search

rorygraves / scalac_perf

The Scala programming language
http://www.scala-lang.org/
16 stars 3 forks source link

Initial structure and parallelization of parser #66

Open pkukielka opened 6 years ago

pkukielka commented 6 years ago

Results from perf_tester:

Parser phase comparison:

19:30:25 after 0 best 10%, phase parser
19:30:25                  Run Name  Cycle   samples            Wall time (ms)          All Wall time (ms)                     CPU(ms)              Idle time (ms)              Allocated(MBs)
19:30:25                  baseline     1       5     1,075.80 [-0.45% +0.39%]    1,075.80 [-0.45% +0.39%]    1,075.80 [-0.45% +0.39%]          0.00 [ NaN%  NaN%]      450.69 [-4.45% +2.69%]
19:30:25                  baseline     2       5       990.40 [-0.55% +0.67%]      990.40 [-0.55% +0.67%]      990.20 [-0.53% +0.59%]          0.00 [ NaN%  NaN%]      451.92 [-4.95% +3.39%]

18:43:04 after 0 best 10%, phase parser
18:43:04                  Run Name  Cycle   samples            Wall time (ms)          All Wall time (ms)                     CPU(ms)              Idle time (ms)              Allocated(MBs)
18:43:04                    parser     1       5       308.00 [-1.62% +0.97%]      308.00 [-1.62% +0.97%]       14.40 [-2.78% +4.17%]          0.00 [ NaN%  NaN%]        2.97 [-0.04% +0.03%]
18:43:04                    parser     2       5       280.60 [-1.64% +0.86%]      280.60 [-1.64% +0.86%]       14.20 [-8.45% +5.63%]          0.00 [ NaN%  NaN%]        2.96 [-0.02% +0.01%]

Total comparison:

19:30:25 after 0 best 10%
19:30:25                  Run Name  Cycle   samples            Wall time (ms)          All Wall time (ms)                     CPU(ms)              Idle time (ms)              Allocated(MBs)
19:30:25                  baseline     1       5    21,656.80 [-5.55% +1.83%]   21,656.80 [-5.55% +1.83%]   21,471.40 [-5.28% +2.00%]          0.00 [ NaN%  NaN%]    3,804.04 [-0.83% +0.50%]
19:30:25                  baseline     2       5    20,870.00 [-0.95% +1.10%]   20,870.00 [-0.95% +1.10%]   20,773.40 [-0.91% +1.13%]          0.00 [ NaN%  NaN%]    3,808.93 [-0.95% +0.52%]

18:43:04 after 0 best 10%
18:43:04                  Run Name  Cycle   samples            Wall time (ms)          All Wall time (ms)                     CPU(ms)              Idle time (ms)              Allocated(MBs)
18:43:04                    parser     1       5    22,126.00 [-2.44% +1.67%]   22,126.00 [-2.44% +1.67%]   21,640.20 [-2.21% +1.76%]          0.00 [ NaN%  NaN%]    3,421.77 [-0.29% +0.18%]
18:43:04                    parser     2       5    20,689.60 [-0.72% +0.59%]   20,689.60 [-0.72% +0.59%]   20,267.60 [-0.61% +0.72%]          0.00 [ NaN%  NaN%]    3,356.09 [-0.01% +0.01%]

This is with rangepos enabled.

mkeskells commented 6 years ago

Its looking pretty good

I added some helper classes to https://github.com/rorygraves/scalac_perf/commits/mike/2.13.x_parallel_base in commit 2 that should ease some of the code I hope

mkeskells commented 6 years ago

Looks very close 👍 just a few small comments. Lets discuss on Monday and squash this