On 10/30/2015 01:52 AM, Markus Armbruster wrote: >>>> I'm assuming you temporarily patched check-qjson to use larger constants >>>> when you hit your ~100K token testing? Because I am definitely seeing a >>>> lot of execution time spent on large_dict when running tests/check-qjson >>>> by hand, in relation to all the other tests of that file, but not >>>> minutes worth. Care to post the diff you played with? >>> >>> I tested on a slow machine. >> >> I guess it was all the malloc pressure on a low-memory system that would >> make it so much slower than what I'm seeing, if you stuck with the >> default gen_test_json(gstr, 10, 100). > > Something must have been wrong with this machine yesterday, because > today it runs the exact same test much, much faster. Computers...
Or maybe you had still reverted 65c0f1e9 to figure out where the regression was at asserting rather than gracefully failing as fixed by patch 2/4? Still, even if it is much faster, it is noticeably slow enough to wonder if we have some inefficiencies beyond the poor memory usage. > > I'll update the commit message. Thanks for your sanity check! > -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature