Send News. Want a reply? Read this. More in the FAQ.   News Forum - All Forums - Mobile - PDA - RSS Headlines  RSS Headlines   Twitter  Twitter
User Settings
LAN Parties
Upcoming one-time events:
Germany 08/31
Chicago, IL USA, IL 10/19

Regularly scheduled events

Op Ed

Eurogamer - Why OnLive Can't Possibly Work.
"So, let's say that Grand Theft Auto V is released via OnLive, and (conservatively) one million people want to play it at the same time. We can talk about Tesla GPUs, server clusters, the whole nine yards, but the bottom line is that the computing and rendering power we're talking about is mammoth to a degree never seen before in the games business, perhaps anywhere. There may be a way how this can be handled (more on that later), but even having capacity for 'just' 5,000 clients running at the same time is a monumental effort and expense. It would be the equivalent of us running a single Eurogamer server for every reader who connects to the site at the same time. The expense involved is staggering (not to mention the heat all this hardware would generate - think of the children!)."

Wired - In Praise of the 3-Hour Game. Thanks Mike Martinez and Techdirt.
"Forty hours might sound like a reasonable amount of play. But the truth is that very few games offer an experience that truly requires — and rewards — 40 hours of play. After all, one of the chief joys of gameplay (which nongamers tend to misunderstand) isn't in having mastered it. It's in the process of mastering it. You start off stumbling around, not really knowing what your goals are, how your enemies and obstacles behave, or the complexities of your weapons and abilities."

GameZone - Tough Cookies: Are video games too easy now?
"From the development side of things, the desire to create a game that can be easily completed by the player makes a lot of sense. These developers and publishers are pilling millions of dollars into the creation of these games, from the production of the story elements and cutscenes, to its continuation and franchise potential. If you haven’t completed the first game in a trilogy, are you usually ready to jump into the sequels? After all, if you’d only made it halfway through Star Wars, would you really give a crap about The Empire Strikes Back?"

29 Replies. 2 pages. Viewing page 1.
< Newer [ 1 2 ] Older >

29. Re: Op Ed Mar 28, 2009, 00:03 Creston
To be fair, I would believe encoding done via hardware could hit that number

A) Then why isn't Youtube running it?

B) The cost of such hardware would be astronomical. You'd probably be talking about tens of thousand of dollars just for the encoding PER CLIENT. How on earth are they ever going to earn that back?

Then again, it's smart if they're scamming. Come up with some ludicrous idea coupled with bullshit marketing words (psychophysical research!) and then say you need to build datacenters that cost 50 million dollars each.

Edit : I share many of the concerns with the article but the biggie is, having this tech, why jump into a domestic service, and even then, why gaming?

The guy in the article actually said that very nicely. If they REALLY built this amazing encoding technology, why on EARTH would you waste it on gaming? Sell it to the military for 900 billion dollars and buy yourself a planet.


This comment was edited on Mar 28, 2009, 00:05.
Avatar 15604
Reply Quote Edit Delete Report
28. Re: Op Ed Mar 27, 2009, 19:25 DG
I share many of the concerns with the article but the biggie is, having this tech, why jump into a domestic service, and even then, why gaming?

A big red flag to me is, in three words, "cost of capital".

What they're saying is they got investors to stump up the gigantic amount of money not only to develop tech for 7 years, but buy data centres and all the other costs you get in starting up a big company from scratch. And working capital. That's a *lot* of funding.

People investing in new companies require a pretty big return, and that goes double when it's a tech firm and triple when the tech hasn't even been invented. You're looking at 30%pa minimum. That may sound like a great return but it's what you need to convince you to part with the cash, because for each winner you hit losers too.

Let's say you invest equal instalments annually at the start of each year. If you invested $1 each time, at the end of the 7 years you would be expecting nearly $23 back. Even if you invest nothing in the following year, you would be needing nearly $30 back. 2 more years, $50 please.

So, given the basics of investing, do you then go straight into domestic service rollout? Your balance sheet looks like shit and you have no realistic idea of when you will be able to repay those loans.

No way, it's total lunacy. You would licence the tech to non-competing companies. This gives an instant flood of cash with which to pay off investors, so your cost of capital is now like 10%. It also funds the working capital for the domestic service rollout (meaning you need less cash from the investors anyway).

I expect it works for tax too. To be honest I don't know about the US and I'm hardly hot on the topic over here either, but I expect it will still be the case where you do not get tax relief for R&D expenditure in one go, you release say 20%pa over 5 years. But, if you sell the tech it all comes off.

