Marius Millea
Marius Millea
Fixes #10 #24 With this, e.g.: ```julia julia> using CUDA, AutoPreallocation julia> _,f = preallocate() do x = CUDA.zeros(10) 2 * x .+ 3 end; julia> CUDA.@allocated f() 0 ```
Here's a MWE: ```julia using AutoPreallocation N = 256^2 # preallocation doesn't work here struct Foo end Foo() = Array{Float64}(undef, N) _,preallocated_f = preallocate(Foo) @allocated preallocated_f() # 52440 # preallocation...
This seems erroneous since the output is actually real: ```julia using CUDA, Zygote, LinearAlgebra x = cu(rand(10)) y = complex(cu(rand(10))) Zygote.gradient(1) do A norm(@. A * x * y) end...
I'm not an expert and don't know the internals, but is there a reason PyCall can't do whatever PythonCall does that lets the user choose any Python executable at runtime?...
It seems at least on Linux, setting the environment variable `PYTHONFAULTHANDLER=1` (enabling a Python-builtin package as of recent versions https://docs.python.org/3/library/faulthandler.html) causes segfaults with PyCall. The MWE I've been using is...
Here's my maybe naive way to fix #882. Trying to aquire the GIL from the finalizer led to deadlocks. This is instead basically option 3 from [here](https://docs.julialang.org/en/v1/manual/multi-threading/#Safe-use-of-Finalizers). Tests pass locally...
The `@profview` GUI shows an inline graphic and tooltip hover of how much time was spent on a given line relative to the other lines in the parent function. It...
The following seems unexpected: ```julia using FFTW, LinearAlgebra y = rand(4,4) p = plan_rfft(y) x = p * y x_orig = copy(x) ldiv!(y,p,x) x == x_orig # unexpectedly returns false...
What the title says. Once you `using CUDA`, Revise doesn't work for any subsequently loaded other packages. I made a small MWE with pinned package versions (CUDA v3.12.0, Revise v3.3.4)...
Random thing I unearthed w/ Soss v0.20.9 and GeneralizedGenerated v0.3.3, not sure if it belongs here or there but I only know how to trigger it from here: ```julia model...