Hacker News Reader

Perhaps you could point out the hazards? The generated code is longer, but so what, so long as it is semantically correct - longer code is not a hazard. And without knowing how the code was built, with which compiler, with which options etc. it's very hard to comment.

View on HN

Bigger ASM means you can fit less code in cache, which makes your code slower. If this happens in a hot loop, it may be the difference between "my code fits entirely in L1 cache" and "my code needs to be fetched from L2/L3".

This matters in high-performance scenarios. The fact that you can introduce a performance issue simply by creating a temporary is absolutely bonkers, and makes writing high-performance code more complicated than it needs to be (and god knows it's already very complicated).

Depending on the processor, it can also be the difference between instructions in a tight loop only needing to be decoded once (because of micro-op caching) vs. decoded each loop iteration.

> without knowing how the code was built

The article links out to Compiler Explorer (https://godbolt.org/), which shows the exact compiler version and options and lets you play with different ones. It shouldn’t matter which compiler — the standard prohibits optimizing this.

Reply on news.ycombinator.com