Informal test the "video" Transmission program written with DDS

In fact, it's a bit boring. I want to test how many video subscribers can fill up the network bandwidth.

Let's talk about the test environment:

M LAN

DDS inforepo runs on a notebook with a 54m wireless network card.

The video publishing end runs on a machine with a m Nic

Multiple external video subscribers run on four M NICs desktops.

Windows XP and Windows 2003

 

When the subscription end is crazy, more than 120 instances are started, and the CPU usage of the released machine Nic exceeds 90%.

ProgramNote:

Video Publishing end: capture frames through dshow, 15 frames per second, 320*240 size, compressed to about 5 k/frame in JPG format. Publish to subscriber through DDS.

When the subscriber receives data, the jpg image is displayed on the interface.

The default QoS policy is used by the publisher and subscriber.

When the percentage of NICs exceeds 90% (P: S> +), the system is unstable. First, the USB camera strikes and then the publishing end exits directly.

 

It is a publisher and two subscribers.

During the test, we found that the publish side (the upper-layer small screen) USB camera was sluggish and there was no significant delay in network transmission.

The test is only fun and no other results are recorded.

If you are interested in this, please contact us via emailSource code.