Closed Manishearth closed 5 years ago
This does make Scope
one byte bigger. However, it's stored on the stack, so this doesn't really matter. In fact we could probably make this into a 4-element SmallVec
and still win.
The compiler probably is able to fast-path the two-element case, too.
Yep, definitely a nice improvement in the one test that's affected by the tracing (preferences):
Gnuplot not found, disabling plotting
resolve/"simple" time: [9.3731 us 9.4333 us 9.4966 us]
change: [+0.4569% +1.7164% +2.9955%] (p = 0.01 < 0.05)
Change within noise threshold.
Found 7 outliers among 100 measurements (7.00%)
2 (2.00%) high mild
5 (5.00%) high severe
resolve/"preferences" time: [103.70 us 104.21 us 104.72 us]
change: [-11.756% -11.077% -10.394%] (p = 0.00 < 0.05)
Performance has improved.
Found 1 outliers among 100 measurements (1.00%)
1 (1.00%) high mild
resolve/"menubar" time: [26.571 us 26.828 us 27.101 us]
change: [-2.2364% -1.3536% -0.4172%] (p = 0.00 < 0.05)
Change within noise threshold.
Found 3 outliers among 100 measurements (3.00%)
3 (3.00%) high mild
resolve/"unescape" time: [1.7138 us 1.7233 us 1.7356 us]
change: [-10.769% -9.5668% -8.2277%] (p = 0.00 < 0.05)
Performance has improved.
Found 6 outliers among 100 measurements (6.00%)
1 (1.00%) high mild
5 (5.00%) high severe
Gnuplot not found, disabling plotting
Thank you
I seem to see improvements in the benchmarks, but my results seem to swing wildly.
Worth checking yourself first.
r? @zbraniecki