Bouncner

But there are no benchmarks for this change, right? I only see the benchmark for all three changes together.

Why did you use a scale factor og 10? The working set almost fits into the caches.

Closing this PR. This PR has nicely shown that we need to address several NUMA issues in Hyrise. Before these optimizations are done, it is hard for this PR to...

Continue later.

Reopening as scheduler issues are hopefully fixed.

~Taking a look at performance. The current state cannot be merged.~

Benchmarks after changes to NULL writing. ### AMD **System** cx28 - click to expand | property | value | | -- | -- | | Hostname | cx28 | |...

Trying to address regressions in TPC-DS and TPC-C. ~Benchmarks are running.~

~Running benchmarks against new merged #2611 right now.~ Benchmarks done. But since it is a PR to improve multi-threading, I will probably wait for the merge of #2630 which changes...

> Will take a few more days, still benchmarking #2630 and it seems to be a bit flaky (probably due to the fewer runs). Does it help if I run...