Computer Technology in USSR - circa 1955 to 1970
John Passafiume
term working in research labs in the USSR on airborne computer projects.
In our discussions he mentioned some technogies and uses of technologies
that I was unaware of. For example, he discussed the use of core memories
as ROM where "little old ladies" wired the application program bits into
the ROMs. The
use of core memory as ROM is something I am not aware of. Is anyone out
there aware of such a use of core memory. He also mentioned the use of
memory plates with ultrasonic drilling of small holes for the bits (to
be used in RAM). Is there a US counterpart to this?
Finally, he indicated that they used ferrite transistors. I was not
aware of this either. Can anyone shed light on this??
If anyone can point me to any historical references to this era in either
US or USSR developments it would be greatly appreciated.
--
John Passafiume
School of ICS, Georgia Tech, Atlanta GA 30332
CSNet: JFP @ GATech ARPA: JFP.GATech @ CSNet-Relay
uucp: ...!{akgua,allegra,rlgvax,sb1,unmvax,ulysses,ut-sally}!gatech!jfp
Henry Spencer
> as ROM where "little old ladies" wired the application program bits into
> the ROMs. The use of core memory as ROM is something I am not aware of.
> Is anyone out there aware of such a use of core memory.
This was common in specialized applications some years ago. The on-board
computers of the Apollo spacecraft used program ROMs built this way, for
example. DEC's "PDP-16" (sort of a kit for building specialized computers)
included a wire-your-own-ROM module of this type. I think some of the
telephone switching computers may have used it also. Given these examples,
I would expect that it was used quite a bit in avionics and the like.
> He also mentioned the use of
> memory plates with ultrasonic drilling of small holes for the bits (to
> be used in RAM). Is there a US counterpart to this?
If you mean ROM rather than RAM, there are a few. The 360 series used a
wide variety of strange ROM technologies for their microcode, including
one where the ROM was a sort of aluminized punchcard.
--
Henry Spencer @ U of Toronto Zoology
{allegra,ihnp4,linus,decvax}!utzoo!henry
mike knudsen
> term working in research labs in the USSR on airborne computer projects.
> In our discussions he mentioned some technogies and uses of technologies
> that I was unaware of. For example, he discussed the use of core memories
> as ROM where "little old ladies" wired the application program bits into
> the ROMs. The
> use of core memory as ROM is something I am not aware of. Is anyone out
> there aware of such a use of core memory. He also mentioned the use of
> memory plates with ultrasonic drilling of small holes for the bits (to
> be used in RAM). Is there a US counterpart to this?
> Finally, he indicated that they used ferrite transistors. I was not
> aware of this either. Can anyone shed light on this??
> If anyone can point me to any historical references to this era in either
> US or USSR developments it would be greatly appreciated.
> --
> John Passafiume
I don't think there's anything unique here, but these technologies
are a bit obscure. I remember reading (probably in Bell & Newell's
"Computer Structures" text) that one of the computers in the Apollo
spacecraft used a "wire rope ROM" like the Russian item you described.
An address de-mux shot a pulse down one of many wires (one per address).
There was one big core donut for each bit in the word. If that
address wire went thru toroid #5, then bit 5 of that addressed
word was a ONE. A very reliable technology, and very radiation,
glitch, and temperature proof. I almost used it in a home
project. I suspect there are several on the Moon today!
Bell Labs used various ferrite sheet and "twistor" schemes
for both RAM (sort of like core) and ROM in the original ESS
processors (computerized telephone exchanges) in the 60's and
early 70's. The ROMs were steel sheets with little spots of
north or south magnetism, which biased ferrite cores in the
reading frame into which they were inserted. These cores
were read like usual core RAMs. We had "ROM burners" that
were big machines that fed these sheets thru and magnetized
them. All this neat stuff has long since been scrapped
for plain old DRAM chips.
These schemes work very well -- just a bit bulky and heavy
by today's standards, but pretty solid against
radiation and EMP and power glitches in a military plane.
mike k
Darrel VanBuer
The IBM 360 series used almost every conceivable scheme possible for
implementing a ROM accross the series. This resulted from both the
incredibly broad performance range of the series and that each model was
designed by a different unit within IBM. There certainly were some with
custom threaded magnetic cores. Another used capacitors printed on a sheet
with the plate punched out to create a bit.
