From owner-freebsd-hackers Wed Nov 29 04:00:47 1995 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.6.12/8.6.6) id EAA06912 for hackers-outgoing; Wed, 29 Nov 1995 04:00:47 -0800 Received: from Sysiphos (Sysiphos.MI.Uni-Koeln.DE [134.95.212.10]) by freefall.freebsd.org (8.6.12/8.6.6) with SMTP id EAA06893 for ; Wed, 29 Nov 1995 04:00:30 -0800 Received: by Sysiphos id AA06154 (5.67b/IDA-1.5 for hackers@freebsd.org); Wed, 29 Nov 1995 12:59:55 +0100 Message-Id: <199511291159.AA06154@Sysiphos> From: se@zpr.uni-koeln.de (Stefan Esser) Date: Wed, 29 Nov 1995 12:59:54 +0100 In-Reply-To: J Wunsch "Re: Anyone keen to help me get a Phillips CDD 521 Recorder working?" (Nov 29, 8:55) X-Mailer: Mail User's Shell (7.2.6 alpha(2) 7/9/95) To: joerg_wunsch@uriah.heep.sax.de (Joerg Wunsch) Subject: Re: Anyone keen to help me get a Phillips CDD 521 Recorder working? Cc: hackers@freebsd.org Sender: owner-hackers@freebsd.org Precedence: bulk On Nov 29, 8:55, J Wunsch wrote: } Subject: Re: Anyone keen to help me get a Phillips CDD 521 Recorder workin } As Douglas Ambrisko wrote: } > } > So the data has to come from a hard drive } > attached to the SCSI bus. A lot of burners use a separate SCSI bus for the } > CD-R so it's doesn't have to worry about any other device holding the } > bus. Also people use AV disks to eliminate the problem of thermal } > recalibration. } } I think the problem with thermal recalibration is the biggest one. Not a problem anymore with embedded servo drives (e.g. a Quantum Fireball or any other of the current low cost SCSI drives ...) } On the other hand, with something like `team' doing the data } shuffling, you could use the RAM as an additional cache. One process } in constantly reading the input file, and this one will choke when the } disk performs a thermal recalib. Since the output process reads } buffered data, it will not starve immediately. The only requirement } is that the average CD burn rate is much slower than the average disk } read rate, but with 600 KB/s for the burner, this doesn't seem to be a } problem. } } Nevertheless, the driving process should run at rtprio, so its getting } the highest probability of grabbing the CPU whenever it needs it } (which is seldom anyway, since both of its sub-processes are i/o- } bound). Well, how about a configuration with a dedicated 1GB SCSI drive ($250) connected to a SCSI host adapter ($40) that it shares only with the CDROM writer, and using a SCSI Copy command for the actual transfer ... Seems like a low cost solution, that only requires a command similar to scsi_format to initiate the copy ... Regards, STefan -- Stefan Esser, Zentrum fuer Paralleles Rechnen Tel: +49 221 4706021 Universitaet zu Koeln, Weyertal 80, 50931 Koeln FAX: +49 221 4705160 ============================================================================== http://www.zpr.uni-koeln.de/~se