Anyone playing with the 8x300

[Dwight K. Elvey]

>From: "Allison" <ajp166 at bellatlantic.net>
>
>>
>>Subject: Re: Anyone playing with the 8x300
>>   From: "Dwight K. Elvey" <dwight.elvey at amd.com>
>>   Date: Fri, 13 May 2005 14:57:09 -0700 (PDT)
>>     To: cctalk at classiccmp.org
>>
>>>
>>>Ah the classic first of the fast microcontrollers.
>>>I'd have to dig but I vaguely remember the 8x300
>>>as a disk controller apnote.  Nasty beast to program.
>>>
>>>Allison
>>>
>>
>>Hi
>> Don't know why you'd say this, it only has 8 instructions!
>>I've got the spec posted to Al's site.
>> This controller application is a little interesting in that 
>>who ever designed this board, also must have done a bitslice
>>designs at one time or another. To save a machine cycle, all
>>I/O addresses are selected by a ROM tied to the instruction
>>addressing. Normally it would take two cycles, one to write
>>the I/O address and one to transfer the data. With the
>>ROM, the address is understood by the program's execution
>>address location.
>>Dwight
>
>It's more of a sequencer or state machine with a crude ALU.
>As to those 8 instructions, looks at what the fields are 
>for each one. I've done horizontal microcode and that is 
>similar. Due to it's very harvard design it's not one you 
>will do constans in rom much.  Also the IO devices are 
>really IO specific as in the address of each is coded 
>into the part. 

Hi Allison
 It is interesting that this application didn't use
that method of I/O addressing. All I/O devices are typical
bus type devices. It has a ROM connected to the instruction
addresses that selects the I/O based on the address being
executed. They don't seem to be taking advantage of the
read modify write I/O, either. The ports are all wired as
unidirectional ports.
Dwight

>
>One use for it was an 8bit wide DSP.  I have a real one here
>of the later 8x305 I2L that was a bit faster.
>
>Allison
>
>