[saagarjha]

Isn’t k-means memory bandwidth bound? What was the arithmetic intensity of the final code?

[shihab]

No. Assuming `k` is small enough, which in practice often is, the arithmetic intensity of this kernel is 25-90 Flops/Byte, way above the roofline knee of any modern CPU.

[NohatCoder]

I assume that the image would at least fit in L3.