Hi!

Whether you plan to create Hardware NAT Accelerate for routers?

P.S. Sorry for my English.

Hi!

Whether you plan to create Hardware NAT Accelerate for routers?

P.S. Sorry for my English.

No, and yes they already do (Fast Path) and no/yes, but no.

Why would they want a non-versatile hardware NAT engine and restrict themselves while they have clearly chosen for a software based optimized multi-core design with the CCR series?

So no, it's not needed or desired. Hopefully that answer your question.

Yes and no. It is not needed, as far as the software implementation can meet the performance requirements of a specific application. In the case of massively multicore machines such as CCR, this requires the processing engine to be wholly multicore optimised. Correct me if I am wrong, but RouterOS is not yet 100% multicore. It seems the complete transition is planned for v7.

Στάλθηκε από το GT-I9100 μου χρησιμοποιώντας Tapatalk

Hello,

Correct me if I am wrong, but RouterOS is not yet 100% multicore. It seems the complete transition is planned for v7.

Surely it is not 100% multicore compatible, but if such companies like Microsoft, huge gaming companies, and others like Adobe are only using until today only 4 Cores from the Intel cpu´s
it will be fine to know that a smaller latvian company likes MikroTik is able to deliver it from 9
to 72 cores those things.

its mode in mentality of old-shool devlopers, grown-ups from C and C++ backgrounds/career.
not so much HR on market presently with skills of truly coherent, fluidly conculrent and self-parelezing langs, including Erlang foremost and them Haskell, Ada(even in aerospace and military sector).
three isn't poitn to talk about scaliabity in Reality/Universe(or perceptions of such) where exist things as "threads", "semaphores", "mutexes", "shedulers" and etc.
.http://www.erlang.org/download/erlang-book-part1.pdf
folks who started from oberon, pascal, modula and etc in college - usually more flexible and better adapt that.
same issues with ROS kernel - MT went alpha netowkrs and cavium way and implemented fast path on kernel-level, while Tilera had WASTLY better idea about. with preidcitble benefts, both in scalability and other areas.
http://www.tilera.com/sites/default/fil ... aper_0.pdf