Andy Ferris

Right! This is precisely why I’m interested in dictionaries. A dictionary is a container where lookup from key to value is fast. What you wrote is basically a multidimensional sort-based...

> After playing with this for ten minutes, I'm afraid it drives me completely nuts though! Yes I agree that it definitely needs improving significantly! > For my typical use...

Right... git and I aren't really friends at the moment. I _did_ try to rebase the first commit _on top_ of your changes, but I must have done something backwards.

It is an interesting question - the syntax provided by *TensorOperations* is very convenient, even if the array is small (and there's also block matrix structures).

What's the error for StaticArrays here? Is it `permutedims`?

All the code in *StaticArrays.jl* is manually unrolled by generated functions, which tend to also generate linear (not Cartesian) indices statically (all of this lets LLVM use SIMD). I generally...

Multithreaded - cool! Did you ever consider contributing your faster code to Base, and putting it in `permutedims!(::StridedArray, perm)` in base/multidimensional.jl? Obviously StaticArrays is the complete opposite end of the...

Here I use `mean` instead of the (default) median: ``` ===================================================================== Matrices of size 4×4 and eltype Float64 ===================================================================== SMatrix: m3 = m1 * m2 takes 87.07 ns, 144.00 bytes...

By resolves the issue, do you mean reports more allocations and longer run-times? EDIT: Yes, I see that in the PR. Looks great.

So #28 doesn't seem to affect my results in any meaningful way. I'm going to have a deeper look sometime but I can't right now. The code is at https://github.com/JuliaArrays/StaticArrays.jl/blob/use-size/perf/benchmark3.jl...