From Newsgroup: comp.arch

But what I learned about her wasn't terribly important.
But along with that, I learned that before writing her book about
VLSI, she invented something at IBM. But she didn't get credit for
it...
However, I'm not at all sure just _what_ she invented.
I remember IBM's Future Systems project. James Martin went around
telling people how it would change the world. IBM ended up cancelling
the project as impractical... but they didn't throw all their work
away. Instead, it just got scaled down a bit, and the AS/400 did end
up as a real IBM project.
There was another IBM project that didn't come to fruition. IBM was
going to try to build a supercomputer as a prestige project. The first iteration of the Advanced Computer System was ACS-1, which was going
to have a new architecture; the second iteration, ACS-360, was going
to be compatible with the IBM/360.
IBM cancelled it, and Gene Amdahl was so unhappy, he left the company
and started Amdahl.
The important thing, though, is that the ACS project appears to have
come *after* the System/360 Model 91. You know, the one that embodied
the Tomasulo Algorithm. The one for which out-of-order... well, if you
don't count the 6600, of course, which Mitch Alsup will definitely
complain about.
Anyhow... perhaps it was *superscalar* and not OoO that Lynn Conway
invented during her work on ACS.
But I'm not exactly sure what she invented. Wikipedia named her
parentss at the time of her birth, so I had the surname part of her
deadname to help me in my efforts to search for old publications... it
didn't help.

John Savard
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.arch

On Wed, 17 Jun 2026 20:13:53 GMT, quadibloc@invalid.com (John Savard)
wrote:

The important thing, though, is that the ACS project appears to have
come *after* the System/360 Model 91.

I now see that Lynn Conway invented Dynamic Instruction Scheduling (or Handling) in 1965, while Tomasulo's Algorithm dates from 1967.

So this isn't an extension of OoO to superscalar, but an important
step in the development of OoO.

John Savard
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.arch

On 2026-Jun-17 16:32, John Savard wrote:

On Wed, 17 Jun 2026 20:13:53 GMT, quadibloc@invalid.com (John Savard)
wrote:

The important thing, though, is that the ACS project appears to have
come *after* the System/360 Model 91.

I now see that Lynn Conway invented Dynamic Instruction Scheduling (or Handling) in 1965, while Tomasulo's Algorithm dates from 1967.

So this isn't an extension of OoO to superscalar, but an important
step in the development of OoO.

John Savard

A copy I found a while back...

Dynamic Instruction Scheduling, Conway 1966 https://ai.eecs.umich.edu//people/conway/ACS/DIS/DynInstSched-paper.html https://ai.eecs.umich.edu//people/conway/ACS/DIS/DIS.pdf



--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.arch

On Wed, 17 Jun 2026 20:13:53 GMT, John Savard wrote:

Instead, it just got scaled down a bit, and the AS/400 did end up as
a real IBM project.

Maybe System/38, to begin with?
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.arch


quadibloc@invalid.com (John Savard) posted:

On Wed, 17 Jun 2026 20:13:53 GMT, quadibloc@invalid.com (John Savard)
wrote:

The important thing, though, is that the ACS project appears to have
come *after* the System/360 Model 91.

I now see that Lynn Conway invented Dynamic Instruction Scheduling (or Handling) in 1965, while Tomasulo's Algorithm dates from 1967.

Many people who have read about 360/91 (or 370/195) using Thomosulo
as the FP dynamic scheduler. Most do not remember that Anderson wrote
on the dynamic scheduler on the integer side--I wonder if Lynn had her
ands in there....

So this isn't an extension of OoO to superscalar, but an important
step in the development of OoO.

John Savard

--- Synchronet 3.22a-Linux NewsLink 1.2