Removed the custom allocator for PODRedBlackTree
Benchmark results (using PerformanceTests/Layout/float*): Baseline: floats_100_100.html: mean: 91.1718554492836 ms; stdev: 1.3918589890850632 ms floats_100_100_nested.html: mean: 100.49385774946131 ms; median: 100.05755349629908 ms floats_20_100.html: mean: 277.7535506003914 ms; median: 277.76054949936224 ms floats_20_100_nested.html: mean: 494.03283490028116 ms; stdev: 3.6272360800676324 ms floats_2_100.html: mean: 196.86348635004833 ms; stdev: 4.035362228138497 ms floats_2_100_nested.html: mean: 573.9511935502378 ms; stdev: 2.9780811856821683 ms floats_50_100.html: mean: 182.97691660009147 ms; stdev: 2.5367743906601987 ms floats_50_100_nested.html: mean: 229.38113970049017 ms; stdev: 2.325508081507197 ms Patch: ~/OpenSource/PerformanceTests/Layout> cat results-patch.txt floats_100_100.html: mean: 88.71689805018835 ms; stdev: 1.363018374233766 ms floats_100_100_nested.html: mean: 98.24441215023398 ms; stdev: 2.7629447157770586 ms floats_20_100.html: mean: 279.18915980080783 ms; stdev: 5.142024415145014 ms floats_20_100_nested.html: mean: 495.1849814511661 ms; stdev: 4.086517310884674 ms floats_2_100.html: mean: 199.62681959987094 ms; stdev: 3.2710243141682187 ms floats_2_100_nested.html: mean: 578.9757075501255 ms; stdev: 3.644157356812131 ms floats_50_100.html: mean: 176.08423344972834 ms; stdev: 2.033621745702075 ms floats_50_100_nested.html: mean: 226.00061699959042 ms; stdev: 2.011903990152965 ms Summary: floats_100_100.html: 2.8% faster floats_100_100_nested.html: 2.3% faster floats_20_100.html: 0.5% slower floats_20_100_nested.html: 0.2% slower floats_2_100.html: 1.4% slower floats_2_100_nested.html: 0.9% slower floats_50_100.html: 3.9% faster floats_50_100_nested.html: 1.5% faster ***geometric mean***: 0.9% faster The goal here is just to eliminate unnecessary complexity without regression. A small speedup is a nice bonus.
Created attachment 242003 [details] Patch
Committed r176431: <http://trac.webkit.org/changeset/176431>