gaujay

Hi! Thanks for your feedback. It's nice to know JOMT is useful to people. To clarify, in what scenario would you like to see settings saved? It seems that the...

Hello, For what I can see, except for steps 5 and 6, all the others configs should be restored on re-opening the same benchmark. Are you up-to-date on the repository?...

Did you build the last JOMT version? (the one with the update to Qt6): https://github.com/gaujay/jomt/commits/master/ In my scenario: 1. Open a benchmark file in JOMT 2. Plot chart 3. Change...

A config file per benchmark could be an interesting addition. But in your case, fixing the missing config files per chart type would already address most of your needs. Check...

Thanks for taking time to share feedback. I appreciate it :) Hope you like the new "baseline" option I added few weeks ago

Like Abseil original design, Hashbrown uses tombstones and an anti-drift mechanism. The idea is to avoid performance degradation by forcing a grow + rehash when too much have accumulated. In...

I updated my repo to add benchmark results for LLVM using the same custom hash (murmur3/FNV-1a) for every map. This should give a better view about what could be gained...

I made some new benchmarks to monitor lookup hit/miss performances depending on load factor. Here are the typical results on my machine: ``` // FoldHashMap, RandomKeys test lookup_size_14500 ... bench:...

Yes, not sure which one could benefit Hashbrown without extensive benchmarks. I already noticed some performance difference between random keys and serial ones for example (due to hash quality and...

Updated some preliminary benchmarks in a branch for an 8-bits Tag implementation (using SSE2). Gains might not always be clear, but it might shows some improvement room for iteration at...