From owner-freebsd-multimedia Tue Mar 14 4:34: 6 2000 Delivered-To: freebsd-multimedia@freebsd.org Received: from dorifer.heim3.tu-clausthal.de (dorifer.heim3.tu-clausthal.de [139.174.243.252]) by hub.freebsd.org (Postfix) with ESMTP id E195D37B748 for ; Tue, 14 Mar 2000 04:34:02 -0800 (PST) (envelope-from olli@dorifer.heim3.tu-clausthal.de) Received: (from olli@localhost) by dorifer.heim3.tu-clausthal.de (8.9.3/8.9.3) id NAA70280; Tue, 14 Mar 2000 13:33:47 +0100 (CET) (envelope-from olli) Date: Tue, 14 Mar 2000 13:33:47 +0100 (CET) Message-Id: <200003141233.NAA70280@dorifer.heim3.tu-clausthal.de> From: Oliver Fromme To: freebsd-multimedia@FreeBSD.ORG, vallo@matti.ee Reply-To: freebsd-multimedia@FreeBSD.ORG Subject: Re: gogo + 3dnow X-Newsgroups: list.freebsd-multimedia In-Reply-To: <8ala1u$1cmj$1@atlantis.rz.tu-clausthal.de> User-Agent: tin/1.4.1-19991201 ("Polish") (UNIX) (FreeBSD/3.4-19991219-STABLE (i386)) MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit Sender: owner-freebsd-multimedia@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org (Posted & mailed according to Reply-To) Vallo Kallaste wrote in list.freebsd-multimedia: > As I understand the PIII SSE instructions have no use at the moment. > What are the reasons FreeBSD doesn't support SSE? Is it hard work to do > or mostly missing developer time? What do you mean by saying "FreeBSD doesn't support" it? FreeBSD doesn't "support" 3DNow either, but that doesn't prevent anyone from writing programs that run under FreeBSD and use 3DNow instructions. In fact, I already did that. If you mean that FreeBSD's compiler (gcc) should automatically create SSE (or MMX, or 3DNow) instructions from standard code, well, that's a different issue. And it's very non-trivial to _automatically_ identify code sections that could be improved using SSE (or ...) instructions. In particular, you cannot use FPU instructions at the same time, and switching between those instruction sets involves a certain delay penalty. Therefore I think it is _much_ more useful to improve critical code sections _manually_ by replacing them with SSE (...) code if appropriate. This can result in _dramatic_ performance improvements (+100% and more). This is not just theory -- I experimented a lot with 3DNow in the past months. BTW: I also have some patches to nasm to support the new instructions (enhanced K7 3DNow / SSE) and submitted them to the nasm author in Autumn last year. He replied that he already had patches himself and would release them along with a new version. I haven't heard anything since then, but I haven't actively looked for it. Regards Oliver -- Oliver Fromme, Leibnizstr. 18/61, 38678 Clausthal, Germany (Info: finger userinfo:olli@dorifer.heim3.tu-clausthal.de) "In jedem Stück Kohle wartet ein Diamant auf seine Geburt" (Terry Pratchett) To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-multimedia" in the body of the message