[zeromq-dev] LMAX/Distruptor code project

[Daniel Cegiełka]

Here is an interesting discussion:

http://groups.google.com/group/lmax-disruptor/browse_thread/thread/4a47a0a9b5837ca2/4ca5e4fb17593f75

--- CUT ---

Hello,

I've made a port called disruptor--, it's quite different than Dan's
disruptor-cpp.

-no dependency on boost, only C+11 features
-no compilation needed, 100% templated
-autotools integrated for tests/benchmark
-almost up to date with latest LMAX's disruptor 2.6
-using Google C++ style guide

This is an early port, probably not production ready. Next phase is
testing and benchmarks.

Now for the number crunchers, benchmark are run on a (old) Core2 Duo
E8400.

For comparison, I've run Martin's inter thread latency (http://
mechanical-sympathy.blogspot.com/2011/08/inter-thread-latency.html) 50
times getting a mean of 32,451,655 ops/sec with standard dev. of
55,616. The only test supported ATM is 1P- 1EP (1 publisher, 1 batch
event processor), getting a mean of 29,785,805 ops/sec, standard dev.
of 1,475,971, and maximum of 32,211,178 (pretty close to the inter-
thread ops/sec).

I have also run the benchmark on a core i7 930, but there's a really
high variance (12m-60m ops/sec). I'm still trying to find out why
there is such a high variance before posting numbers.

Francois


--- CUT ---




W dniu 6 października 2011 13:10 użytkownik Daniel Cegiełka <
daniel.cegielka at gmail.com> napisał:

>
>>     It may be worth speaking to the guy then. Maybe he had something
>>>    like replacing yqueue with disruptor in the mind himself.
>>>
>>>
>>> I think this is a very good idea. Disruptor is a fresh look at
>>> concurrency.
>>>
>>
>> Great, are you going to contact him and find out?
>>
>>
> I think it would be better if ask someone who knows much better than I
> zeromq structure.
>
> daniel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.zeromq.org/pipermail/zeromq-dev/attachments/20111006/c01eff46/attachment.htm>