Hacker News Reader

Comment by johnklos | Hacker News Reader

It's simply amazing what people did with the limited computing they had available to them at the time. Even when it was available, you only got an allotment of it and had to figure out how to do everything you wanted within that allotment.

Here we are with many orders of magnitude of computing power that we have to ourselves, 24/7, and mostly we're using all that power to browse the Internet :P

Calculating the first 50,000,000 primes takes less than ten minutes (using no memory - that is, not a sieve). The 50,000,000th prime, BTW, is 982,451,653. I wonder what the author of this paper would've been able to do with the kind of processing available to us.

View on HN

The IBM mainframe I had access to in the 80s at University of Illinois of Chicago had quotas of compute time (that reset on a weekly basis). Since I worked in the computer center, I had an account that had a larger quota (or possibly unlimited, I don’t remember now, perhaps the unlimited was when I advanced to a staff level?). One of the more entertaining things was when one class started having people code in Ada. Unfortunately, the Ada compiler was a CPU hog and a single compilation was enough to use up a week’s quota. As a result, students had to work in teams and share access to their online disks (VM/CMS did allow different modes that allowed privacy for things like email to be kept even while allowing read access to the rest of the stuff on the disk—also noteworthy was the lack of subdirectories). The team members had to switch to a different account after each day‘s work since they would exhaust each account’s quota. If you inadvertently logged out too soon, it was possible that a homework assignment could not be completed.

> Even when it was available, you only got an allotment of it and had to figure out how to do everything you wanted within that allotment.

IIRC computing primes was a popular way to test hardware; it’s fairly easy to compare results between machines, and both having a faster CPU, more CPUs and having more memory (simple example: if you do trial division, you can keep a larger table of ‘small’ primes around to quickly weed out most integers)) will speed up computations.

It then sort-of became a marketing goal to beat your competitors, so cleverer and cleverer algorithms were developed.

Because of that, those records had less trouble with resource allotments.

Reply on news.ycombinator.com