[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH 1/7] posix-timers: tidy up clock interfaces and consolidate dispatch logic
George Anzinger wrote:
As I understand it modern machines, the indirect call does really bad
things to the pipeline.
Unless the call happens often, really often, you'll not be able to
measure differences. Indirect jump obviously cause pipeline stalls
which often, but not always, can be avoided with direct calls. But we
are not talking here about huge delays. And the delays are equivalent
in cost to a mispredicted branch. If you have a couple of ifs in a row
you are likely to incur problems with the branch prediction. I would
just use the indirect jumps but if you want, do some timing and see
whether you can filter the difference out of the noise.
â Ulrich Drepper â Red Hat, Inc. â 444 Castro St â Mountain View, CA â
Description: OpenPGP digital signature