Ask HN: In which types of software is raw computational power the bottleneck?

10 points by cloogshicer 5y ago ↗ HN

Things that come to my mind:

- ML/AI

- Games (but only some)

- Scientific simulations

Any other ideas?

8 comments

[ 3.5 ms ] story [ 26.0 ms ] thread

yuppie_scum 5y ago ↗

Crypto right?

cloogshicer 5y ago ↗

Really? How so? (Genuine question, not being snarky)

mikewarot 5y ago ↗

Cryptography - Not really, encoding data can happen at quite high speed these days.

Cryptocurrency - "Proof of Work" will ALWAYS be a hard problem, the faster you make your solved, the harder the problem scales.

geoalchimista 5y ago ↗

Anything that involves solving the Navier-Stokes equations, e.g., weather forecast, hurricane prediction

d--b 5y ago ↗

- Video processing

- Ray tracing

- Data mining

- Data compression

- Financial market modeling

- Any kind of large-scale optimization

errantmind 5y ago ↗

Agreed, financial modelling for sure. There is a breakpoint in terms of development cost and velocity when data exploration / processing can be done on a single machine vs. needing to scale out across multiple machines. I've had a lot of success optimizing my data oriented projects to being able to run on a single beefy computer.

mbrodersen 5y ago ↗

Optimisation problems (pairing/rostering).

whateveracct 5y ago ↗

Games are kinda, but in terms of development it's definitely bound by creativity and art.