6 ms·

Linear-Log Bucketing: Fast, Versatile, Simple

• pacaro 11y ago
  I wish that I knew the name, but this reminds me of the banknote sequence 1 2 5 10 20 50 100 ...
  • rw 11y ago
    Hyperinflation sequence for banknotes: http://oeis.org/A051109 http://oeis.org/A051109
• bluejekyll 11y ago
  I believe memcached uses a similar algorithm for storing objects in its heap space. It's a nice way to strike a balance between waste and speed.
• jibsen 11y ago
  Is it just me, or are the bin sizes listed from jemalloc (assuming 150 should be 160, and 182 be 192) not dividing each power-of-2 range linearly? 16, 32, 48, 64, 80, 96, 128, 160, 192, 256, 320, 384, … It looks like, from 64 and up, each power-of-2 range is divided into two quarter ranges and one half range.
  • pkhuong 11y ago
    That's just bad mental arithmetic on my part. Fixed!
    • hyperpape 11y ago
      I think you now subbed 102 for 192 "128, 160, 102, 224". Sorry to nitpick, but I stared at that for longer than I'd really like to admit before thinking it was a typo.
      • pkhuong 11y ago
        Fixed, thank you!
• to3m 11y ago
  This sounds like what the TLSF allocator does: http://www.gii.upv.es/tlsf/main/docs http://www.gii.upv.es/tlsf/main/docs