It sounds like the Russians may have had a few novel twists, but basically
they are simply stuck with technology a decade behind US/Japan.
--
Darrel J. Van Buer, PhD
System Development Corp.
2525 Colorado Ave
Santa Monica, CA 90406
(213)820-4111 x5449
...{allegra,burdvax,cbosgd,hplabs,ihnp4,orstcs,sdcsvax,ucla-cs,akgua}
!sdcrdcf!darrelj
VAN...@USC-ECL.ARPA
ir...@uiucdcs.cs.uiuc.edu
Core memory as ROM, yes, it will work, but it is a outdated technology.
(I will probably draw some flames here) One might compare it with the
old fashion push type reel lawn mower. There is not much market for
them any more.
One must understand the core memory, and then make a judgement. First,
the core memory takes up a lot of physical space, compared to the latest
and current ROMS. The core is made of ferrite and has a donut shape. It
requires one for each bit, and for however many bits wide the "word" is,
a row of donuts is necessary for each address. When applied, it requires
a "write winding" wound in the core, made up of wire, which requires a
"core driver". Also, there is a "sense lead" which is threaded through
the core. To read the core, one must attempt a write. If the core changes
from it's current magnetic state, the sense wire will sense the change,
so one will know that it's state changed. The sense lead by the way, must
be attached to a "sense amplifier", (one for each sense lead) which will
amplify the tiny pulse induced into the sense lead, to a usable logic level.
If changed on a read, it must be then written back to it's previous state
to preserve the original state.
In other words, to read a core, you actually attempt a write. If it changes
or not tells you what was stored there and if you changed it, you must
then put back what you just altered.
In the "heyday" of core memory, a typical 4096 X 16 bit core plane was
in the order of 6 inches square, not including any electronics.
Core will retain the store when the machine is shut down, so one can think
of it as "ROM". I think however that time required by core and physical
size (added weight in aircraft) and computers no longer the size of a
railroad boxcar, but instead the size of a lunch bucket make it an
outdated memory.
Second, it was a very good memory in it's day, but that day has passed.
P.S. I did NOT cross post this to net.space, we all read both anyway.
Why use up additional disk on everyone's machines?
Ferrite transistors, never heard of them!
Patrick Powell
Ahh!!! The good old days... Yep. I actually have used such a beasty.
We needed to build a PDP8 based system, and needed a ROM. Back in those
days, you had do it with diodes, sort of like building a gigantic
open collector NAND gate system. Well, I forget which maniac thought
of the idea, but we had a 256 word core board.
We munged the sense amps, so that they would report 1 if the core bits
were present, 0 if absent. Then into the core boards with a needle nose
diagonal cutter. We marked all of the core rings to be removed with
red nail polish first, then snip!!! away they went.
Don't!!! make a coding error.
Patrick ("Real programmers do it directly on the hardware!") Powell
--
(Prof.) Patrick Powell, Dept. CS, 136 Lind Hall, U. of Minn.,Minn.,MN 55455
-- Where winter is just a bad dream... 9 months long
jo...@anasazi.uucp
>> ...the use of core memories
>> as ROM where "little old ladies" wired the application program bits into
>> the ROMs. The use of core memory as ROM is something I am not aware of.
>> Is anyone out there aware of such a use of core memory.
stored in core. While it really was writable, it was functionally
a "core ROM" in that nobody ever wrote it more than once.
--
John Moore (NJ7E/XE1HDO)
{decvax|ihnp4|hao}!noao!terak!anasazi!john
{hao!noao|decvax|ihnp4|seismo}!terak!anasazi!john
terak!anasazi!jo...@SEISMO.CSS.GOV
(602) 861-7607 (day or evening)
7525 Clearwater Pkwy, Paradise Valley, AZ, 85253 (Home Address)
The opinions expressed here are obviously not mine, so they must be
someone else's.