On that subject are there straightforward ways to tweak inproc latency in
the ZMQ configuration, besides msg and socket options?
On Thu, Apr 18, 2013 at 8:46 AM, Martin Hurton wrote:
> You have some control over memory allocation when transporting message
> using ypipes.
> See message_pipe_gran
You have some control over memory allocation when transporting message
using ypipes.
See message_pipe_granularity parameter in config.hpp.
- Martin
On Thu, Apr 18, 2013 at 5:26 PM, Timothee Besset wrote:
> Isn't the memory allocation part of those inproc exchanges blowing away any
> speed advant
I'm using inproc and doing delete after my recv calls. My code is stable and
not crashing, so that stuff must be allocated from somewhere?
I understand the zero copy stuff is on the sender side only, I've been meaning
to ask about it in the inproc context?
TTimo
Sent from my iPad
On Apr 18, 2
On Thu, Apr 18, 2013 at 5:26 PM, Timothee Besset wrote:
> Isn't the memory allocation part of those inproc exchanges blowing away any
> speed advantages of the lockless pipes in the first place anyway?
Inproc doesn't copy data, so it can all happen with no memory allocation.
-Pieter
___
Isn't the memory allocation part of those inproc exchanges blowing away any
speed advantages of the lockless pipes in the first place anyway?
(I haven't looked at the innards of the implementation)
TTimo
On Thu, Apr 18, 2013 at 10:23 AM, Martin Hurton wrote:
> > My main question at the moment
> My main question at the moment is about the current implementation of INPROC
> transport.
> Digging it seems like the actual implementation is the class pipe_t, am I
> right?
Yes, producer side writes the message to the pipe and consumer reads
it from it. The pipe is implemented as lockfree on s
Hello Davide,
I'm not certain about the implementation of the inproc in ZMQ but I'm
pretty certain it is using pipes. I'm interested in working on this issue
since I've an MPMC queue implemented using ZMQ inproc and ZMQ_ROUTER and
latency is critical of course. If we can improve latency in this a
Hi,
first of all, congratulation for the amazing ZeroMQ library.
I would like to use ZeroMQ of real time operative systems, In particular
Linux (RT-preempt and later Xenomai) and QNX.
To do this I am planning to do some changes to the core of ZMQ, in
particular:
1) Use O(1) memory allocation.
2)