Am 24.03.2024 um 11:50 schrieb Jonas Hahnfeld:
Just out of interest, do we tweak the JIT_THRESHOLD somewhere or do we
use the default value?
I don't think we change it from LilyPond, so likely using the default
value.
I did a test with the opposite extremes.
Again, 10 subsequent runs with the score I tested before.
Note that the times are not quite comparable with my last findings,
because my laptop
ran on a different energy saving state.

GUILE_JIT_THRESHOLD=0 (compile each function at first sight)
real    4m8.780s
user    0m0.075s
sys     0m0.167s

GUILE_JIT_THRESHOLD=-1 (disable JIT compilation)
real    4m4.925s
user    0m0.030s
sys     0m0.136s

So either the env variable is not working as intended, or there is only
a very small effect.
At least setting `GUILE_JIT_LOG` does work, but one would have to read
the source
to understand the output.

Michael
          • ... Kenneth Wolcott
            • ... Jonas Hahnfeld via Discussions on LilyPond development
    • Re: LilyPond... Karlin High
      • Re: Lily... Jonas Hahnfeld via Discussions on LilyPond development
        • Re: ... Karlin High
          • ... Jonas Hahnfeld via Discussions on LilyPond development
    • Re: LilyPond... Jean Abou Samra
      • Re: Lily... Jonas Hahnfeld via Discussions on LilyPond development
    • Re: LilyPond... Michael Käppler
      • Re: Lily... Jonas Hahnfeld via Discussions on LilyPond development
        • Re: ... Michael Käppler via Discussions on LilyPond development
          • ... Mathieu Demange via Discussions on LilyPond development
  • Re: LilyPond 2.25... Aaron Hill

Reply via email to