TROS: How IBM mainframes stored microcode in transformers

[Marco Scholz]

Ken Shirriff's blog article about a Transformer Read-Only Storage
(TROS) module that stored microcode in an IBM System/360 mainframe
computer.

TROS: How IBM mainframes stored microcode in transformers
http://www.righto.com/2019/11/tros-how-ibm-mainframes-stored.html?m=1

[Douglas Miller]

Very similar technology was used by Wang Laboratories for the ROM microcode in their early calculators. Doesn't really matter who was first, but I have heard that An Wang and IBM crossed paths on several occasions - mostly amenably, I think. There were occasional "patent swap fests" I believe.

[Marco Scholz]

Post by Douglas Miller
Very similar technology was used by Wang Laboratories for the ROM
microcode in their early calculators.

[...]

Wang calculators were microprogrammed? Wow, I didn't know that. Did they
use core rope memory?

The idea of microprogramming the control unit of a computer dates back
to the early 1950s. First machines were the ESDAC and Zuse Z22 - IIRC.
The ESDAC used magnetic core, the Z22 magnetostatic delay lines.

All 1964 IBM System/360 models were microprogrammed to make them
compatible. They used Transformer Read-Only Storage (TROS), Balanced
Capacitor Read-Only Storage (BCROS) and Card Capacitor Read-Only
Storage (CCROS).

Fun fact: IBM invented the 8" floppy disk to distribute microcode
patches.


Interesting read:
IBM's 360 and Early 370 Systems. Pugh, Johnson, Palmer. MIT Press 1991
https://bit.ly/3rA6wz6
https://www.researchgate.net/publication/274770131_IBM's_360_and_Early_370_Syste
ms_By_Emerson_W_Pugh_Lyle_R_Johnson_and_John_H_Palmer_Cambridge_Mass_MIT_Press_1
991_xx_810_pp_Charts_illustrations_appendixes_notes_references_and_index_3750


D985D9B7F64A18003122D7E51C3EEBEA41E77EEE 00010011 00010011 ▞

[Christian Corti]

Post by Marco Scholz
Wang calculators were microprogrammed? Wow, I didn't know that. Did they
use core rope memory?

Yes, see:
http://computermuseum.informatik.uni-stuttgart.de/dev/wang700/wang700_rom.html

Christian

[[e-mail]]

Post by Marco Scholz

Post by Douglas Miller
Very similar technology was used by Wang Laboratories for the ROM
microcode in their early calculators.

[...]
Wang calculators were microprogrammed? Wow, I didn't know that. Did they
use core rope memory?
The idea of microprogramming the control unit of a computer dates back
to the early 1950s. First machines were the ESDAC and Zuse Z22 - IIRC.
The ESDAC used magnetic core, the Z22 magnetostatic delay lines.

I remember in the 1970s MV Wilkes bringing the original transformer setup into his lectures to show it to the students.

[Marco Scholz]

Post by Douglas Miller
Very similar technology was used by Wang Laboratories for the ROM
microcode in their early calculators.

[...]

Wang calculators were microprogrammed? Wow, I didn't know that. Did they
use core rope memory?

The idea of microprogramming the control unit of a computer dates back
to the early 1950s. First machines were the ESDAC and Zuse Z22 - IIRC.
The ESDAC used magnetic core, the Z22 magnetostatic delay lines.

All 1964 IBM System/360 models were microprogrammed to make them
compatible. They used Transformer Read-Only Storage (TROS), Balanced
Capacitor Read-Only Storage (BCROS) and Card Capacitor Read-Only
Storage (CCROS).

Fun fact: IBM invented the 8" floppy disk to distribute microcode
patches.

Interesting read:
IBM's 360 and Early 370 Systems. Pugh, Johnson, Palmer. MIT Press 1991
https://bit.ly/3rA6wz6

[Andreas Kohlbach]

Post by Marco Scholz
Ken Shirriff's blog article about a Transformer Read-Only Storage
(TROS) module that stored microcode in an IBM System/360 mainframe
computer.
TROS: How IBM mainframes stored microcode in transformers
http://www.righto.com/2019/11/tros-how-ibm-mainframes-stored.html?m=1

Does it work with Autobots and Decepticons alike? ;-)

SCNR

[m. thompson]

Post by Marco Scholz
Ken Shirriff's blog article about a Transformer Read-Only Storage
(TROS) module that stored microcode in an IBM System/360 mainframe
computer.
TROS: How IBM mainframes stored microcode in transformers
http://www.righto.com/2019/11/tros-how-ibm-mainframes-stored.html?m=1
--
D985D9B7F64A18003122D7E51C3EEBEA41E77EEE 00010011 00010011 ▞

DEC used transformers to store microcode in the PDP-9 and PDP-15. There were 64 36-bit words. In the PDP-9 four words were never used, and there were rumors of customer modifications to the transformer wiring to add instructions.

