Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!usc!wuarchive!udel!udccvax1!meese
From: meese@vax1.acs.udel.EDU (Tim Meese)
Newsgroups: comp.sys.mac.hardware
Subject: Re: IO mgmt on hard disks: why not?
Message-ID: <5848@udccvax1.acs.udel.EDU>
Date: 12 Mar 90 06:00:57 GMT
References: <7974@sdcsvax.UCSD.Edu>
Reply-To: meese@vax1.acs.udel.EDU (Tim Meese)
Distribution: na
Organization: University of Delaware
Lines: 27

In article <7974@sdcsvax.UCSD.Edu> joe@amos.ucsd.edu.UUCP (Joe) writes:
>Why aren't there any dedicated-scsi chip/cards for the mac
>which work? The idea is simple: you have a processor with
>a large cache; in theory, writes to disk are handed-off
>instantly from the cpu to the coprocessor. If the cache is
>large enough, many types of reads could be nearly instantaneous
>as well.
>
>Why not?
>
>Joe.

Apple will announce yet another high-priced Mac (er, workstation?) on Mar 19
called the IIfx (40 MHz 68030) which will include among other things, such as
a Fast Memory Controller that will allow burst mode reads and writes to RAM,
a I/O chip called an IOP.  The IOP will have a 6502 microprocessor and a
32KB buffer to relieve the '030 of I/O to the floppy disks and ADB, as well
as LocalTalk network I/O.  The report I read didn't mention anything about 
intercepting SCSI I/O, but I would imagine that might be the next data
stream Apples engineers may work on.

Tim

-- 
Tim Meese                            \\\\   Internet: meese@vax1.acs.udel.edu
Electrical Engineering Department    ////       UUCP: ...!udel!udccvax1!meese
University of Delaware               \\\\      Voice: (302) 475-8756