From owner-freebsd-hackers Wed May 17 6:14: 8 2000 Delivered-To: freebsd-hackers@freebsd.org Received: from duke.cs.duke.edu (duke.cs.duke.edu [152.3.140.1]) by hub.freebsd.org (Postfix) with ESMTP id BB34F37BC2A for ; Wed, 17 May 2000 06:14:04 -0700 (PDT) (envelope-from gallatin@cs.duke.edu) Received: from grasshopper.cs.duke.edu (grasshopper.cs.duke.edu [152.3.145.30]) by duke.cs.duke.edu (8.9.3/8.9.3) with ESMTP id JAA24962; Wed, 17 May 2000 09:14:03 -0400 (EDT) Received: (from gallatin@localhost) by grasshopper.cs.duke.edu (8.9.3/8.9.1) id JAA06311; Wed, 17 May 2000 09:14:03 -0400 (EDT) (envelope-from gallatin@cs.duke.edu) From: Andrew Gallatin MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Date: Wed, 17 May 2000 09:14:03 -0400 (EDT) To: Issei Suzuki Cc: hackers@freebsd.org Subject: Re: SSE support for FreeBSD In-Reply-To: <20000516.234638.15656027.issei@issei.org> References: <20000516.234638.15656027.issei@issei.org> X-Mailer: VM 6.43 under 20.4 "Emerald" XEmacs Lucid Message-ID: <14626.39292.549162.216866@grasshopper.cs.duke.edu> Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Issei Suzuki writes: > I add SSE support for FreeBSD 4.0-RELEASE and 5.0-CURRENT. Some more > work is needed everything to go well, but it works functionally now. > > The patches and documents are avaiable at: > > http://home.jp.freebsd.org/~issei/sse/sse.html > > Limitations: > - Multiprocessor system is not supported. > - There is no exception handler to deal with SSE Exceptions. > - No spaces are available for SSE registers in struct fpregs. So, we > cannot record SSE registers when dumping core, nor access to SSE > registers via procfs. > > Comments, bug reports etc. are welcome. Cool! Have you (or anybody else) considered using the PIII SIMD functionality for optimizing kernel bcopy/copyin/copyout and checksums? There are linux kernel patches for this at http://www.redhat.com/~mingo/mmx-patches I have a trivial user-level app which shows a bcopy speedup of between 40% (PIII Xeon, 733MHz) and 50% (PIII, 450MHz) when using the kni memcpy stolen from those patches rather than the bcopy from libc, so I think it could be worth looking into. Cheers, Drew ------------------------------------------------------------------------------ Andrew Gallatin, Sr Systems Programmer http://www.cs.duke.edu/~gallatin Duke University Email: gallatin@cs.duke.edu Department of Computer Science Phone: (919) 660-6590 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message