Hello, folks! David, I could suggest you to try memcpy analogues from SIMD, SSE4 instruction set. They are significantly faster than standard memcpy.
On Thu, May 7, 2015 at 11:01 AM, Luca Deri <d...@ntop.org> wrote: > David > nDPI is reentrant so from this point of view you should be safe. Like Alfredo > said ZC might help ypu. In fact sometimes the memcpy is the most expensive > step in apps, so avoiding it is desirable > > Regards luca > >> On 07 May 2015, at 07:20, David Britt <d...@britt.com.au> wrote: >> >> I am looking to build an analysis tool based on nDPI and would like to share >> the load across multiple CPU cores. I notice that there is some indication >> of thread safeness in nDPI 1.5 but I am wondering whether anyone has built, >> or has ideas on building a multi threaded nDPI application. >> >> I suppose my main question is how to ensure each packet from libpcap reaches >> the correct nDPI thread for processing and analysis - I guess I need to >> have some kind of 5 tuple hashing mechanism to classify each packet into a >> flow and push to an appropriate queue that each nDPI thread can pick from? >> Is this something that PF_RING can do, or perhaps some other simple >> technique? >> >> Any help, guidance of examples would be very much appreciated if anyone has >> gone down this path and is willing to share some ideas. >> >> Dave Britt. >> _______________________________________________ >> Ntop-misc mailing list >> Ntop-misc@listgateway.unipi.it >> http://listgateway.unipi.it/mailman/listinfo/ntop-misc > > _______________________________________________ > Ntop-misc mailing list > Ntop-misc@listgateway.unipi.it > http://listgateway.unipi.it/mailman/listinfo/ntop-misc -- Sincerely yours, Pavel Odintsov _______________________________________________ Ntop-misc mailing list Ntop-misc@listgateway.unipi.it http://listgateway.unipi.it/mailman/listinfo/ntop-misc
