The bottleneck here is network, network driver, network adapter. To CUDA it will not come even CUDA here is redundant.
And here is the server's network adapter
, which has its own iron on Board, not a job for the operating system driver, in contrast to cheap mass consumption counterparts - meaning.
The quality of the switch.
Where to dance nginx.org/ru/docs/events.html
Once you have solved the problem with the network then it makes sense to optimize on:
CUDA, the programming language, etc.But it all very much depends on what you are treating.
Maybe you have each package requires a 10 minute reflection on CUDA. Question is not clear.Real-time you requirements or a request queue and a response to poison an hour is also not clear from the question.
Constant load or variable, can implement a queue - from question is not clear.
Can you look at CUDA to load the data with such speed?In CUDA it is possible to load large chunks of data. But few pieces. But quickly.
About can I ship on CUDA and to answer to many small pieces - is doubtful.
Did anyone signals with a frequency of 30 MHz
MHz? 30 millihertz
? that is, of 0.03 times per second??? what's the problem?
when talking about MHz - megahertz - and write correctly.
To start your educational program:SI prefix
That to UDP so also the size of the package
depends on the speed.