Gabriel Baraldi
Gabriel Baraldi
So I ran your code (1.10 + https://github.com/JuliaLang/julia/pull/50682) and I see a wall time regression, but not necessarily a GC, because max GC time might not be the best way...
The GC might decide to coalesce the garbage collections to bring down overall time while having potentially longer pauses.
So there is a bug on 1.10 beta that should be fixed on the next beta, but in any case is there a way for you to share an open...
Oddly enough I think this might not be a GC bug at all, this is where perf says we are spending time. Which makes me think this is a subtyping...
@vtjnash it hangs while calling ```c ijl_matching_methods (types=0x7f6a96ddbb90, mt=0x7f6a8c1c78a0 , lim=-1, include_ambiguous=0, world=31777, min_valid=0x7f69267e4ea0, max_valid=0x7f69267e4eb0, ambig=0x0) ``` `types` being `Tuple{typeof(Base.:(*)), Any, Any}` Any clue?
Could you open a separate issue, and I wonder if you can make an example that runs a bit quicker on 1.9?
I believe `return_type` does https://github.com/JuliaLang/julia/blob/a889575cc168d5efca211766a6ddad7b3069594a/base/compiler/typeinfer.jl#L1125
I hid those comments because they actually belong to https://github.com/JuliaLang/julia/issues/50859
I wonder if we can do something like this for the chunks, because currently only one thread can take pieces of a chunk at a time, and ideally we should...
I guess it wouldn't be pointer tagging for the chunks, it would probably need to be a heap allocated object, or at least something that's reachable by all threads, but...