dtoa-benchmark

Benchmark results

Each card below summarises one dtoa-benchmark run. The mini chart shows the top 5 fastest methods for that run; click through for the full report with per-digit timings, interactive charts, and a copyable table.

epyc-genoa-processor2026-04-30
Latestlinuxgcc11.58e6015f
zmij
14.15 ns
xjb64
15.21 ns
yy
31.78 ns
dragonbox
40.93 ns
uscale
55.71 ns
12 methodsView report →
apple-m1-pro2026-04-30
macosclang21.045ff4d4
zmij
6.45 ns
xjb64
6.99 ns
yy
24.63 ns
dragonbox
28.95 ns
fmt
36.84 ns
12 methodsView report →
apple-m1-pro2026-04-30
macosclang21.0359383b
zmij
6.51 ns
xjb64
6.92 ns
yy
13.38 ns
dragonbox
20.92 ns
fmt
22.13 ns
12 methodsView report →
ryzen-threadripper-3960x2026-02-10
linuxgcc13.3a10220f
zmij
16.24 ns
xjb64
19.62 ns
yy
21.40 ns
dragonbox
24.76 ns
uscale
28.96 ns
12 methodsView report →
apple-m1-pro2026-01-21
macosclang17.003adc84
zmij
8.66 ns
xjb64
10.54 ns
yy
14.01 ns
dragonbox
20.53 ns
fmt
22.22 ns
11 methodsView report →
epyc-milan-processor2026-01-21
linuxgcc11.569644a9
zmij
19.19 ns
xjb64
19.27 ns
yy
26.96 ns
dragonbox
30.90 ns
fmt
40.10 ns
11 methodsView report →
corei7920@2.672014-08-08
win32vc20139a4863a
milo
288.10 ns
grisu2
385.39 ns
doubleconv
412.39 ns
fpconv
470.22 ns
sprintf
1,338.50 ns
8 methodsView report →