[MobiusHorizons]

This is really cool work! I am surprised to see this level of tuning without using cache profiling or other performance counters to identify the bottleneck and quantify the improvement.

[remram]

They probably did.

[reconditerose]

Haha, we definitely used profiling tools. :) Perf is great and definitely the first tool to use. I know vtune has been used as well.