OnLive Begins Beta Testing

Steve Perlman, CEO of OnLive, has announced that beta testing is now underway for the cloud gaming service that aims to take the processing burden for cutting-edge games off a player's computer and use remote servers instead. Reaction to this service and competitor GaiKai has been interest tempered with skepticism, but users can now sign up to test it themselves and see if the reality matches the hype. There will be hardware and connectivity restrictions to start: "When you sign up for OnLive Beta, you tell us some general information about your ISP, your computer configuration and your location. We use this information to organize Beta testers into test groups so that our engineering team can focus at different times on testing different situations. If you are a potential fit for a particular test group, we'll send you an invitation email, asking you to run a detailed Performance Test on your network connection and your computer configuration."

Comments

world ?

[Spaham]

somehow they assume that the whole world lives in the United States...

Re:world ?

[Anonymous Coward]

Their servers are only in the US. They just handpick ideal candidates, so that no sub-optimal beta tester has anything negative to report about the service.

Re:world ?

[Tom]

It's a minor error. The average american knows very well that there's "stuff" outside the USA. It's just that the stuff doesn't matter, so it occasionally gets forgotten.

Re:world ?

[Desler]

Because there are no services outside the US that ever only restrict their services to their own country. No, never it's only Americans that do that.

Re:world ?

[slim]

Their servers are only in the US. They just handpick ideal candidates, so that no sub-optimal beta tester has anything negative to report about the service.

To do that would be stupid. I have to assume they're not stupid.

Even when OnLive goes gold, customers will be expected to be within a certain distance of a server farm. So beta testers should be selected within the same constraints.

They should, however, be picking beta testers at the edge of that distance, on streets with lots of contention, with crappy PCs, with ropey old DSL modems, etc., so that they can iron out problems.

The purpose of the beta test is not to demonstrate that everything works perfectly first time. It's to find problems and solve them. You don't do that by rejecting problematic testers.

Of course they must set expectations accordingly. You also don't want a load of beta testers blogging about how their free beta is crappy.

Re:world ?

[Drakkenmensch]

What really raises a red flag for me is that they want to test your computer for hardware compatibility. Wasn't the whole point of Onlive that any system could connect to their network regardless of hardware because all the actual processing is done server side? By their own words you'd think that a 300$ netbook should be able to play Crysis as long as it's connected to a solid, low-ping cable modem.

Re:world ?

[RemoWilliams84]

They should, however, be picking beta testers at the edge of that distance, on streets with lots of contention, with crappy PCs, with ropey old DSL modems, etc., so that they can iron out problems.

You also don't start your beta test with those crappy situations, especially when it is a completely new service. That's why you often start with an alpha or a small closed beta.

It's much easier to find the problems when you start with a small set of potential problems and work your way down to those users running a 28k modem on windows 95.

Re:world ?

[slim]

What really raises a red flag for me is that they want to test your computer for hardware compatibility. Wasn't the whole point of Onlive that any system could connect to their network regardless of hardware because all the actual processing is done server side? By their own words you'd think that a 300$ netbook should be able to play Crysis as long as it's connected to a solid, low-ping cable modem.

This simply means they want data on what kind of hardware their testers are using. I'd guess that at some point in the beta program (probably not the early stages), you'll be *more* likely to get picked if you're on a $300 netbook.

Stupid idea

[Anonymous Coward]

In the midst of crazy bandwidth and hardware improvements over the year, one absolute truth remains:

Latency is here to screw you over.

Re:Stupid idea

[GospelHead821]

This post makes me sad. I was always disappointed when somebody with a slightly higher latency would join a Counterstrike game I was in and the other players would vote him out so that they could all retain their better latency. People out in the boonies are no less deserving of games than anybody else.

Re:Stupid idea

[Nursie]

That's not really the latency that's the problem here, though it would affect the folks "in the boonies" more, most likely.

No, the problem here is that the absolute minimum time between you hitting a button and the corresponding action occurring on screen is the time it would take if the game was running locally PLUS the network round-trip time. That extra, in most cases, would dwarf the local latency (unless I'm very wrong).

I've got some sort of natural scepticism of all this cloud stuff as it is, but this does seem to me to be a real, no-quick-fix problem. I'll watch for people's reactions over the coming months as it gets tested.

Re:Stupid idea

[grumbel]

Which is why they plan to have their servers spread all over the globe, instead having them all in a single central location.

Re:Stupid idea

[wizardforce]

correct me if I'm wrong but the latency problem you speak of should also manifest to some extent in all multiplayer games [which it does] owing to the fact that when two or more players interact, their actions are limited by the latency of the network.

