestone|--- |11.0 --- Comment #6 from Jonathan Wakely <redi at gcc dot gnu.org> --- I've just realised there's a second place I need to replace long double ... which explains why the commit above didn't fix the test failures I was seeing on POWER9 when mixing ibm128 and ieee128 long double formats. I probably should have looked at this commit sooner: (In reply to James Greenhalgh from comment #0) > It was pointed out that some forks of GCC ( > https://github.com/FEX-Emu/gcc/commit/ > 8a2b7389f50a50a4e26ec98101d47fb1fc1c1bcd ) reduce the hashtable policy > implementation from a long double to a double. Doing this reduces it from a > soft-float calculation to hardware floating-point. N.B. the commit msg in that fork talks about Oracle making this change. I don't think it has anything to do with Oracle at all, except that one of the main libstdc++ contributors happens to work for them. But he didn't change it because of anything Oracle was doing or wanted, it was just normal libstdc++ maintenance.