Algorithms Versus Buying RAM

Optimization is almost always virtuous when directed at large numbers of consumers, but for in-house applications it can be a poor use of time -- multithreaded, pipelined raw IO striped over multiple SATA channels versus just buying a fast NVMe drive. Still fun, though! "Algorithms are for people who don't know how to buy RAM."

A Twitter exchange shows John Carmack discussing elaborate optimization for in-house applications, comparing multithreaded raw I/O across SATA channels with simply buying a fast NVMe drive. The reply jokes that algorithms are unnecessary when you can solve performance problems by buying more RAM, mocking overengineering and hardware spending. EN: A Twitter exchange shows John Carmack discussing elaborate optimization for in-house applications, comparing multithreaded raw I/O across SATA channels with simply buying a fast NVMe drive. The reply jokes that algorithms are unnecessary when you can solve performance problems by buying more RAM, mocking overengineering and hardware spending.

• @ithueti / 1895 2020-01-19
• @right_wing_dev_squads / 4741 2022-08-26
• @ithueti / 7591 2022-08-29