-
As part of the work I was doing on [minimum alignment](https://github.com/fitzgen/bumpalo/pull/242), I wrote a [blog post](https://coredumped.dev/2024/03/25/bump-allocation-up-or-down/) analyzing bump…
-
For reasons of efficiency, events in `evenio` are allocated in a [`bumpalo`](https://docs.rs/bumpalo/latest/bumpalo/) bump allocator. This allocator is cleared once the top level `World::send` call re…
rj00a updated
3 months ago
-
I propose to explicitly allow to allocate and deallocate with different implementations of `Allocator` trait under specified circumstances. I'm not sure how exactly it must be done, but I guess it wou…
-
Currently there's only `Allocator` trait that provides both allocations and deallocations.
And `Box`, `Vec` and other types has `A: Allocator` generic parameter.
However there are allocators with …
-
I wonder how much time could be gained by having all formula data stored using arenas. We should investigate using `bumpalo` for this.
-
-
Write a malloc to be used in kernel space. There's a bunch of ways to do this and it might be worthwhile to have multiple implementations; it'd be an easy feature to implement.
There's a reference …
-
Many groups want a simple "bump allocator" which can be used for per-frame data.
This should be sufficiently simple to setup and also maintain safety.
* The bump alloc has a pile of data and an …
-
TLDR: CUDA memory API is slow, and Halide allocates and frees all intermediate buffers on the fly _within_ the pipeline. Hence, buffer reuse is not in my control. Envisioned elegant solution: have a b…
-
Early experiments with `jemalloc`, `mimalloc`, `snmalloc`, and `rp_malloc` show potential up to 50% improvements. This is understandable when we optimize to the microseconds.
Currently, `jemalloc` …