Closed henryso closed 7 years ago
My variability is quite a bit higher than yours (on the order of several minutes), but based on the limited number of tests I've run (and without any attempt to control what else might be going on while the tests are running), it appears that using the cache manager actually makes things slower (runs of 96m, 99m, 94m, and 91m with v. 91m, 90m, 91m, and 93m without). It also appears to be more expensive in terms of CPU time (~81m with vs. ~78m without).
Based on my inspection of your code, I'm reasonably certain you're running it right (or at least according to the instructions in the repository). If I get ambitious I might try consulting with @kalrish about what we might or might not be doing wrong, but for now I'm going to say that this appears to be a dead end in terms of optimization of the test script.
Thanks for giving it a try.
I think percentage-wise, it's about the same variability.
Are you saying you have a nearly 60 fold speed improvement over what I get? That makes me wish I could invest in a better system (there is that whole communal property thing though...).
My tests run in just under 4 minutes.
Is that for the whole test repository? Dang my system is much slower than I thought. But then, it is a 2009 system. That's 7 years old at this point (and may not even be upgradable to Sierra, based on what I've read).
Maybe it's time to talk to the Abbot about an upgrade...
Yes, that's all tests except luatex-ja and the long test. My computer will be two years old in January.
@rpspringuel If it's OK with you, lets close this as a failed experiment.
Agreed.
Fixes #155.
@rpspringuel Please try if this helps things on your computer. You will need to clone https://github.com/kalrish/luatex-lua-module-cache-manager somewhere and then run ...
... along with whatever other arguments you need.
On my computer, using the cache manager has zero effect. Runs with and without the cache manager are within 3 seconds of each other, with either beating the other about 50% on any given run.
I may be running it incorrectly, but the cache file is created and updated during the run.