Recommender system benchmark

[lkuper]

From @ViralBShah :

https://github.com/JuliaLang/julia/issues/15874

Folks, this is another non-trivial benchmark. It is the ALS algorithm. We have some work to improve the multi-threading performance, but figured that the folks at Intel may find it interesting. Abhijith, from Julia Computing, has been doing this work. The issue describes where the code is and how to run it with the datasets.

Good example for ParallelAccelerator too.

This might be interesting to look into for HPAT, too.

[ehsantn]

Yes, this is interesting for both ParallelAccelerator and HPAT. We have already looked at ALS to some extent for HPAT but we need sparse matrix support first.

[ViralBShah]

Yes, sparse matrix support is a must.

[lkuper]

For background on what the RecSys.jl package is doing, here's the Julia Computing blog post on it. And, to understand that, the context from the ALS paper is helpful.