In contrast to the way they have chosen, this way the company has cheap capital, a decent balance sheet and patented technology. Tech that is somewhat proven (by the guys you sold it to).

That's how you make money.
Avatar 14793
Reply Quote Edit Delete Report
27. Re: Op Ed Mar 27, 2009, 17:18 Tiver
And do you remember how much people complained when they were 20+ms about how they were unsuitable for gaming? Though that was often more of a problem of ghosting. Still, it's a big concern for people for LCD's to have low latency when used for gaming, to the point where 8ms difference can be huge. We're talking about adding 20-120ms.  
Reply Quote Edit Delete Report
26. Re: Op Ed Mar 27, 2009, 17:03 Z9000
The large full featured LCD HDTVs have a 4-12ms delay as well.  
PS3 resurgance by GOW3 - Check! Mass Effect for PS3 - Check! Diablo 3 for consoles? I say "For sure"!
Reply Quote Edit Delete Report
25. Re: Op Ed Mar 27, 2009, 16:43 Tiver
Fang, I understand all that but I just don't see it being a simple 20ms of additional latency That seems to be an absolute best case. On top of the normal latency you'd have client side they'll be adding compression latency, transmission latency, and a quasi additional display latency.

8.33ms First, the fact they likely only poll and transmit an image at 60fps. We'll ignore the end display device refresh time as that exists in both but this is still something additional added. 16.67ms as you said, average of 8.33ms delay added. It is possible they've got a solution that knows every time the graphics card finishes rendering and only starts compressing then, in which case they could even transmit a variable bitrate stream, but nothing they've said so far seems to indicate this.

1-16.67ms Next, the encoding. The data has to be compressed, even using hardware compression if they reached 1000fps speeds it'd be a huge breakthrough and useful elsewhere. It's far more likely this is going to be much closer to the 16.67ms.

4.17-16.67ms Bandwidth, the video data has to be transmitted. Both network latency and bandwidth will affect this. They've mentioned 5mbps of bandwidth. If a user only barely has enough bandwidth to cover this then it'll incur a 16.67ms of latency, if they instead have a connection capable of say 20mbps and the data was capable of using all of that in bursts this could add as little of 4.17ms of latency.

5-40ms Network Latency, the time it takes the input packet to hit the server and the time it takes the data to cross the network. The longer the path the greater the latency. This one unfortunately varies drastically based on connection users have and where the servers are situated. Best case I'm thinking they can get this down to 5ms when colocating on every users ISP and the user is on a low latency connection to their ISP. More likely is 40ms but varying on users connections could easily be more.

1-16.67ms Decoding, as you mentioned they then have to decode the frames on the client end. I doubt they'd put as expensive of hardware into decoding but since decoding is easier maybe it could accomplish 1ms. At worst though it has to manage 60fps.

Total 19.5-116.84ms

At 19.5ms, this could actually work. At 116.84ms that's almost 6x the normal input lag and would not be enjoyable to most players. If there's latency induced through any other part of the setup or the servers have even more network latency between them and customers then it could get really bad. It is possible that initial 8.33ms delay I put in there could be removed but I'm less inclined to believe they accomplished that when they spout so much other crap.

It'll be interesting if they ever open this up to beta testing and we get actual real user reviews on it. Even then though, what are they going to have to charge to make a profit off this?

This comment was edited on Mar 27, 2009, 16:48.
Reply Quote Edit Delete Report
24. Re: Op Ed Mar 27, 2009, 15:57 Fang
To be fair, I would believe encoding done via hardware could hit that number, with the caveat that I'm not a video encoding expert. If they had software algorithms that could hit that number, I would be very interested in seeing them as a computer scientist.

But the follow up question should have been what is the latency introduced by the decoder which would have to be done in software since that is their claim.

And the economics question of the additional cost of the hardware encoders.

Anything can be done with technology (well, almost anything). The real question is the cost, and will a business model support it.
Reply Quote Edit Delete Report
23. Re: Op Ed Mar 27, 2009, 15:38 Fang
Trust me, I know about client-side prediction. You're ignoring the many sources of latency for even something like a single player game.

Though your RockBand example is a good case for why this tech can be "good enough." We're talking < 20ms additional latency and for the most demanding gametype for low latency, it was still playable. Albeit, annoying, but still playable. So maybe selling Guitar Hero on it is not a good idea.

But for something like a console shooter which already has auto-aim and a 30 Hz internal game clock, it will probably be acceptable to the masses.

For comparison, let's take a standard console game, with a wireless controller and hooked up to a 60 Hz display (updates every 16.7 ms). It will most likely be running at a 30 Hz internal game clock. Depending on when your input comes in, it can have a 33.3 ms delay. Though there will be a window in which an input can processed for that "time slice". So your input will really be delayed something like 4 - 37.3 ms, which would mean an average of 20.7 ms delay. This is before taking into account the wireless controller delay and the video display delay.

The flaw in many people's thinking here is that they assume that there is 0 latency in current single player games. Going from 0 to 120 ms latency is a big deal in their minds. But the true jump is more like from a current 30 ms latency to 50 ms. Sure, it's not negligible, and you have to make sure it's not the straw that breaks the camel's back. But it's not a non-starter, at least for the non "PC FPS" crowd.

Do you know the internal clock that old arcade games were running at? When you play those games, do you feel frustrated by the lack of instant responsiveness to the controls? My guess is probably not. For most gametypes, you probably will be able to deal with an additional 20 ms.

Sure you may not want to play CS in a competitive setting on a system like this, but something like Halo 3? For guys who don't even feel the need to use a mouse?

This comment was edited on Mar 27, 2009, 15:58.
Reply Quote Edit Delete Report
22. Re: Op Ed Mar 27, 2009, 15:17 Cutter
More than that, OnLive overlord Steve Perlmen has said that the latency introduced by the encoder is 1ms. Think about that; he's saying that the OnLive encoder runs at 1000fps. It's one of the most astonishing claims I've ever heard. It's like Ford saying that the new Fiesta's cruising speed is in excess of the speed of sound. To give some idea of the kind of leap OnLive reckons it is delivering, I consulted one of the world's leading specialists in high-end video encoding, and his response to OnLive's claims included such gems as "Bulls***" and "Hahahahaha!" along with a more measured, "I have the feeling that somebody is not telling the entire story here." This is a man whose know-how has helped YouTube make the jump to HD, and whose software is used in video compression applications around the world.

Ahahahahaha! Love it, pure gold! Where's OnLive's shill today?
Avatar 25394
"Lights going out and a kick in the balls. That's entertainment, that's entertainment"
Reply Quote Edit Delete Report
21. Re: Op Ed Mar 27, 2009, 14:24 Tiver
"I believe latency is being overblown as a problem by those who don't know much about internet topology."

Client-side prediction, ever hear of it? It's why your lag to servers doesn't show up as input lag. When you move your mouse left in an FPS the client tells the server to turn you left but also does it immediately client side, predicting the server will accept it. This cause the 18-50 or whatever ping time you have to a server to not affect your input lag.

Because this will be a thin-client, there will be no client-side prediction. Your inputs will be lagged at the time it takes for them to be sent to the server and for the encoded video to be sent back and displayed. For input lag it starts becoming noticeable around 10-20 ms and gets progressively worse. By 100ms many users will find it unbearable. I was playing rockband and we had the video latency setting off by 20ms and it was noticeable and annoying to everyone playing, now imagine if we had it off by 100ms? Then imagine if instead of a predictable shown sequence to offset our input for it was an FPS with human opponents?

Go find one of those crappy wireless mice with high input lag, that'd give you a good preview of what this service is going to be like.
Reply Quote Edit Delete Report
20. Re: Op Ed Mar 27, 2009, 14:09 MMORPGHoD
I still have a hard time believing this technology is that much more advanced than existing remote desktop software.. which sucks at 1.5Mbs DSL. That's essentially what this software is, except now you're streaming a game rather than a fairly static 2D desktop.

This comment was edited on Mar 27, 2009, 14:13.
Reply Quote Edit Delete Report
19. Re: Op Ed Mar 27, 2009, 13:28 Psyclone
As for the heat, what do you think all those Xbox's and PS3's are doing already?

lol wut?
Reply Quote Edit Delete Report
18. Re: Op Ed Mar 27, 2009, 11:58 Fang
Heck, when I was back on DSL, my pings were around 25 ms along the coast.

And according to, the US has an average of 6 MB/s download speed from their testers. Of course, that's counting the number who voluntarily take their test, so it will probably skew high.

My point is that DSL is being dropped with the switch to Fiber.

They have already stated that the servers will be located on each of the coasts. That's all you need to be close to most of civilization, right?

