A quick lookup for CPU speeds of yesteryear shows the Motorola 68040 in those macs did about 35 MFLOPS [1].

If you could parallelize this program to use all 2.6 TFLOPS in a Macbook Air M1's 8-core GPU, you're seeing about 75,000x speed improvement. So a 12 hour "overnight" job may take just over half a second.

[1] https://tech-insider.org/mac/research/1991/1021-e.html

I generated two random 35 digit primes, multiplied them, and then tossed the product into the first hit for "factor integer online": https://www.alpertron.com.ar/ECM.HTM

1731 322587 637083 705865 442033 654804 904216 446486 784497 444181 466938 498269 (70 digits) = 25317 847509 490819 861309 522806 413689 (35 digits) × 68383 482718 586893 688113 139227 257221 (35 digits) Time elapsed: 0d 0h 0m 23.8s

This is in JavaScript on a single core on my laptop.

Ran the same program and input on my phone.

Time elapsed: 0d 0h 5m 38.2s

Took about 50s on my iPhone.

I'd very much like to revive my old code (which I still have) at some point and see how well it runs on modern computers. My guess is that a typical 64-bit ~3 GHz quad-core machine could accomplish the same task in a few minutes today.

Hate to be the choosing beggar, and I appreciate all the answers, but ... everyone is answering about how long that 70 digit job would take now, when what I meant was, how big a number could you factor “overnight” with similar cost hardware.

Not exactly the same, but close.

https://github.com/eniac/faas

The purpose of the FaaS (Factoring as a Service) project is to demonstrate that 512-bit integers can be factored in only a few hours, for less than $100 of compute time in a public cloud environment.

512b = 155 decimal digits.

Cool, thanks!

Or on a single modern smartphone?

On that note, I'd like to share a clip about how fast modern machines are relative to early 90s supercomputers (the relevant part is in the first 5 minutes):

https://youtube.com/watch?v=oJ4GcZs7y6g

Ha, I did a small bunch of calculations on the CM5. Thanks for the memories.