[Simh] VAX Tape Emulation?

Clem Cole clemc at ccc.com
Thu Jan 25 20:47:33 EST 2018


On Thu, Jan 25, 2018 at 7:46 PM, Johnny Billquist <bqt at softjar.se> wrote:

>
> Domain OS using 1056 bytes? On what systems? I used a lot of DN3000, 4000
> and 5000 systems, and they used bog standard disk drives with 512 byte
> sectors. (Interesting systems in some ways, but their windowing system was
> pretty horrible.)


​All of the early ones -- until the DN10000 I'm pretty sure.   They used a
1024 basic block, but Mike Spourer had used the Xerox Alto trick of putting
some of the FS metadata into the controller microcode. That's what the
extra 32 bytes were.   Masscomp and Apollo used the same SMS 'OMPTI 7000'
controller for the ST506/ESDI drives, and you'll see it in the
SMS marketing literature as it says something like "Supports 512/1024/1056
block sizes".  I have controller doc somewhere in my basement - thinking
about it, SMS might have even supported blocks as large as 4K - but I
definitely remember them putting the 1056 stuff in for Apollo.   Remember
we all knew each other, the firms were one town apart and couple of people
had somebody working at each.

Anyway, with the DN10000, Apollo started to use the Xylogics 4xx
controllers and Xylogics either wanted too much money for custom microcode
or the Apollo guys decided custom microcode was too expensive.   In our
Stellar days, we used to kid Mike about the 1056 bytes choice. BTW, Mike
might have done the same thing at Prime, I never looked at/knew much the
low level I/O of those systems. [@Stellar we used 4096 as the basic size
but that was a multiple of 512 so it was never an issue].

I've lost track of Mike, but Scott Baden (who is now a Prof at UCSD) was a
CPU guy on Prime 750 and I'm still pretty close too, and he might know.
I'll ask him if he remembers next time I see him.

Clem
ᐧ
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.trailing-edge.com/pipermail/simh/attachments/20180125/813afb1f/attachment.html>


More information about the Simh mailing list