Investor Money Goes To Magic Lag Reducing Tech

Gamasutra reports on Texas technology company Bigfoot networks, which just received a $4 Million investment to develop a lag-reducing hardware PC card. From the piece: "According to the firm, it will bring to market the world's first Gaming Network Accelerator card, which will allow online gamers to play their favorite games with less lag. The company explained: 'Lag is the number one problem in online video games today, and Bigfoot Networks is the only company in the world whose sole mission is to fight lag', but gave no specific technical explanation about how it intends to do this." Greg Costikyan spells it out on the Games*Design*Art*Culture blog: "So yes, there might be a business here. But if so, it will be a business built largely on bullshit."

Comments

Reducing lag?

[Andr0s]

Honestly, I don't see how user-side hardware (or software, for that matter) can reduce online activity lag. Sure, you might try to implement some sort of protocol that evens out the lag a bit by pulling excessive amount of data when 'lag is low' and use it to fill in the gaps when 'lag is high' - but that'd require a certain, no small, amount of heuristics and second-guessing. I'm certain many of early MMO veterans remember the ancient lag issues from the times of real-time simulations - fast ones in particular, such as flight simulators, suffered tremendously from lag-related issues such as phantom opponents (where your 'second guessing' lag-compensators assumed that opponent would continue in a straight line or at the same turn radius/speed, whereas he actually went into some wild maneuver). In the current state of affairs, I'm honestly not sure how much, if any, of the lag in your average MMO is user/connection-side and thus corrigible; games such as World of Warcraft, City of Heroes and Battlefield 2 are actually playable over dial-up - the trickle of packages isn't a lot of challenge even for a stable 56k modem. The bottleneck of modern day MMOs seem to be game servers going slightly ballistic when a certain area gets swarmed by a large number of active player objects (think Ironforge in WoW or Atlas Park in CoH) and therein lies the catch... how do you expect client-side hardware to correct server-side problems?

Sure, why not

[smooth wombat]

After all, there's a sucker born every minute.*

This isn't any different than the phantom console, magnets which supposedly help your arthritis or whatever book that Kevin Trudeau is bilking people into buying claiming this is information that the government doesn't want you to know about.

This shouldn't surprise anyone. Not the least of which that there are VC idiots who will gladly pony up the money for a non-existant, never-to-be-made product simply because it has oodles of neat sounding words in its description.

*PT Barnum never actually said those words but people routinely attribute the phrase to him.

The "gold cables" of gaming?

[Philus]

I guess it was just a matter of time before something like this appeared.. The hi-fi industry has cables and magic boxes all over the place, now we get magic hardware.. I'm VERY curious to how they plan to eliminate lag introduced by routers that they have no control over. Not to completely blow them off, but I'm not holding my breath. Seeing is believing.

I have a semi-decent 5.1 surround setup, and have avoided expensive cables because I simply don't believe in it. Audio cables might benefit from better shielding and low capacitance wiring, but digital signals.. come on man. A bit across the wire that's "worn in the edges" is still a bit, unlike a sound wave.

Re:Lag attack

[steinnes]

Analyzing the connection? How do you propose that happens, I doubt anyone is using a protocol for their online game which is so complicated that it has routing functionality built into it at the application level. Also there would have to be a choice of different servers available, and usually (for MMO's at least) people choose their general area in the world, and I think most game services worth their salt then try and find a server for the client to play on which has relatively low latency. When it comes to "analyzing the connection", I think people just have to rely on the internet to do their routing, and a client machine with one gateway pretty much doesn't have much choice besides their one gateway with regards where to send the packets.

Re:Lag attack

[stienman]

If you're suggesting static routing (ie, put route information into the packets so they go the route you want them to) then I doubt it's going to help. IIRC many routers ignore that option. Further, it would simply add to the size of the packet and increase processing time at each juncture.

There are really only two places for lag: the PC itself, and the network. I can imagine a card that optimizes itself for gaming packets. Imagine, for instance, the card estimating the arrival of a new packet coming from the computer. It can start sending out the packet header even before the computer gives it the packet information and data. At minimum it can start sending out the packet as it's received from the system rather than waiting for the complete packet (and CRC check).

Still, we're looking at 1-5mS speed increase, and that increase is significantly less than, for instance, removing the router/firewall software/hardware.

I'm interested in what the hardware does, but I can't see it improving the experience enough that even a heavy online player will notice the difference.

-Adam

Not Impossible, Just Improbable

[jevvim]

First, they're terming "lag" as anything that delays your game - network latency, network loss, system latency. I can't envision their card reducing network latency or loss, but it certainly could be optimized for system latency. First off, a lot of network adapters try to play nice by interrupting the system only after multiple packets have arrived (resulting in nearly-full buffers) or after data has been waiting in a buffer for a certain period (sometimes up to 100ms, depending on the card). Creating a "gaming card" that reduces these delays by default (which really avid gamers can reduce as well, since most drivers allow these parameters to be tuned) could allow a company to market a "new network card" that's really just a differently tuned driver set.

More improbably, though, is that Bigfoot Networks could implement and expose a programmable protocol processor on the card. This won't help existing games, but would enable developers to move some of their protocol closer to the wire, where it may be possible to buffer data more efficiently (send one "game state" packet to the protocol engine, which can then create the multiple unicast packets needed, instead of sending multiple wrapped network packets with effectively the same data across the PCI bus multiple times). However, this will require games to be adapted for it - somewhat unlikely - and even then would only provide significant help for game servers. But since many games - Quake, Half Life, et al - are hosted by home users, it might reduce lag in some situations.

Of course, without a product to play with or any real announcements from the company, it's just speculation at this point. But I'd love to play with a programmable protocol processor - such a device could open up new opportunities for network efficiency innovation (running PPPoE in hardware, integrated firewalls like the nForce ethernet, not to mention TCP, segmentation, and checksum offloading).