.ETA10 system reference manual, PUB-1005, ETA Systems, Inc., 1987

[luke.l...@gmail.com]

folks i am embarrassed to say that after considering myself to have
become somewhat of an afficionado of Vector ISAs, there is one i
have only just come across, called the ETA10-Q:

https://dl.acm.org/doi/10.1145/76263.76352

apparently ETA Systems was a spinoff of CDC, and lost money and
even had two of its systems destroyed by JVNC when no buyer could
be found for them https://en.wikipedia.org/wiki/ETA10

i'm looking rather urgently for the ISA so as to ensure it is preserved for
posterity online, apparently it's available at the Computer History Museum
http://50.204.185.175/collections/catalog/102641713

if anyone knows where (or how) this can be made available online please do
advise.

best,

l.

[Quadibloc]

On Wednesday, June 29, 2022 at 2:40:22 AM UTC-6, luke.l...@gmail.com wrote:

> if anyone knows where (or how) this can be made available online please do
> advise.

Al Kossow, who maintains Bitsavers, the large online collection of computer
documentation, also has a close connection to the Computer History Museum.
So if it is possible, no doubt he will make scanning it a priority. From what
information I could find online, apparently it's quite similar to the Cyber 205.

John Savard

[Quadibloc]

I've been informed that the correct procedure to follow would be to send an E-mail
request to rese...@computerhistory.org about this.

John Savard

[Al Kossow]

On Wednesday, June 29, 2022 at 1:40:22 AM UTC-7, luke.l...@gmail.com wrote:
> folks i am embarrassed to say that after considering myself to have
> become somewhat of an afficionado of Vector ISAs, there is one i
> have only just come across, called the ETA10-Q:

you can find a decent technological history of ETA here
https://ethw.org/First-Hand:The_First_CMOS_And_The_Only_Cryogenically_Cooled_Supercomputer

The ETA10-Q was an air-cooled lower performance machine
P 24nS cycle time 750MFLOPS
Q 19nS cycle time 947MFLOPS
E 10.5nS cycle time 6858MFLOPS
G 7nS cycle time 10286MFLOPS

we have an ETA10 in the CHM collection

https://www.computerhistory.org/collections/catalog/X1691.99B

[Al Kossow]

> you can find a decent technological history of ETA here
> https://ethw.org/First-Hand:The_First_CMOS_And_The_Only_Cryogenically_Cooled_Supercomputer
>

also, the papers in
http://bitsavers.org/pdf/cdc/cyber/cyber_200/Programming_Methodology_for_CDC_CYBER_205_Vector_Processor_Aug80.pdf

[luke.l...@gmail.com]

Al, and John, thank you so much. the computer history museum very kindly
scanned the ETA10 ISA document and made it available for anyone to access.
i'm astounded and delighted that they were able to respond so quickly on
such a rare and historic machine.

l.

[luke.l...@gmail.com]

i took a quick look and my initial assessment is that ETA10 had a Scalable (variable-length)
Memory-to-Memory Vector ISA

* Page 3-220 of its ISA shows that it had Predicate Masks and Horizontal Reduction.

* Vector operands were designated A,B (source) and C (destination)

* Appendix H-1 shows it is likely a Memory-to-Memory Vector Architecture, and overcame the
penalties normally associated with this by adding an explicit "Vector operand forwarding/chaining"
instruction (Page 3-69). It is however clearly Scalable, up to Vector elements of 2^16.

i'm profoundly astonished.

l.