Do you guys still buy the beefiest Intel Xeons in order to fit your main application + OS entirely within the L3 cache? There was a CppCon talk from a HFT dev about this 10 years ago.
Not 10 years, merely 6. I was mistaken and thought it was posted in 2015.