19 Dec
2006
19 Dec
'06
4:26 p.m.
Brent Hilpert wrote: I always
kind of liked the idea of making an entire processor out of just one basic
gate/IC type.
-------------------------
Didn't Seymour Cray do this on one of his later machines? The Cray 2?
Seems like it was a 5 input And gate.
But then he always did like to minimalize. The 6600, his first big iron,
used one transistor type tied together in different ways to make gates, flip
flops and timing circuits. He used circles and squares to denote gates -
but whther they were And or Ors depended on what fed them. Ah, twas fun to
work with. And I loved the esthetics of the logic diagrams. All those
little squares and circles had a kind of magic to them.
Billy
19 Dec
19 Dec
4:57 p.m.
On 19 Dec 2006 at 16:26, Billy Pettit wrote:
Didn't Seymour Cray do this on one of his later
machines? The Cray 2?
Seems like it was a 5 input And gate.
Before that, Neil Lincoln was talking up and showing his "Chiclet"
apprach for the Cyber 203. The basic idea was that the whole shebang
was to be built out of a selection of ICs that could be held in a box
of Chiclets chewing gum (IIRC, something like 13 different types).
As a test vehicle, I believe he did an implementation of a 6400 CPU.
Cheers,
Chuck
6:29 p.m.
Billy Pettit wrote:
Brent Hilpert wrote: I always
kind of liked the idea of making an entire processor out of just one basic
gate/IC type.
-------------------------
Didn't Seymour Cray do this on one of his later machines? The Cray 2?
Seems like it was a 5 input And gate.
But then he always did like to minimalize. The 6600, his first big iron,
used one transistor type tied together in different ways to make gates, flip
flops and timing circuits. He used circles and squares to denote gates -
but whther they were And or Ors depended on what fed them. Ah, twas fun to
work with. And I loved the esthetics of the logic diagrams. All those
little squares and circles had a kind of magic to them.
Billy
I have a web page (http://www.cs.ubc.ca/~hilpert/e/gates/index.html)
where I'm accumulating the diversity of symbols that have been used for
logic gates. If you have a scan or reference for the Cray/6600 symbols I'd be
interested in adding them.
7:43 p.m.
Didn't Seymour Cray do this on one of his later
machines? The Cray 2?
Seems like it was a 5 input And gate.
I think it was the early Cray-1 line. Four chip types - 2 gates, a
flip flop, and a memory chip.
To me, it seems like a really dumb idea. If he could get the chip
house (Fairchild?) to make a JK flip flop, certainly he could have had
them make some more complex gates, like AOIs or adders. With a few
more chip types, I bet the Cray-1 could have been a much better
machine (cheaper, more reliable, maybe a tad faster).
For that matter, one has to wonder why he just did not use off the
shelf MECL III. Methinks it was NIH syndrome.
--
Will
8:22 p.m.
In article <e1d20d630612191943r673776e4qa28990a829b95162 at mail.gmail.com>,
"William Donzelli" <wdonzelli at gmail.com> writes:
To me it seems like a smart idea. Concentrate on a few chips so that
you can get the manufacturing yields up, the cost per component down,
and fewer parts to stock and manage from manufacturing all the way
through repair servicing. There are lots of overhead factors that
increase with the number of kinds of parts you have.
--
"The Direct3D Graphics Pipeline" -- DirectX 9 draft available for download
<http://www.xmission.com/~legalize/book/download/index.html>
Legalize Adulthood! <http://blogs.xmission.com/legalize/>
Didn't
Seymour Cray do this on one of his later machines? The Cray 2?
Seems like it was a 5 input And gate.
I think it was the early Cray-1 line. Four chip types - 2 gates, a
flip flop, and a memory chip.
To me, it seems like a really dumb idea. [...]
8:41 p.m.
To me it seems like a smart idea. Concentrate on a
few chips so that
you can get the manufacturing yields up, the cost per component down,
and fewer parts to stock and manage from manufacturing all the way
through repair servicing. There are lots of overhead factors that
increase with the number of kinds of parts you have.
Adding just ten more chip types will not significantly increase
manufacturing overhead. How many parts total were on the Cray-1 BOM?
