back
2 comments
I was slightly disappointed by this article. Rather than being directly about the speed of loading a csv which is what I was expecting, it more like comparing the speed of execution of column oriented math on streamed reads, vs pandas.

The article seemed thorough enough just not what I was expecting. I also feel like any comparison like this should also look at memory use.

Anyways, there should be no surprises here, pandas is great at column wise math.

I expected a lemire style full SIMD implementation :(