Date: Sat, 21 Feb 2026 15:57:34 +0100 From: Daniel Engberg <diizzy@FreeBSD.org> To: Robert Clausecker <fuz@fuz.su>, =?UTF-8?B?w4RsdmVu?= <alven@freebsd.org> Cc: ports-committers@freebsd.org, dev-commits-ports-all@freebsd.org, dev-commits-ports-main@freebsd.org Subject: Re: git: ddd386c470a1 - main - textproc/krep: [NEW PORT] High-performance string search utility Message-ID: <f6db520b-3c67-481f-a10f-b7514f55f5be@FreeBSD.org> In-Reply-To: <aZm6hSgYs52FZAEk@fuz.su> References: <69961a51.197dd.744ceecd@gitrepo.freebsd.org> <39e6740c-ab82-42db-a68d-10739f84943e@FreeBSD.org> <7z7bs6ei1b.fsf@nacka.vinterdalen.se> <aZm6hSgYs52FZAEk@fuz.su>
index | next in thread | previous in thread | raw e-mail
There's no need for this and adds unnecessary special cases, all upstream do is adding flags which CPUTYPE already handles (indirectly as they're inherited by whatever -march is set to) and establish. Best regards, Daniel On 2026-02-21 15:00, Robert Clausecker wrote: > > Am Sat, Feb 21, 2026 at 10:18:24AM +0400 schrieb Älven: >> Daniel Engberg <diizzy@FreeBSD.org> writes: >>> >>> All SIMD "hacks/detection" should be disabled as we have CPUTYPE and >>> the current approach is also broken as host might not share the same >>> CPU features as target. > > The feature detection is gated behind a non-default option, which is fine. > The usual name for this option is NATIVE though. I don't see why we should > not offer this, it just shouldn't be the default. > > Long term, upstream should perhaps be motivated to do runtime dispatch > for different levels of SIMD. > >>> Best regard, >>> Daniel > > Yours, > Robert Clausecker > >> Thank you! Agree. Committed [0]. >> >> [0] https://cgit.freebsd.org/ports/commit/?id=17c13889a5137e45011a083759a1879fec109613 >> >> -- >> Best regards, >> Älven > > >home | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?f6db520b-3c67-481f-a10f-b7514f55f5be>