STINNER Victor added the comment: > I'm somewhat surprised at the 2-3x numbers you're seeing, as I was > consistently getting 4-5x in the Linux tests I did. But it does depend quite > a bit on what file system you're running, what hardware, whether you're > running in a VM, etc. Still, 2-3x faster is a good speedup!
I don't think that hardware matters. As I wrote, I expect the whole /usr/share tree to fit in memory. It's sounds more like optimizations in the Linux kernel. I ran benchmarks on Fedora 20 with the Linux kernel 3.14. > Anyway, where to from here? Are we agreed given the numbers that -- > especially on Linux -- it makes good performance sense to use an all-C > approach? We didn't try yet to call readdir() multiple times in the C iterator and use a small cache (ex: between 10 and 1000 items, I don't know which size is the best yet) to also limit the number of readdir() calls. The cache would be an array of dirent on Linux. scandir_helper() can return an array of items instead of a single item for example. I can try to implement it if you want. ---------- _______________________________________ Python tracker <rep...@bugs.python.org> <http://bugs.python.org/issue22524> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com