Yes yes, they'll include a midwest server farm for "fly-over" country

So if you have 18 ms round trip times for most of your users on the West and East coasts (assuming your users are of the high tech/first adopter crowd), this is small potatoes compared to other sources of latency that's there even when you're playing with your current Xbox 360 setup. Okay, more like a big watermelon. But it's not the Titanic that everyone here is making it out to be.

But then economics is an impact here. The users who have the high speed internet to use your service at a reasonable level, are also the ones with the high end PC's and multiple game consoles. They don't need your service.

My prediction is that this will die on the vine, not because of the technical challenges but because of the business challenges in creating enough revenue to support the computing requirements.
Reply Quote Edit Delete Report
17. Re: Op Ed Mar 27, 2009, 11:31 NKD
NKD, you do know that ping times are round trip times, right? So 50 ms isn't exactly being generous since that's implying a 100 ms ping time.

Yeah I'm aware. I was assuming an average Internet connection like DSL or low-end Cable, and that the servers were not very close. So yeah I guess I wasn't really being generous.
Avatar 43041
"America is like an MMO where an early guild already took 90% of the good farming grounds, boss raids and rares while everyone else is still trying to level up on low-level mobs" -Some Guy on Reddit
Reply Quote Edit Delete Report
16. Re: Op Ed Mar 27, 2009, 11:28 GT
I think what the heat joke means is the added heat to the system "earth", not to just one enclosed space.
The "think of the children" part was the joke, but he was serious about the heat because it would be a serious problem in the server farms.

This comment was edited on Mar 27, 2009, 11:31.
Reply Quote Edit Delete Report
15. Re: Op Ed Mar 27, 2009, 10:54 Fang
NKD, you do know that ping times are round trip times, right? So 50 ms isn't exactly being generous since that's implying a 100 ms ping time.

I get 18 ms when I ping along the east coast. But then I don't have a crappy cable ISP.

Don't get me wrong, I agree OnLive has formidable challenges with computing power and the underlying economics, but I believe latency is being overblown as a problem by those who don't know much about internet topology.

Though I should say while average latency for customers with non-crappy (read non-cable) ISP's shouldn't be a problem, consistency of service (which Eurogamer points out) can be troublesome.
Reply Quote Edit Delete Report
14. Re: Op Ed Mar 27, 2009, 10:22 Grendle
I think what the heat joke means is the added heat to the system "earth", not to just one enclosed space.  
Reply Quote Edit Delete Report
13. Re: Op Ed Mar 27, 2009, 10:01 GT
As for the heat, what do you think all those Xbox's and PS3's are doing already?
They're not all stacked in the same server room like this would be. Heat would certainly be a major issue when it comes to scale.
Reply Quote Edit Delete Report
12. Re: In praise of the three hour game Mar 27, 2009, 09:41 eRe4s3r
Indeed, what is it with people wanting shorter games, I mean choice and all is good, but the trend for shorter games that cost more is somehow.. not good...

That wired article states that the path to mastering is what makes a game good, but that is non-sense, people don't always play games for mastering the gameplay, they want engrossing story/scares/npc's and back-story, intelligent quests etc. That wired article is way way too focused on shooters, apparently all the editor ever played, because i certainly don't play most of the games for mastering them... i play them because they are fun, whatever that is ^^

Avatar 54727
Reply Quote Edit Delete Report
11. Re: In praise of the three hour game Mar 27, 2009, 08:54 Riker
I'm happy for the Wired writer who enjoys short games, but I don't. There's a market for those people, and there's a market for my people. Deus Ex, System Shock 2, or Half-Life 2 as 3-hour games? I don't think so.

Fallout 3 did a reasonably good job in offering a great deal of game play, but people with short attention spans were welcome to only follow the main story line, which ties up relatively quickly. I think that's probably the best way to go; offer different types of experiences for different types of people.

This comment was edited on Mar 27, 2009, 08:54.
Avatar 6580
Reply Quote Edit Delete Report
10. Re: Op Ed Mar 27, 2009, 06:12 InBlack
Finnaly a good article from Eurogamer, its been a while since Ive read a good one there.

Im so sure that OnLive is a glorified scam, the investors will get swindled out of millions of dollars, the managers of the company will get massive paychecks and nice resumes, and the actual developers will all get laid off regardless of the fact that they may have created a product that works in THEORY.

Think "Flagshipped" to the n-th potential!

All in all I still hope that Im wrong and this thing magically works, but hope is one thing, common sense is another.
Avatar 46994
I have a nifty blue line!
Reply Quote Edit Delete Report
29 Replies. 2 pages. Viewing page 1.
< Newer [ 1 2 ] Older >


Blue's News logo