10,000? 20,000? making it 10,010 or 20,010 is not going to strain the
inventory folks.
Also, by adding just ten more chip type, the total number of chips in
the system will have gone down - perhaps by ten percent. That will
greatly increase yield. Back in those hand-placed hand-soldered
surface mount days, board errors were far more likely to be bad solder
joints (using that old fashioned solder! horrors!) than misplaced
parts.
One of the Cray-1s faults was the somewhat miserable servicing
requirements, and the crummy downtime that forced it. Fewer parts
would have increased uptime (although many of the issues with the
machines were not chip related).
--
Will
20 Dec
20 Dec
1:49 p.m.
"William Donzelli" writes:
I think it was the early Cray-1 line. Four chip types
- 2 gates, a
flip flop, and a memory chip.
To me, it seems like a really dumb idea. [...]
Richard wrote:
To me it seems like a smart idea. Concentrate on a
few chips so that
you can get the manufacturing yields up, the cost per component down,
While you two duke it out over the manufacturing practicalities :), I just have
to add that my interest in a processor made from a single gate type was just
for the esthetics of a 'universal machine made from a universal gate', or (to
be high-falutin' about it) a physical embodiment of a fundamental principle.
While the Cray-1 may have been a little late for such a design philosophy,
apparently there was a period in IC development where it made practical sense.
Aside from the AGC, I have an application note from RCA from 1967 for a small logic
family (ECCSL Emitter-Coupled Current Steered Logic) composed of just 3 IC types:
CD2150 dual 4-in gate
CD2151 dual 4-in gate (different output drive specs from 2150)
CD2152 8-in gate
The app note includes diagrams for building standard functions (flip-flops, shift
register, counters, adders, etc.) from only those ICs.
No idea what equipment/machines they may have been used in.
2:20 p.m.
On 20 Dec 2006 at 13:49, Brent Hilpert wrote:
> While the Cray-1 may have been a little late for
such a design
philosophy,
apparently there was a period in IC development where
it made practical sense.
Aside from the AGC, I have an application note from RCA from 1967 for a small logic
family (ECCSL Emitter-Coupled Current Steered Logic) composed of just 3 IC types:
Some families of ECL were built along this line with gates with up to
6 outputs (NAND/AND and NOR/OR) that could be wire-ORed to form all
sorts of interesting functions. (e.g. Moto MC1001).
Cheers,
Chuck
19 Dec
19 Dec
11:40 p.m.
On Dec 19, 2006, at 7:26 PM, Billy Pettit wrote:
Brent Hilpert wrote: I always
kind of liked the idea of making an entire processor out of just
one basic
gate/IC type.
-------------------------
Didn't Seymour Cray do this on one of his later machines? The Cray 2?
Seems like it was a 5 input And gate.
But then he always did like to minimalize. The 6600, his first big
iron,
used one transistor type tied together in different ways to make
gates, flip
flops and timing circuits. He used circles and squares to denote
gates -
but whther they were And or Ors depended on what fed them. Ah, twas
fun to
work with. And I loved the esthetics of the logic diagrams. All
those
little squares and circles had a kind of magic to them.
Actually I believe this was the case for the Cray 1 and earlier
XMP machines.
-Dave
--
Dave McGuire
Cape Coral, FL
31 Dec
31 Dec
12:16 p.m.
Billy Pettit wrote:
But then he always did like to minimalize. The 6600,
his first big iron,
used one transistor type tied together in different ways to make gates, flip
flops and timing circuits. He used circles and squares to denote gates -
but whther they were And or Ors depended on what fed them. Ah, twas fun to
work with. And I loved the esthetics of the logic diagrams. All those
little squares and circles had a kind of magic to them.
I am not sure what the computer was but it was the first he one he built.
It was all using the same surplus transistor that he got very cheap. The
second computer was a problem since they had no more surplus transistors
to sell him.
Billy
.
6347
days inactive
6358
days old
9 comments
7 participants
participants (7)
-
bfranchuk＠jetnet.ab.ca -
Billy.Pettit＠wdc.com -
cclist＠sydex.com -
hilpert＠cs.ubc.ca -
legalize＠xmission.com -
mcguire＠neurotica.com -
wdonzelli＠gmail.com