Re:Generics can make your Go code slower
by siftrics on Wednesday March 30, 2022 @02:14PM
(#30858785)
Assuming your generic functions take _pointers_ to Components as input, full monomorphization does not occur and you're suffering a performance hit similar in magnitude, if not strictly greater empirically, to interface "dereferences".
On this basis, I don't believe your generic implementation is as faster than an interface implementation as you claim.
Re:Generics can make your Go code slower
by pphysch on Wednesday March 30, 2022 @03:08PM
(#30859522)
You're right, here's what one of my hot loops look like:
func (cc *ComponentContainer[T]) ForEach(f func(*Component[T])) {
for _, page := range cc.pool.pages {
for i := range page {
if page[i].IsActive() {
f(&page[i])
}
}
}
}
Still, the interface approach is a total nightmare from a readability + runtime error perspective so I won't be going back & will just hope for some performance freebies in 1.19 or later :^)Just to clarify things here: a pointer receiver on a generic type isn't what was covered in the article. That was covering if your `T` is a pointer type.
If T here is a value-type, you're probably getting monomorphization (at the cost of copying the value all over, and possible benefit of inlining to reduce that copying).
Thanks, that makes more sense!
My components are all small structs (coordinates, states, health values, etc) so I am happy to copy them around
Component[T] is something like
{
Metadata_fields...
Data T
}
Re:Generics can make your Go code slower
by nikki93 on Wednesday March 30, 2022 @04:20PM
(#30860297)
Sweet! I've been using it for the same. Example game project (did it for a game jam): https://github.com/nikki93/raylib-5k -- in this case the Go gets transpiled to C++ and runs as WebAssembly too. Readme includes a link to play the game in the browser. game.gx.go and behaviors.gx.go kind of show the ECS style.
It's worked with 60fps performance on a naive N^2 collision algo over about 4200 entities -- but also I tend to use a broadphase for collisions in actual games (there's an "externs" system to call to other C/C++ and I use Chipmunk's broadphase).
Sounds interesting, is it available somewhere?
I get to write one set of generic methods and data structures that operate over arbitrary "Component" structs, and I can allocate all my components of a particular type contiguously on the heap, then iterate over them with arbitrary, type-safe functions.
I can't fathom that doing this via a Component interface would be even as close as fast, because it would destroy cache performance by introducing a bunch of Interface tuples and pointer dereferencing for every single instance. Not to mention the type-unsafe code being yucky. Am I wrong?
FWIW I was able to update 2,000,000 components per (1/60s) frame per thread in a simple Game of Life prototype, which I am quite happy with. But I never bothered to evaluate if Interfaces would be as fast