Thank you, again. Sorry for not looking at your Godbolt link last night. My mistake in my own testing with GCC was not specifying -march. Also, I see in the psABI that __m512 has size and alignment 64. So, all is well, and I hope I did not disturb the _BitInt conversation too badly. -----Original message----- From: Thiago Macieira <thiago_at_[hidden]> Sent: Saturday, September 6 2025, 8:19 am To: std-proposals_at_[hidden] <std-proposals_at_[hidden]>; Paul Caprioli <paul_at_[hidden]> Subject: Re: [std-proposals] D3666R0 Bit-precise integers On Friday, 5 September 2025 21:03:51 Pacific Daylight Time Paul Caprioli wrote: > But given the compiler-dependent alignment differences, it seems one cannot > portably use this type in a structure because different compilers on the > same platform will pad differently. I don't want to do this anyway, but > where is it written I can't? There aren't alignment differences: they all have the same alignment. Look at the Godbolt link I posted. > If I understand correctly, these types are not part of the platform ABI. > That's the answer, right? They are. See the x86-64 psABI https://gitlab.com/x86-psABIs/x86-64-ABI, figure 3.1. > So, what's the plan for _BitInt? See the next page (18) of the psABI above. -- Thiago Macieira - thiago (AT) macieira.info - thiago (AT) kde.org Principal Engineer - Intel Platform & System Engineering