[zeromq-dev] Mb vs. MB in tests & on the Wiki

[Holger Hoffstätte]

In case anybody cares..

Holger Hoffstätte wrote:
> FWIW lmbench says this rusty tin pushes ~1.1GB (not bit) between processes
> with AF_UNIX but of course that is without zmq overhead. Still, shouldn't
> throughput over loopback be higher than ~60-70 MB even with only one core?

Now on the dual-core Thinkpad T60 things look markedly different. Here
lmbench pushes >2GB/sec and zmq results are much better too, obviously
because the two cores can push/pull in tandem. For single-threaded
producer/consumer I get ~200-350 MB/sec throughput depending on message
size, and that goes up to ~495 MB with two threads each and messages of
128k. Honor restored!

> Then again, latency is much more interesting to me anyway. I'll run some
> tests on a dual-core machine with the -rt kernel later and can report back
> if it yields anthing useful.

2.6.26-rt9 seems generally a good bit slower in throughput and has higher
average latency than stock 2.6.27 - ~55us vs. ~27-35us. I didn't measure
avgerage deviation from min/max but somehow don't think it would be much
better either. This is consistent with the results I got when tracing
RabbitMQ.

cheers
Holger