Author: Sune Fischer
Date: 11:03:30 03/10/05
Go up one level in this thread
On March 10, 2005 at 11:25:03, Anthony Cozzie wrote: >>Generally speaking communication costs $. >> >>Imagine if your chess program created a socket (e.g. via GPRS) to a >>chess server, and used UCI to relay moves. >>You'd be ruined fast. > >Lets do the math here. Each move is 5 bytes (4 coord + space). If we assume >the average number of moves is 50 (generous) than UCI involves an overhead of >50*2 (ponder+normal) * 5 = 500 bytes / move. On the same computer, this >probably 1 us. On a broadband connection, it is something like 2.5 ms. So, >even giving you the benefit of the doubt on every point, it still doesn't >matter. > >anthony I was thinking more along the lines of general communication, say mobile devices. I bet it won't be too long before we can play on ICC from our cell phones. It might also be interesting to make a two-component engine. Write a GUI shell to run on a phone or Palm, then let it connect to the main engine back home on your PC. It shouldn't be too hard to listen to some port and handle the socket stuff. By remote controlling a real PC you can of course walk around with a conventional strength chess program! Even the XBoard protocol would be too expensive to run I think, every byte you save is money in your pocket. Regarding regular PC's I don't know the actual latency of transmitting 500 bytes, I suspect it might be more than 1 us. You also need to parse and make all the moves. I won't claim it's the deciding factor, just a small minus on the graceful side. -S.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.