cache size affect the data rate of queues

Hi Community !!!

I’ve try to detect the data rate of neutrino message queues.

And I have a problem in data rate, there is strong decrease
in data rate between the blocksize of 28500 byte and 29000 byte.

So the one possibility is, as some community members mentioned,
the cache size.

please explain me exactly how cache size of the processor is able to
affect the data rate of the message queue I dont understand this fact.

Thank you

Read this: … 299_3.html

Thank you mario

very interesting article!!!

But I am using a P4 1800 L2 Cache: 512 KB => 524288 byte
and the problem Blocksize is 28500-29000 byte
so the data area fit into the processor’s cache

and other Tests with message passing, shared-memory, file don’t have
this problem.

so whats special by queues this problem appears by 28000 byte ?

if I increase the time for measurement from 5 seconds in neutrino to 30 seconds in QNX4
the same effect happens in QNX4 too whereas 5 seconds works without the effect.

Thank you !!!

is the solution because of
no “send-blocked” “receive-blocked” “reply-blocked” like in
message passing and no semaphor using for synchronisation
is data accumulating over the 512 KByte
because it is a message queue and data will be buffered?

Thank you !!!