On Tue, 6 Sep 2016, Richard Biener wrote: > > On 32-bit x86, max_align_t is currently 8-byte aligned, but > > _Decimal128 and _Float128 are 16-byte aligned, so the alignment of > > max_align_t needs to increase to meet the standard requirements for > > these types. > > As doubles on 32bit x86 are 4 byte aligned I suppose the choice could have > been to make those types 8 byte aligned to avoid changes of max_align_t? > (with the obvious eventual performance penalty of needing to do unaligned > loads/stores to xmm registers).
However, the choice made some years ago (and documented in HJ's ABI document) was 16-byte alignment. -- Joseph S. Myers jos...@codesourcery.com