Re:Stupid idea

[RalphSleigh]

There is much that can be done with a clever client to hide and compensate for latency, see:

http://developer.valvesoftware.com/wiki/Latency_Compensating_Methods_in_Client/Server_In-game_Protocol_Design_and_Optimization

Camoflage drm ?

[pinkishpunk]

is it just me or does this might alot more advantages to the right holds of the game than it do to the user ? the whole second hand market, copying, lending would be "fixed" with this.

Why?

[ledow]

I get 25ms latency in an ideal online gaming situation (i.e. it takes 25ms for my input to reach the server and the server information to reach me) with a good, nearby server (same country). That's at rates like 4-5 kbps using retranmission, UDP, etc. to keep losses to a minimum. I'm not affected by peak periods because I have a very good ISP.

How is anything which requires significantly more data going to work anywhere near those latencies? First, my router can kill a gaming session if someone opens a couple of webpages - people's connections will have to be *dead* to allow multi-Mbps connections anywhere near reliably. Then you have that data having to be received and processed at both ends - not a big task for the consumer but acting on Mbps takes much longer than acting on Kbps no matter what you do. Then you have the lack of ANY sort of "predictive" technology - even Doom, Quake etc. knew to do input smoothing and not send every input event and have the client/server compensate by basically guessing if the connection lagged for a few ms - that's not possible here.

Then you have that the BBC iPlayer streams can effectively kill a business-broadband connection on their own without proper QoS and they are talking significantly more bandwidth, and some of it in the other direction too. So even in the *ideal* situation, with an *ideal* ISP it'll be *worse* than an average game of Counterstrike to play. Translate that to what most people who would be interested in this service have (noisy wireless, crappy broadband, slow ISP connection, etc.) and it just makes for a disaster.

I'd love it to succeed. I'd also love it to have beta testing somewhere other than the US - but I have to admit my main factor in taking up the beta program would only be to see just how bad it is.

Re:Why?

[slim]

And stop there. No remember how great gaming was on a TFT with a latency of 25ms.

Some hardcore gamers playing twitchy hardcore games were bothered by TFT latency. We're talking about the type of people who think it's worth investing in hardware so they can play Quake at 120FPS.

Most people can't tell the difference between 30FPS and 60FPS. Most modern games aren't reflex-oriented enough for it to matter. (At 30FPS, you're talking an average of 17ms lag, even if every input is guaranteed to be reflected in the next frame).

Yes, this is a concern for some racing games, the more hardcore fighting games, and the most intense FPSs. But for the vast majority of consumers, for the style of games popular today, it's not relevant.

Re:Because....

[wizardforce]

actual games are SO FAST NOW on so cheap hardware,

on newer hardware running windows that may be true... for hardware more than a few years old or on alternative OSes or for games like Chrysis this isn't true.

That 500MB files butchers my poor quadcore.

again, that's only true for newer hardware. Photoshop can use over two gigs of RAM to handle larger files in a timely manner. Don't have the RAM? You may just be screwed.

Re:Linux

[wizardforce]

quoting Onlive's beta sign-up page:

and have a broadband-connected PC running Windows Vista/XP, or an Intel-based Mac.

so not right now.

20 computers at 250ms?

[h00manist]

I wonder if we install this on all the computers here, will it play well on any pc, or just the newer ones..? and as for latency? "Cheap hardware" isn't exactly cheap for the whole world. I'm in South America, so 50ms is just not happening.

More Importantly...

[sonicmerlin]

Where can I buy some of their stocks?

Re:More Importantly...

[smartr]

Did you mean, where can you buy some put options?

The timing is perfect!

[Drakkenmensch]

Just in time for the american ISPs clamoring to lower the DSL speed requirements, potentially giving this new technology its death sentence before it even starts.

Re:Linux

[Nerdfest]

The work actually required to create a Linux (or OSX) client is much less than it is to create a complete game, and theoretically they only need to do it once for all games. I hope they take the opportunity.

Translation: Positive publicity ONLY, please!

[Scott Kevill]

If you are a potential fit for a particular test group, we'll send you an invitation email, asking you to run a detailed Performance Test on your network connection and your computer configuration.

So they can counter their critics by saying they had a positive public beta, yet with a carefully controlled group, to ensure pesky real-world situations don't damage their hype for gaining investors.

Re:Translation: Positive publicity ONLY, please!

[Keiran Halcyon]

Or, you know, so that they don't get five thousand uber-gamers all testing with the same general hardware range and then end up discovering at release day that anyone not running an i7 over FIOS is unable to play. You know, the other 90% of their target audience. Please tell me that you're not in any way related to QA in anything that you've ever done in your life.