Jonathan Blow

With 64 bit address spaces, thread stacks should have GBs reserved by default, and programming with alloca should be encouraged. A lot of string and container work could be done at high performance with alloca and no frees if you don't let them escape.

Having a per-thread temporary allocator that is a standard convention, so that everyone including libraries can use it, is better... you can return temporary strings / arrays / nodes from functions, etc. (I do this all the time in the language I am working on).

You just pretend like you have a GC for all these things, and it just gets completely reset once per frame at a known point.

We had this on one of the engines I worked on, a per frame linear buffer that would reset at a known time. It was shared among threads though, and it would be double buffered so GPU could access it on a different frame as well.

The problem with sharing across threads is that now it's a contended resource. Also the threads have to sync to reset it, but, if that is part of your design already then it is fine.

It depends on how you use your threads I guess. That linear buffer was used for all the jobs and memory needed to be passed around between them. Jobs could also have their own local linear allocator. Long running threads would not use this memory.