The Philco 212, Philco 2000, Philco S2000-TRANSAC... and the RCA 601
John Savard
amazingly quickly.
I had also noted that while on Al Kossow's site, there was a
description of the GE-645 computer, the 20-bit GE-225 wasn't there.
But, thanks to Ed Thelen's site, I found that there is a site reviving
the Dartmouth Time Sharing System, and it had GE-235 manuals (and
GE-265 manuals) there.
I vaguely remember someone requesting information about a mysterious
computer front panel, and being told it was a Sigma. I saw a picture
in an old book on computers I had, and realized that this was the
computer!
But after doing some searching, I found that the Computer Museum
already knew the computer was a Philco 212.
And it had a 24-bit word size. For various sentimental reasons (I
never actually worked on an old SDS 930 that, during my years as a
grad student, was being occasionally used for running dusty deck
jobs... nor had I any advance knowledge of its being scrapped those
long decades ago, but I presume, and hope, some prof kept the front
panel at least) and reasons connected with my web page
http://home.ecn.ab.ca/~jsavard/comp/cpint.htm
I had been looking through the web for information about the
architectures of 24-bit computers.
During my web searches, though, I found that 212 was really the model
number of the central processor. The computer was officially a Philco
2000, or even a Philco 2000 Model 212.
Then I found that the Philco S2000 TRANSAC, which looked different
from a 212, having a horizontally-extended front panel (was this the
model 200 central processor? Or was the model 2000 completely
different, or a later iteration of the same architecture?) was the
*first* commercial transistorized stored-program computer.
Shipped to customers in either 1957 or 1958, depending on which web
site you believe.
Oh, yes. That qualification "stored-program" computer is really
necessary. IBM, to get its name in the history books, and lower the
electrical bills of small businesses everywhere, beat Philco to the
punch with a transistorized version of its card-programmed calculator,
the IBM 608.
The RCA 601 is another 48-bit machine that doesn't seem to be
described on the instruction set level anywhere I could find.
And I remember seeing the manual for a machine whose name I can't even
remember that was user microprogrammable, with microprograms going
into a special fast core memory with dual-hole rectangular cores, and
which had a 24-bit word length. I remember that when the computing
department library was starting up a collection of old manuals, I
helped arrange for the area where I worked to contribute that manual
to it.
I suspect, because I think I bumped into a mention of it on the web,
that someone will probably remember which machine this was; it long
predated Microdata and Nanodata.
John Savard
http://home.ecn.ab.ca/~jsavard/index.html
Charles Richmond
>
> [snip...] [snip...] [snip...]
>
> I had been looking through the web for information about the
> architectures of 24-bit computers.
>
Harris 800 and Harris 1200 as late as the mid-1980's at least.
Their FORTRAN77 compiler did an *excellent* job of optimizing,
and of course there was a cache to speed things along. The
main memory of the Harris 800 was made up of 100 nanosecond
64k RAM chips, and used Hamming code to recover from single
bit errors. Each 24 bit word had one bit in each of 24 RAM
chips, plus five more RAM chips for the Hamming code.
The Harris 1200 used ECL and ran very hot. I am *not* sure how
the memory on that one was set up...
--
+----------------------------------------------------------------+
| Charles and Francis Richmond richmond at plano dot net |
+----------------------------------------------------------------+
John Savard
> Harris took over the Data Craft 24 bit designs, and sold the
> Harris 800 and Harris 1200 as late as the mid-1980's at least.
> Their FORTRAN77 compiler did an *excellent* job of optimizing,
> and of course there was a cache to speed things along.
Thank you for your response to my post; I've been unable to see new
posts, or reply, due to newsserver problems lately.
John Savard
John Savard
(John Savard) wrote, in part:
>And I remember seeing the manual for a machine whose name I can't even
>remember that was user microprogrammable, with microprograms going
>into a special fast core memory with dual-hole rectangular cores, and
>which had a 24-bit word length. I remember that when the computing
>department library was starting up a collection of old manuals, I
>helped arrange for the area where I worked to contribute that manual
>to it.
It's mentioned in the BRL report: the machine I was thinking of was
not the (also microprogrammable and 24-bit!) Burrougs B1700, but the
Packard Bell PB440.
John Savard
http://home.ecn.ab.ca/~jsavard/index.html
Sarr J. Blumson
John Savard <jsa...@excxn.aNOSPAMb.cdn.invalid> wrote:
>
>But after doing some searching, I found that the Computer Museum
>already knew the computer was a Philco 212.
>
>And it had a 24-bit word size.
Sort of. It accessed memory in 48 bit words, but stored two instructions
per word.
The 212 was a very potent machine; they ran the AEC (now Department of
Energy) atomic power labs (Los Alamos, Livermore, etc) before the CDC
6600, and ran the North American Air Defense system at least to the late
60s. A group a Ford built a time sharing OS for it in 68 and 69 and
designed a follow on, but Ford pulled Philco out of the computer business.
--
--------
Sarr Blumson sarr.b...@alum.dartmouth.org
http://www-personal.umich.edu/~sarr/