The Illiac IV Computer

I have been attempting to locate software for the Illiac IV. I corresponded with as many of the people involved with it I could easily find online and with NASA Ames (which has a history office!).

I've been looking for software such as this:

ASK - assembler for the Illiac IV on the Burroughs B5500 and later B6700

SSK - simulator for the Illiac IV on the Burroughs B5500 and later B6700

CFD - a FORTRAN like language targeted at CFD

Glypnir - ALGOL like language, likely not used at Ames

VECTORAL - vector processing language used at Ames

So far nothing has been found except the manual for CFD.

> Generally considered a failure due to massive budget overruns, the design was instrumental in the development of new techniques and systems for programming parallel systems. In the 1980s, several machines based on ILLIAC IV concepts were successfully delivered.

Such as it ever was. Breaking boundaries takes guts, a lot of time and resources, and is rarely appreciated in the moment for the advances generated (even if a 'failure' by original operating criteria).

Article doesn't mention SILLIAC, a similar computer at Sydney University.

> Like most of the IAS family, SILLIAC was not an exact copy of ILLIAC. One important change was the use of 2C51[6] valves in place of the more common 6J6.[7] The 2C51 had been developed by Bell Labs for use in undersea telephone repeaters and had about 5 times the life (for 6 times the cost). This decision significantly improved the reliability of SILLIAC compared to its contemporaries.

https://en.wikipedia.org/wiki/SILLIAC

I've been reading The Dream Machine by Mitchell Waldrop and it tells the story of early computing centered on J.C.R. Licklider. It's an easy recommendation for the HN audience.

If your eyesight is less than perfect I recommend buying both the hardcover and the digital version of the book. The hardcover is printed by Stripe (yes, the payments people) and it's a beautiful thing but the text is small.

> The ILLIAC IV was the first massively parallel computer. The system was originally designed to have 256 64-bit floating point units (FPUs) and four central processing units (CPUs) able to process 1 billion operations per second.

Modern consumer-grade GPUs and CPUs have attained these numbers, eh? It's great to see that we have achieved this goal.

It was the first network-available supercomputer, beating the Cray-1 by nearly 12 months.