|
Title: 1k player FPS Tech demo Post by: Mrbloodworth on March 11, 2011, 05:35:13 AM Quote Pikkotech, a networking technology firm, is aiming for what it says is the world record for the number of players in a single multiplayer battle: 1000. Pikkotekk consulted with CCP, the noted developer of EVE Online and a pioneer in the field of single-shard massively multiplayer technology, and has created a server backend that works in a manner similar to the tech used by overlapping cell phone towers. The firm has thrown together a tanks vs. robots video demonstration complete with a humorous disclaimer. "We were so excited about our revolutionary network technology that we wanted a game that had it all [...] But our CEO said that Pikkotekk is a company that only does networking for games. So here you go: A crappy looking 1000-player shooter that you can play in the browser." Link to Preview video. (http://pikkotekk.com/worldrecord/) Title: Re: 1k player FPS Tech demo Post by: Sheepherder on March 11, 2011, 09:02:14 PM You still have to fit the data for a thousand entities through the small pipe that is a consumer grade connection.
Title: Re: 1k player FPS Tech demo Post by: DayDream on March 12, 2011, 02:34:32 AM Yeah, I agree. I think MMOs aren't going to make any substantial changes in gameplay design until the personal connection infrastructure improves, purely because of technical limitations. Improvements a la EQ>WoW, sure, but only along that spectrum. Plenty of situations in the current design style already run into that, raiding etc. At least in the western hemisphere, to my limited knowledge.
Title: Re: 1k player FPS Tech demo Post by: Lantyssa on March 12, 2011, 05:42:09 AM You still have to fit the data for a thousand entities through the small pipe that is a consumer grade connection. I think that's the point of their entire tech demo. If it works then they can market themselves as a tech solution.Title: Re: 1k player FPS Tech demo Post by: Sheepherder on March 12, 2011, 06:54:30 AM It just seems to be dynamic load balancing in a nice tidy package. They have some .pdf's from the main page if you're curious.
Numbers. (http://pikkotekk.com/tanks-vs-robots-in-numbers.pdf) By their numbers a snapshot of the server state is 3000 bytes large with 125 entities. Ergo, 24 bytes per entity, and they claim there's junk data for each entity. Which means that they're using single precision floats, three axis for position, and (likely) one axis for rotation (the other two being extrapolated at the client) with two junk bytes. Given that internet spaceships would take up all 24 bytes just with position data, assuming they didn't go for a double-precision float to match their new seamless world, that test is far less generous than one would be lead to believe. Title: Re: 1k player FPS Tech demo Post by: Mrbloodworth on March 12, 2011, 07:55:26 AM You still have to fit the data for a thousand entities through the small pipe that is a consumer grade connection. I think that's the point of their entire tech demo. If it works then they can market themselves as a tech solution.They already are, but they seem to be reaching for a 1k symmetric network solution. Title: Re: 1k player FPS Tech demo Post by: TripleDES on March 12, 2011, 09:11:58 AM You still have to fit the data for a thousand entities through the small pipe that is a consumer grade connection. Not really. Your client doesn't need to know about players that are far away and not visible. Furthermore, you could cut down the update rate for visible players depending on their distance to you. If someone's far enough away, their smoothened interpolated motions won't really be noticeable.But oh if everyone decides to meet in the same place... Title: Re: 1k player FPS Tech demo Post by: Sheepherder on March 12, 2011, 11:35:42 AM Quote number of players in a single multiplayer battle: 1000. That number of players with their test setup is already pushing 52 kByte/s even with the tiny amount of data per entity. Also according to them, they're already culling data update frequency based on distance(?). EDIT: Actually, it's probably a tree/node arrangement they use to determine what is culled. |