3 ms·

Are GC pauses really significant? I can understand there being problems if you have to churn through gigabytes of world data, but the author's games don't appea

Are GC pauses really significant? I can understand there being problems if you have to churn through gigabytes of world data, but the author's games don't appear to be on that scale.

Narishma 11y ago
Their unpredictablity is a bigger concern than their duration when it comes to games.
0xcde4c3db 11y ago
They often are if specific care is not taken to minimize garbage in the game loop, particularly in resource-constrained environments like some console and mobile platforms [1]. In GC languages, this often means deliberately avoiding common idioms that perform allocation under the hood. The threshold for a noticeable pause is much lower for action games than it is for general applications. 100ms is often used as a rule-of-thumb "instantaneous reaction" threshold for general UX [2], but that's a lot of time to a highly competitive player (e.g. this guy playing Super Punch-Out blindfolded [3]). In fact, in high-reliability real-time software, even non-GC heap allocation is often avoided (or done once at startup and then left alone). Heap allocation is not a very predictable operation beyond "probably fast enough". [1] https://stackoverflow.com/questions/5063190/game-jitters-on-xbox360 https://stackoverflow.com/questions/5063190/game-jitters-on-... [2] https://www.nngroup.com/articles/response-times-3-important-limits/ https://www.nngroup.com/articles/response-times-3-important-... [3] https://youtu.be/vSr3aXd4XuQ?t=358 https://youtu.be/vSr3aXd4XuQ?t=358
- weavejester 11y ago
  I can see that you'd probably have to take care on mobile and older platforms, and with resource-heavy games. But I have difficulty imaging that there'd be much issue with anything else, as long as the code wasn't overly terrible. Do you happen to know of any benchmarks, or non-anecdotal evidence?
davexunit 11y ago
Like in any program written in any language, you want to minimize allocation. People love to play the garbage collector blame game.

Are GC pauses really significant? I can understand there being problems if you have to churn through gigabytes of world data, but the author's games don't appear to be on that scale.

Narishma 11y ago

Their unpredictablity is a bigger concern than their duration when it comes to games.

0xcde4c3db 11y ago

They often are if specific care is not taken to minimize garbage in the game loop, particularly in resource-constrained environments like some console and mobile platforms [1]. In GC languages, this often means deliberately avoiding common idioms that perform allocation under the hood. The threshold for a noticeable pause is much lower for action games than it is for general applications. 100ms is often used as a rule-of-thumb "instantaneous reaction" threshold for general UX [2], but that's a lot of time to a highly competitive player (e.g. this guy playing Super Punch-Out blindfolded [3]). In fact, in high-reliability real-time software, even non-GC heap allocation is often avoided (or done once at startup and then left alone). Heap allocation is not a very predictable operation beyond "probably fast enough". [1] https://stackoverflow.com/questions/5063190/game-jitters-on-xbox360 https://stackoverflow.com/questions/5063190/game-jitters-on-... [2] https://www.nngroup.com/articles/response-times-3-important-limits/ https://www.nngroup.com/articles/response-times-3-important-... [3] https://youtu.be/vSr3aXd4XuQ?t=358 https://youtu.be/vSr3aXd4XuQ?t=358

weavejester 11y ago
I can see that you'd probably have to take care on mobile and older platforms, and with resource-heavy games. But I have difficulty imaging that there'd be much issue with anything else, as long as the code wasn't overly terrible. Do you happen to know of any benchmarks, or non-anecdotal evidence?

davexunit 11y ago

Like in any program written in any language, you want to minimize allocation. People love to play the garbage collector blame game.