3 ms·

Unpacking (or splatting) can be excruciatingly slow in Julia if you have more than a few dozen elements in the iterable. That happens because it gets splatted

Unpacking (or splatting) can be excruciatingly slow in Julia if you have more than a few dozen elements in the iterable. That happens because it gets splatted into a tuple with N elements, which in Julia has a type parameter for every element - i.e. a generic type Tuple{N1, N2, N3 ... N} with as many type parameters as there are elements in the iterable. And the compiler has a hard time working with an instance of a type with e.g. 10,000 parameters.

dunefox 6y ago
That seems like a niche case. Surely flattening arrays is faster.
cbolton 6y ago
Was this improved in recent versions? I tried the following in Julia 1.6-rc1 and they run instantly: @time v = [(1:10_000)...]; -> 0.000639 seconds f(args...) = sum(args) @time f(v...); -> 0.001039 seconds