May I ask you if you could re-run the script with the changes that @xjakub suggested? I created a file with those changes at github.com/realFlowControl/1brc
I am very curious about the time you get on that machine
Interestingly it doesn't seem to really scale past 24 threads. Between 16 and 24 it gets faster the more threads I chuck at it. Over 24 and it starts to slow. nice/ionice make no noticable difference
For further actions, you may consider blocking this person and/or reporting abuse
We're a place where coders share, stay up-to-date and grow their careers.
I got you. I have the exact same AX161 from Hetzner. The final version of his code executes in 15.4 seconds (with JIT & parallel).
Whoop whoop, that's awesome
May I ask you if you could re-run the script with the changes that @xjakub suggested? I created a file with those changes at github.com/realFlowControl/1brc
I am very curious about the time you get on that machine
With pleasure.
Ran it five times as my server is a bit busier today: 13.4/12.8/12.9/12.6/12.5
And re-ran the old version five times, for completeness: 16.3/16.0/15.9/15.9/15.9
Interestingly it doesn't seem to really scale past 24 threads. Between 16 and 24 it gets faster the more threads I chuck at it. Over 24 and it starts to slow. nice/ionice make no noticable difference