This project collects some unscientific measurements of the performance of biber and bibtex on different platforms.
git clone https://github.com/thomwiggers/biber-benchmark.git
./benchmark.sh ~ git biber-benchmark ./benchmark.sh
/tmp/tmp.YJjf3lghB8 ~/git/biber-benchmark
Cloning into 'cryptobib'...
remote: Enumerating objects: 8, done.
remote: Counting objects: 100% (8/8), done.
remote: Compressing objects: 100% (8/8), done.
remote: Total 8 (delta 3), reused 1 (delta 0), pack-reused 0
Receiving objects: 100% (8/8), 4.74 MiB | 2.62 MiB/s, done.
Resolving deltas: 100% (3/3), done.
This is pdfTeX, Version 3.141592653-2.6-1.40.24 (TeX Live 2022/Arch Linux) (preloaded format=pdflatex)
restricted \write18 enabled.
entering extended mode
This is pdfTeX, Version 3.141592653-2.6-1.40.24 (TeX Live 2022/Arch Linux) (preloaded format=pdflatex)
restricted \write18 enabled.
entering extended mode
Biblatex version: 3.17
~/git/biber-benchmark
biber version: 2.17
biber execution time was 98 seconds
BibTeX 0.99d (TeX Live 2022/Arch Linux)
bibtex execution time was 1 seconds
Model name: Intel(R) Core(TM) i7-8550U CPU @ 1.80GHz
| Platform | Back-end | Processing time |
|---|---|---|
| XPS 13 9370 i7-8550U | Biber 2.17 | 85 seconds |
| XPS 13 9370 i7-8550U | Bibtex 0.99d | 2 seconds |
| Ryzen 3950X | Biber 2.17 | 61 seconds |
| Ryzen 3950X | Bibtex 0.99d | 1 second |
| Epyc 7401P | Biber 2.17 | 109 seconds |
| Epyc 7401P | Bibtex 0.99d | 2 seconds |
| Apple Mac Mini M1 | Biber 2.17 | 43 seconds |
| Apple Mac Mini M1 | Biber 2.18 | 43 seconds |
| Apple Mac Mini M1 | Bibtex 0.99d | 2 seconds |
| Apple M2 Pro | Biber 2.18 | 39 seconds |
| Apple M2 Pro | Bibtex 0.99d | 2 seconds |