[Peter Flass]

Post by m. thompson

Post by Marco Scholz
Ken Shirriff's blog article about a Transformer Read-Only Storage
(TROS) module that stored microcode in an IBM System/360 mainframe
computer.
TROS: How IBM mainframes stored microcode in transformers
http://www.righto.com/2019/11/tros-how-ibm-mainframes-stored.html?m=1
--
D985D9B7F64A18003122D7E51C3EEBEA41E77EEE 00010011 00010011 ▞

DEC used transformers to store microcode in the PDP-9 and PDP-15. There
were 64 36-bit words. In the PDP-9 four words were never used, and there
were rumors of customer modifications to the transformer wiring to add instructions.

64 words? That’s not a heck of a lot of microcode!

[John Levine]

Post by m. thompson
DEC used transformers to store microcode in the PDP-9 and PDP-15. There
were 64 36-bit words. In the PDP-9 four words were never used, and there
were rumors of customer modifications to the transformer wiring to add instructions.

64 words? That’s not a heck of a lot of microcode!

The PDP-9 only had a four bit opcode, 13 instructions that reference memory, an opcode
for a group of "operate" instructions that manipulated and tested the accumulator, an
opcode for I/O instructions, and an optional extended arithmetic feature.

[Bill Findlay]

On 15 Mar 2021, Peter Flass wrote
(in

Post by m. thompson

Post by Marco Scholz
Ken Shirriff's blog article about a Transformer Read-Only Storage
(TROS) module that stored microcode in an IBM System/360 mainframe
computer.
TROS: How IBM mainframes stored microcode in transformers
http://www.righto.com/2019/11/tros-how-ibm-mainframes-stored.html?m=1
--
D985D9B7F64A18003122D7E51C3EEBEA41E77EEE 00010011 00010011 ???

DEC used transformers to store microcode in the PDP-9 and PDP-15. There
were 64 36-bit words. In the PDP-9 four words were never used, and there
were rumors of customer modifications to the transformer wiring to add instructions.

64 words? That´s not a heck of a lot of microcode!

KDF9's Main Control Unit had 64 BITS of microcode (in transformers).

[m. thompson]

Post by Bill Findlay
KDF9's Main Control Unit had 64 BITS of microcode (in transformers).
Bill Findlay

The Microcode in a PDP-9 is 64 words (60 used) and 36-bit words. I booted a real PDP-9 last Saturday...

[Bill Findlay]

Post by m. thompson

Post by Bill Findlay
KDF9's Main Control Unit had 64 BITS of microcode (in transformers).
Bill Findlay

The Microcode in a PDP-9 is 64 words (60 used) and 36-bit words. I booted a
real PDP-9 last Saturday...

I have to (somewhat) recant: the KDF9 case is more complicated than I
implied.

Main Control was one of several concurrently running units,
each having its own microcode matrices.

Moreover, although each Main Control matrix had 64 pulse transformers,
each transformer could (and did) have multiple outputs and the
outputs could take part in several instruction execution sequences.

There were microcode matrices also in Arithmetic Control, Shift Control,
the Multiply/Divide unit, I/O Control, and in each up up to 16 "buffers",
which were actually DMA device controllers.

The latter had very optimised implementations.
For example a tape punch buffer had two arrays of only 5 transformers,
each accessed by a 3 bit microcode address.

[Peter Flass]

Post by m. thompson

Post by Bill Findlay
KDF9's Main Control Unit had 64 BITS of microcode (in transformers).
Bill Findlay

The Microcode in a PDP-9 is 64 words (60 used) and 36-bit words. I booted
a real PDP-9 last Saturday...

Probably horizontal microcode, but still, I’d be interested to see what
could be done in 60 words.