Hacker News Reader

Comment by kaashif | Hacker News Reader

I don't think there's really a credible alternate reality where Moore's law just stops like that when it was in full swing.

The ones that "could have happened" IMO are the transistor never being invented, or even mechanical computers becoming much more popular much earlier (there's a book about this alternate reality, The Difference Engine).

I don't think transistors being invented was that certain to happen, we could've got better vacuum tubes, or maybe something else.

View on HN

As someone has brought up, Transputers (an early parallel architecture) was a thing in the 1980s because people thought CPU speed was reaching a plateau. They were kind of right (which is why modern CPUs are multicore) but were a decade or so too early so transputers failed in the market.

CPU cores are still getting faster, but not at the 1980/90s cadence. We get away with that because the cores have been good enough for a decade - unless you are doing heavy data crunching, the cores will spend most of the time waiting for you to do something. I sometimes produce video and the only times I hear the fans turning on is when I am encoding content. And even then, as long as ffmpeg runs with `nice -n 19`, I continue working normally as if I had the computer all to myself.

If you're on Linux, I'd highly recommend trying `chrt -i 0`. Not quite night-and-day compared to nice 19, but anecdotally it is noticeable, especially if you game.

When MC68030 (1986) was introduced, I remember reading how computers probably won't get much faster, because PCB signal integrity would not allow further improvements.

People that time were not actually sure how long the improvements would go on.

We were stuck with 33MHz PCBs for a long time as people kept trying and failing to get 50MHz PCBs to work. Then Intel came out with the 486DX2 which allowed you to run a 50MHz processor with an external 25MHz bus (so a 25MHz PCB) and we started moving forward again, though we did eventually get PCBs to go much faster as well.

The Transputers (mentioned in other comments) had already decoupled the core speed from the bus speed and Chuck Moore got a patent for doing this in his second Forth processor[1], which patent trolls later used to extract money from Intel and others (a little of which went to Chuck and allowed him to design a few more generations of Forth processors).

[1] https://en.wikipedia.org/wiki/Ignite_(microprocessor)

> We were stuck with 33MHz PCBs for a long time as people kept trying and failing to get 50MHz PCBs to work.

What is the current best symbol rates we get on PCB traces? I know we’ve been multiplexing a lot of channels using the same tricks we used with modems to get above 9600bps on POTS.

How conservative of a lower bound would next-gen PCI-e give?

Reply on news.ycombinator.com