Date: Thu, 25 Oct 2001 23:46:00 +0100 (BST) From: Doug Rabson <dfr@nlsystems.com> To: Luigi Rizzo <rizzo@aciri.org> Cc: John Baldwin <jhb@FreeBSD.org>, Jonathan Lemon <jlemon@FreeBSD.org>, <cvs-all@FreeBSD.org>, <cvs-committers@FreeBSD.org> Subject: Re: cvs commit: src/sys/dev/fxp if_fxp.c Message-ID: <20011025234439.M549-100000@salmon.nlsystems.com> In-Reply-To: <20011025151935.G57392@iguana.aciri.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 25 Oct 2001, Luigi Rizzo wrote: > > > does it mean that integer multiply in the kernel is bad on the ia64 :) > > > > Not really - its reasonably quick. They just didn't bother to add any > > integer multipliers when there were two perfectly good floating point > > units going idle. Moving integers in and out of float registers is quick > > and easy so it actually makes sense. Divides are done with floating point > > too. > > the point for not having FP in the kernel is not speed, but rather > the fact that you do not have to save those registers in a context > switch. This could just be an artifact of ancient times where FPU's > were separate units... > > So, i guess the FP registers are saved in the IA64 context switch routines ? The kernel only uses a cut-down set of fp registers - ten registers are saved/restored on exceptions and four on context switches. -- Doug Rabson Mail: dfr@nlsystems.com Phone: +44 20 8348 6160 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20011025234439.M549-100000>