How fast can a computer go

[Barney]

China and the USA have been challenging each other on the development of the fastest computer.

The USA may had put their head in front and has now developed a machine that can work at a mind blowing speed of
1 quintillion calculations per second.
Thats 1,000,000,000,000,000,000 / second. 18 zeros for you, like me, non math's guys.

[noosard]

is that terminal velocity

[Whistler]

Computing speed is not just a factor of CPU capabilities, most commercial systems are limited by bandwidth to handle immense amounts of data. If you want a classic example of this, start using AI in anger, it is constrained by data not compute power.

[tamada]

I'm only familiar with teraflops so I have no idea what a quintillion can do for me.

That, and my calculator doesn't have a wide enough display.

[Bandung_Dero]

The fastest my Notebooks have gone is about Mach 0.85, average cruising speed of a jet liner. 5555

[FrazeeDK]

as Moore's law continues to show it is relevant, the scientists are ensuring that data transmission and throughput also keep improving. https://www.independent.co.uk/tech/inte ... 19333.html

[Udon Map]

I'm not sure how much of a factor this is any more, given that some computers (phones) are so small, but the speed of a computer may be limited by the speed at which the electrical signals can travel over wires, circuits, or fibre. This was the primary reason that Cray super-computers were round. By making them round (well, C-shaped, actually), Seymour Cray was able to shorten the distance that signals had to travel within the computer. They were enormously successful because they were so much faster than their competition.

[tamada]

I believe the vast increases in computing speed are down to "distance" in the way that technology has allowed for thinner and thinner substrates inside the processors.

https://www.forbes.com/sites/antonyleat ... rformance/

https://www.tomshardware.com/tech-indus ... 5-and-2026

[Whistler]

These days parallel computing has won the race. Seymour Cray initially worked for IBM and alongside Dr An Wang built the first big IBM mainframes. Dr Wang invented core memory and the electronic calculator. The Cray was a mono CPU, rather crude CPU. It cycled so fast it kept cool by encasing circuit boards in liquid nitrogen, it pushed single CPU tecnology about as far as possible. It also predated the silicon chip. When the Digital Vax 11/780 came out in the late 1970's it introduced MULTICs, parallel computronics. Later you could cluster multiple CPUs and massively and dynamically increase compute power.

AWS and Microsoft Azure now deliver parallel computing on a massive scale with server farms consisting of thousands of CPU'S hugely scalable computing. It harnessed cluster computing invented by Digital.

The first 11/780 in Australia went into Mt Stromlo observatory in Caberra, it was almost was as fast as a Cray but at a twentieth the price. The first commercial 11/780 in the world went into Illawarra County Council in August 1979 and blew their old mainframe to pieces.

[Udon Map]

Seymour Cray initially worked for IBM and alongside Dr An Wang built the first big IBM mainframes.

I think you may have it mixed up a bit. Cray worked for CDC, where he did his development and research. His first big success in coaxing speed out of computers was the CDC 6600. CDC's main competitor at the time was IBM. IBM's mainframe architecture (the System 360 line, followed by the 370 and eventually the 308x series) was designed by Gene Amdahl, who eventually left IBM to do his own thing, founding Amdahl Corp., where he developed computers which competed directly with the 360 series and successor machines.

Dr. Wang wasn't an IBMer. He started Wang Labs right out of Harvard and never worked anywhere else.

(In a previous life, I was a computer geek. In the late 1960s through the 1970s, when all of this was going on, I was in the thick of it, living this stuff daily.)

[Whistler]

Interesting. I got the IBM thing from Fred Wang who was Wang computers VP in Dallas. Wang was certainly involved in commercial ventures with IBM and he eventually sold the patent for core memory to IBM in the mid 1950's I never questioned what Fred told me, however, Fred was a real party animal and I had a few sessions with him when I lived in Texas. He also told me that Wang senior was a mate of Cray and they both used to enjoy designing systems and helped design the 1401. I see on his biography that he while he sold the patent of core memory to IBM but says nothing on the 1401 or actually working for IBM. Maybe Fred was BSing or my memories were fogged by tequila slammers.

I was also told that US intelligence convinced Dr Wang to leave China as they did not want this talented man to work for the communists after WW2.

I operated 1401s in the mid 1960s at Colgate Palmolive, we upgraded to a 360/30 in 1966 when I started programming.

[tamada]

Maybe you're talking about the wong Wang?