[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ih] NCP and TCP implementations

On 10 Mar 2020 at 13:23, Steve Crocker via Internet-hi wrote:

> The TAC was an extension of the IMP.  The original IMP was built on
> the
> Honeywell 516 (and later 316) platform, which was a 16 bit twos
> complement
> computer.  I assume Hinden's reference to 15-bit arithmetic reflected
> the
> fact that the arithmetic was signed.

I honestly cannot remember what the TAC was!!   Was that the TIP?   Regardless, 
yes, the x16s had 16-bit signed arithmetic with 10 bit addressing 9 bits of page 
address, 1 bit of "this page" or the 0 page, 16Kwords of memory. 

Things got more complicated with the 316 -- it supported 32K words.  What we 
did for the TIP [and maybe the TAC, whatever that was] was to keep the IMP 
*unchanged* in the bottom 16K, and then in the upper 16K we wrote a 
self-contained "host". There was some [small!] hack to fake interrupts and 
input/output to this host but to the IMP it thought it was just another NCP 
connected host.  It'd set up a host output buffer and instead of doing a hardware 
"send" it'd pass control to the upper 16K.  Similarly [at least for the TIP], when it 
got something in from a terminal it'd copy it into a host-input buffer and then 
issue an "interrupt" down to the IMP.   Worked quite well.

            Bernie Cosell
       bernie at fantasyfarm.com
-- Too many people; too few sheep --