More on Grid Computing and Gaming

securitas writes "Sony, IBM and Butterfly.net will announce and demonstrate a new grid computing network for PS2 online gaming at the Game Developers Conference next week. The network is based on Linux and the Open Grid Services Architecture (OGSA) and is designed to support millions of players. This is believed to be the first major consumer application of grid technology. Read the details at the NY Times, CNET and the Washington Post."

Now We Need Games!

[6e7a]

It seems to me that the more technology these companies throw at games, the less I feel the desire to play them. Don't get me wrong: I love excellent graphics and sound. I just think the playability suffers when a game developer spends so much effort on the technology. I'm glad we have such a scalable platform for online gaming. I just haven't seen games that are as compelling as they used to be to take advantage of the platform. Am I getting too old for video games?

I'm excited

[arvindn]

about the implications that this could have for other applications. As an example, consider IBM's Deep Blue chess playing program that defeated Kasparov in 1997. It used a massively parallel grid for evaluating positions using custom-built hardware costing millions. Now imagine if the same thing could be achieved over a grid on top of the internet. You have a world champion beating chessplayer right on your desktop!

Another application would be in natural language processors. They require huge databases and computing power to process them. A grid would be a perfect way to build such a system.

Mind you, these applications are equally commercially viable. You could charge say $1000 per game against the world champion chess program, or $100 for 30 minutes of conservation with the most intelligent bot ever, and so on.

Re:What's so special about the grid?

[rw2]

ARPANET was about creating a network that was resiliant to bad things.

There are parallels. When an arpanet node goes down routing takes place on the other nodes instead.

In a grid there are many nodes. Some have speciallized resources that are fairly single point of failure suseptable (e.g. mass storage systems, large experimental devices), but most can be supplanted by another node.

That's where the analogy stops though. Where arpanet was concerned with networking, grids are concerned with networking on in that they use them. They are really about job movement, data movement, resource discovery and _security_.

Sony's mysterious moves

[thatguywhoiam]

I've been following the development of Sony's console-based online efforts for a little while now, and I have to say they are definitely up to something big. Between this so-called Cell chip (or Grid or whatever), and now this interesting collusion with IBM (again) on their Butterfly.net... it raises some intriguing possibilities.

However no one I've spoken to has the slightest clue as to how they plan on using this Grid stuff. Does anyone know any details? All I see are people saying 'no bandwidth, latency', etc.... I still can't figure out what it's supposed to do. Which is maybe on purpose.

If you look at the chess pieces on the board, so to speak... MS with Xbox, MSN, flavours of XP with media/TV style abilities... then Sony, aligned with IBM for a new chip and a radical new network... not to mention the Cell sharing some tech with IBM's forthcoming Power derivatives for Apple...

Strange things are afoot at the Circle K...

OGSA for gamers

[tom_conte]

The OGSA stuff is not necessarily about distributed computing, although the emphasis is certainly placed on this aspect in their docs.

The basic service a Grid infrastructure can provide to gamers is "peer groups": you can discover groups of people willing to share a game online, and join their group, and chat and play with them, without having to log on to a central server.

You could then imaging sharing add-ons and various other files with your peer group, again without using any central server.

The next step would of course be sharing the actual CPU time of all the devices, for example to keep your characters "alive" even when your console is switched off. And then you'll receive an SMS whenever he gets attacked :-)

Re:hmm

[s.d.]

How is this different from Parallel computing?

It isn't. Grid computing is the big buzzword of the day these days. I work with, and on, Globus, and this stuff just doesn't work yet. But beyond the fact that it isn't reliable software, what IBM is doing with Butterfly isn't really Grid computing. They're just saying that to get publicity.

Some of the original articles last year attributed features of the "grid" they're setting up to the Globus software, while anyone who has actually installed Globus knows that it can't do (things like accounting, failover services, etc).

g`day...

[m1chael]

grid computing is unfeasible over long distances. maybe they mean when you take your ps3 to your friends house and network them. otherwise this grid computing sounds like sony's little internet.

Re:Console breakdown, reality crashes in...

[EpsCylonB]

Sorry but this is misleading, the CPU's for the PS2 and GC are both specifically designed for games, the xbox is still more powerful but it is a lot closer than those numbers represent.

The xbox is running a x86 intel chip, I'm sure most of us here don't need to be reminded that the current x86 chips we use today are descendants of of chips designed purely to crunch numbers for business applications. Quite a lot of CPU cycles are wasted in current computers, if you were trying to design a CPU for an interactive entertainment system you would not design it like an x86 chip.

Also the fact that the PS2 is 18 months older than the xbox explains why it's spec seems to be unimpressive.

Re:Console breakdown, reality crashes in...

[fbg111]

When presented with the facts, it's clear which machine is superior and which machine is using latent technology.

Cuz we all know that CPU MGz is the defining performance factor, and that theoretical polygon count is the same as actual polygon count.

Currently the Xbox has the most power, but the least utility.

Define "utility".

nobody has really pushed the box to it's limits in a game

Let me guess, Microsoft told you that?

I actually own both consoles. Yes, Xbox wins out graphically, but not by leaps and bounds. PS2's only flaw is its poor texturing ability and lack of hardware shaders for surface effects. But good devs like Naughty Dog and Rockstar (with GTA:VC) are coming up with some nice software graphical implementations. And in the end, it really is about the games, and PS2 continues to dominate in that respect, although Xbox is slowly getting better.

Re:Console breakdown, reality crashes in...

[gergi]

Just so you know...
The Polygon count specified for PS2 & XBox are optimum but not "real-world" while the GC polygon count is "real-world". The numbers are actually fairly even across the board.

The CPU for the PS2 & GC are also designed for games, which the XBox is not. This makes a huge difference.

It boils down to this:
Sony relies on different numbers, the number of people who own a PS2 and the number of games available. Which is why the PS2 is #1 in games sold, by a huge margin.
Nintendo doesn't rely on numbers at all, it relies on games to speak for the quality of the GC. Which is why the Gamecube is by far the most profitable of the current systems, despite selling less games than the PS2.
XBox is relying on the argument the TurboGrax and other failed products rely on: "Better" hardware. They need to focus on more games if they want any market-share. Which is why Microsoft is in last place world-wide and losing a fortune.

It's the games, stupid.