From owner-freebsd-current Fri Apr 5 09:07:23 1996 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id JAA25988 for current-outgoing; Fri, 5 Apr 1996 09:07:23 -0800 (PST) Received: from chrome.jdl.com (chrome.onramp.net [199.1.166.202]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id JAA25973 for ; Fri, 5 Apr 1996 09:07:20 -0800 (PST) Received: from localhost (localhost [127.0.0.1]) by chrome.jdl.com (8.6.12/8.6.12) with SMTP id LAA14190; Fri, 5 Apr 1996 11:06:35 -0600 Message-Id: <199604051706.LAA14190@chrome.jdl.com> X-Authentication-Warning: chrome.jdl.com: Host localhost didn't use HELO protocol To: davidg@Root.COM cc: asami@cs.berkeley.edu (Satoshi Asami), current@FreeBSD.org, nisha@cs.berkeley.edu, tege@matematik.su.se, hasty@rah.star-gate.com Subject: Re: fast memory copy for large data sizes In-reply-to: Your message of "Fri, 05 Apr 1996 02:21:48 PST." <199604051021.CAA00222@Root.COM> Clarity-Index: null Threat-Level: none Software-Engineering-Dead-Seriousness: There's no excuse for unreadable code. Net-thought: If you meet the Buddha on the net, put him in your Kill file. Compiler-Motto: Wintermute is dead. Long live Wintermute. Date: Fri, 05 Apr 1996 11:06:31 -0600 From: Jon Loeliger Sender: owner-current@FreeBSD.org X-Loop: FreeBSD.org Precedence: bulk So, like David Greenman was saying to me just the other day: > >Here are the kind of numbers we are seeing, and hope you will see, if > >you run the program attached at the end of this mail: > > > > 90MHz Pentium (silvia), SiS chipset, 256KB cache: > > > > size libc ours > > 32 15.258789 MB/s 6.103516 MB/s > > 64 20.345052 MB/s 15.258789 MB/s > > 128 17.438616 MB/s 15.258789 MB/s > > This would be a big lose in the kernel since just about all bcopy's fall > into this range _except_ disk I/O block copies. I know this can be done bette >r > using other techniques (non-FP, see hackers mail from about 3 months ago). Don't know how much it would cost (in performance), but would it make sense to have a simple size-based cutoff test for the two different algorithms? jdl