in another thread, tnt/beyond force wrote:
Have you tested how much power does it take to receive steady 50 packets per second and how big the latency is? (The latter would require two C64s with RR-Nets to tell the real performance so it might be harder for single person to test. Unless you own two RR-Nets, of course.)
Hi Jorma!
I have finally done this test, using Six/Style's netlib64. I find it easier to use than IP65. Here are the results:
You'll have to count text lines and multiply by 8 to get specific raster counts.
1) The grey line at the top is the time to detect whether a packet has arrived or not. Only 1-2 raster lines!
2) Blue is a SID playback (WOVIOT by Pegasus/RPG if anyone cares)
3) Green is the time to properly receive and parse the UDP packet. Note that before (at the gray line), it will respond to *any* packet, regardless of type, protocol, or destination. This is a 50 byte UDP packet, and the time to parse it does go up linearly as the packet grows.
4) Red is the time to prepare and send a 10-byte reply. It's a lot longer than I thought, but still less than a frame. Also grows based on packet size, though I haven't properly characterized it.
I think latency 'on the wire' on a local LAN segment is negligible compared to these.