Improve scheduling for thomaswue

[thomaswue]

Check List:

• Tests pass (./test.sh <username> shows no differences between expected and actual outputs)
• All formatting changes by the build are committed
• Your launch script is named calculate_average_<username>.sh (make sure to match casing of your GH user name) and is executable
• Output matches that of calculate_average_baseline.sh

• Execution time: 0.62
• Execution time of reference implementation: 120.37s

Already filing enhancement requests for the GraalVM compiler scheduler. Apparently this innocent looking change gains 6%+ ;-).

[gunnarmorling]

It doesn't seem to move the needle on the eval machine:

Benchmark 1: timeout -v 300 ./calculate_average_thomaswue.sh 2>&1
  Time (mean ± σ):      2.722 s ±  0.008 s    [User: 18.719 s, System: 0.701 s]
  Range (min … max):    2.716 s …  2.741 s    10 runs

Summary
  thomaswue: trimmed mean 2.72014917887, raw times 2.71617197462,2.7215708666199996,2.71872755162,2.7185243566199997,2.7305528786199997,2.7414847446199997,2.72071959562,2.71618575162,2.7173498986199998,2.7175625316199996

Leaderboard

| # | Result (m:s.ms) | Implementation     | JDK | Submitter     | Notes     |
|---|-----------------|--------------------|-----|---------------|-----------|
|   | 00:02.720 | [link](https://github.com/gunnarmorling/1brc/blob/main/src/main/java/dev/morling/onebrc/CalculateAverage_thomaswue.java)| 21.0.1-graal | [Thomas Wuerthinger](https://github.com/thomaswue) | GraalVM native binary |

[gunnarmorling]

Already filing enhancement requests for the GraalVM compiler scheduler.

This is great! Also someone mentioning JDK tickets somewhere. So great to see these improvements as a fallout of this!

[thomaswue]

OK, thanks for evaluating! Looks like this is starting to get quite machine-specific now.

[gunnarmorling]

Yepp, absolutely. It might be as far as we get for now, which is waaay beyond whatever I had been expecting before. Let's see whether someone manages to pull out a surprise bunny out of their hat between now and Jan 31 ;)

[thomaswue]

@gunnarmorling No stress evaluating this, but now I fixed a problem in the hash calculation (which even led to a collision for the test data set) and improved the performance for the new 10k cities test substantially by being a bit better in case of actual collisions. The performance gain on my machine was another 3% over the previous version for the original test data and I have now runs slightly below 0.60s there.

[gunnarmorling]

00:02.552 now on the official data set. I.e. a teenie-tiny bit ahead @merykitty's recorded unsafe entry. So close in fact that I can't identify a clear ordering with the current way of measuring. I'm therefore putting both with #1 into the leaderboard for now. I don't think the order matters an awful lot, as the learning experience is the most important outcome here.

| # | Result (m:s.ms) | Implementation     | JDK | Submitter     | Notes     |
|---|-----------------|--------------------|-----|---------------|-----------|
| 1 | 00:02.552 | [link](https://github.com/gunnarmorling/1brc/blob/main/src/main/java/dev/morling/onebrc/CalculateAverage_thomaswue.java)| 21.0.1-graal | [Thomas Wuerthinger](https://github.com/thomaswue), [Quan Anh Mai](https://github.com/merykitty), [Alfonso² Peterssen](https://github.com/mukel) | GraalVM native binary |

[thomaswue]

Yes, agreed, thank you!