From owner-freebsd-hackers Mon Mar 2 20:17:12 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id UAA26874 for freebsd-hackers-outgoing; Mon, 2 Mar 1998 20:17:12 -0800 (PST) (envelope-from owner-freebsd-hackers@FreeBSD.ORG) Received: from dingo.cdrom.com (dingo.cdrom.com [204.216.28.145]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id UAA26865 for ; Mon, 2 Mar 1998 20:17:07 -0800 (PST) (envelope-from mike@dingo.cdrom.com) Received: from dingo.cdrom.com (localhost [127.0.0.1]) by dingo.cdrom.com (8.8.8/8.8.5) with ESMTP id UAA14993; Mon, 2 Mar 1998 20:14:53 -0800 (PST) Message-Id: <199803030414.UAA14993@dingo.cdrom.com> X-Mailer: exmh version 2.0zeta 7/24/97 To: Jay Nelson cc: hackers@FreeBSD.ORG Subject: Re: Recommendation needed for real time monitor In-reply-to: Your message of "Mon, 02 Mar 1998 21:48:26 CST." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 02 Mar 1998 20:14:52 -0800 From: Mike Smith Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG > Don't know if this is the right list for this, so point me in the right > direction it I'm wrong. > > Given the available hardware supported by FreeBSD, what are your > hardware recommendations for a machine capable of handling 31 > simultaneous serial connections at 19200? This is a real-time machine > monitoring situation, not dial up users. Data is bursty and > non-compressed. I need to insure worst case with no buffer overruns. We need some more information to give you a really good answer; - maximum burst size - net data rate - peak data rate over a small period (eg. minimum/maximum interburst period). If the bursts are small (< 16 bytes) and the interburst period moderately long (so the net data rate is low), then multiport cards based on 16x50 UARTs will do the job. You might want to look at 16650's instead to give you more headroom. For larger bursts, you will want to look at "smart" cards with lots of buffering, eg. some of the supported Digiboard cards have 512 bytes or more per port. Also some of the Stallion and Specialix cards are worth checking out. 21 ports at 19200 is only 60k/sec, or ~120k port accesses/second for 16x50 UARTs (hit me if I screw up here Bruce 8). If you're streaming you get one interrupt every 14 bytes (approx), or a maximum of 4251 interrupts per second. Depending on the line discipline involved, most of your overhead is likely to be in the tty subsystem. More CPU will help there, but without actually measuring it I can't give you really useful numbers. (I would suggest starting with something like a P166 and experimenting.) It also depends on what else the machine will be doing. Busmastering peripherals will help (disk, network adapter, etc.) -- \\ Sometimes you're ahead, \\ Mike Smith \\ sometimes you're behind. \\ mike@smith.net.au \\ The race is long, and in the \\ msmith@freebsd.org \\ end it's only with yourself. \\ msmith@cdrom.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message