8089 was Re: Hyperion Passport, Apricot, Convergent Technologies workSlate, - test-drb@ccmp.vtda.org

31 Dec 1999

Quite true, Hans.  The Intel effort to push the industry ahead fell with
this chipset fell flat because software vendors were not willing to
re-architect their existing constructs in order to capitalize on the
advantages a hierarchical structure for the I/O subsystem would provide.
Software simply never evolved to the point at which it fully exploited the
hardware.  The degeneration of the initial architecture into what it now is
rather than the more highly abstracted (and more highly evolved) construct
that was presented by the hardware was caused by software vendors' inability
to organize themselves around a unified construct.

Dick

-----Original Message-----
From: Hans Franke &lt;Hans.Franke(a)mch20.sbs.de&gt;
To: Discussion re-collecting of classic computers
&lt;classiccmp(a)u.washington.edu&gt;
Date: Friday, December 31, 1999 7:48 AM
Subject: Re: 8089 was Re: Hyperion Passport, Apricot, Convergent
Technologies workSlate,

...
   >a strange
machine -- one of the few MS-DOS boxes to use an 8089 I/O
 >coprocessor (which I why I got interested in it -- nice chip). 
>   Hi Tony,  I have an 8089 in a CPM machine but I haven't been able to
find
...
   out much about
that chip. What can you tell me about it? 
The 8089 is the supposed I/O Processor for 8086/88 systems.
Like the 8087 it can be used in an 8 or 16 bit bus system.

Basicly Intel did go for a structure like a /370 alike mainframe.
A CPU (8086 or 8088), an IOC (8089) and a math extension to the
CPU (FPU, 8087). Due the nature of the 86 Bus the 8087 did become
a bit more independant than a /370 math extension. Basicly the
8086 is designed as ordinary CPU, while the 8089 is optimized for
I/O - you may call it a super DMA chip, but thats like calling
a versitale VW Bus a shoping cart. From a system softwares view
point you may assign the low level I/O drivers to be run on the
8089, while the 8086 executes the high level functions. For a
a Disk drive this may give you an SCSI like interface between
these components - the 8086 supports a control block with (logical)
drive ID, and block number, while the 8089 translates this to
controller address, drive number and head/track/sector number
to programm the FD chips and then initiate the DMA transfer.

For a serial line, this may include low level block drivers
for packing/unpacking and CRC and block repeat to handle the
complete transmission of a given data (chunk).

Especialy in a multi tasking environment this gives an enormus
boost in available processing power. Not to mention the simplified
OS design. As a standard IOC, the 89 also overcomes the driver
problems that you get if every I/O has his own, different
'intelligent' controller.

Well, I guess there may have been some applications, but I doubt
if this has ever used the full potential of the 86/89 combination.
The design did realy borrow a lot of good mainframe ideas. Just to
early ?

Gruss
H.

--
Der Kopf ist auch nur ein Auswuchs wie der kleine Zeh.
H.Achternbusch 