It might very well work, but it's going to need ideal conditions to do so, and my internet connection, even at 24mb/s is not going to be able to support what they are promising.

I think it was mentioned somewhere that for SD resolution a 1.5mbps connection was required, and for HD resolutions, a 5mmbps connection was required!

I think the system will work just fine at the start, the problem I see is them sustaining the performance after 6 months when more and more people sign up. If the system fails or 'lags' at the early stage at all that could be the breaking for them.

I love the principle of the system though, rest in peace £450 8800GTX card, hopefully I won't need to buy another like you again for high spec PC gaming (Well, once its rolled out in the UK)

I think an important factor that OnLive must mention when they say "it works, we've used it fine" is what country this was in and what time of day.

In the UK especially, our upload rates are - and always have been - completely crippled. I'd like to see/hear someone test this service in a busy town, at peak time, 3 miles from the nearest exchange, on a BT line broadband package (BT themselves or wholesale) and then see what the latency is like.

Last time I checked my broadband speed, it came out as 7Mbit download, but ~570Kbps upload... and I live about 1 mile from the exchange...

Like everything else: "seeing is believing", in this case Playing it is believing it will work...

OnLive will work if ISPs ditch fair usage policy.

otherwise it will work for first few month, and then won't work during peak hours

The fact that big publishers are behind it surely isn't the dealbreaker that they claim? For the publishers, they have nothing to lose by providing their support and products. If it doesn't pay off or work as intended, they will continue to distribute their games as they always have. If it does, they get a new and potentially lucrative business model for games distribution.

To be honest, the issues raised by the Eurogamer article are valid. To be honest, I can't help but feel skeptical...not only because I spent just over a grand on a new PC 5 months ago, but because it does seem to good to be true. There are too many things that have to be correct for this to fully work- latency issues being the foremost...

But as the Eurogamer article does conclude, I hope I am proved wrong. Only mass real world tests will see...

Upload speed shouldn't matter (much) since your only sending the same data you would need in any online game.

At least they finally clariffied the lag: 35-40 Ms between user input and screen update. For a strategy, RPG or even RTS, I can't see this being a problem, but twitch gaming genres (FPS, Racers, Platformers etc.) are going to be awful to play. You run to a ledge and press jump, and your charachter hapilly flies off the edge, and then proceeds to attempt to jump half a second later. It's a massive delay in real terms.

Crisys on eee pc? NO WAY! ;-)

Well, of course the publishers are going to be supportive of this. After all, they would love nothing more than for you to keep paying for their games rather than buying a copy and being able to play it for as long as you want.

OnLive is going to revolutionize Gaming. Watch their hour long video from GDC, it answers most of the questions that are posed. Most people dont understand what OnLive is doing.

I think they(OnLive) don't understand that they doing either =)

The biggest hurdle that they'll have to overcome is hardware support on their side of things. They can't magically run any game at a given resolution with any less hardware than anybody else can, so assuming they're using the same games we buy off the shelves, they'll need the equivalent of one fully-fledged PC per subscriber. I just don't see how they'll manage to keep up with that once they get any real kind of subscription base.

Originally Posted by Goty
they'll need the equivalent of one fully-fledged PC per subscriber. I just don't see how they'll manage to keep up with that once they get any real kind of subscription base.

No, they won't. They need one PC per concurrent player. I game for about 2 hours a week, so they only need a PC for me 1.2% of the time. For the other 98.8% of the time, the PC can be assigned to other subscribers. Say a decent gaming PC costs £500, they could charge me £1 per month and have their money back in under half a year. Obviously they have to pay for bandwidth, maintenance, a building etc. but the whole point is that you spread the cost across many subscribers.

they say : data centre no further than 1,000 miles . Yeah right 1000 miles is around 5ms of lag in one(!) direction @ a speed of light (!) in a perfect situation with no network-hardware lags , and thous are far from zero too. Only a signal lag would be up to 10ms + network hardware lag + internet lag + video encoding lag + video decoding lag + rendering takes time too ;) , so OnLive would be more like OnLAG LOL :)

Originally Posted by Bauul
At least they finally clariffied the lag: 35-40 Ms between user input and screen update. For a strategy, RPG or even RTS, I can't see this being a problem, but twitch gaming genres (FPS, Racers, Platformers etc.) are going to be awful to play. You run to a ledge and press jump, and your charachter hapilly flies off the edge, and then proceeds to attempt to jump half a second later. It's a massive delay in real terms.

Even twitch gamers anticipate lag, I find that it takes a little while to fine tune and sync your actions (so to speak) with a servers lag. But when you do...... it's KILLING SPREE!! HOLY SH1T!!

Even if this does work as advertised can you think of the cost of hardware/power!

They recon they can get 10 players per server. Steam have 1.7million concurrent players online at peak. so if onlive get that many players thats means 170,000 high end servers minimum!

Originally Posted by steveo_mcg
Upload speed shouldn't matter (much) since your only sending the same data you would need in any online game.

Actually, you're sending much less than you would need for most normal online games - all you are sending is input data from your control devices, which will be trivial (on the order of a few tens of kb/s at the most) - think about it, even if it sampled control inputs 50 times a second and sent 1kb (far more than is really likely to be needed) for every sample, that's only 50 kb/s.

Originally Posted by Bauul
At least they finally clariffied the lag: 35-40 Ms between user input and screen update. For a strategy, RPG or even RTS, I can't see this being a problem, but twitch gaming genres (FPS, Racers, Platformers etc.) are going to be awful to play. You run to a ledge and press jump, and your charachter hapilly flies off the edge, and then proceeds to attempt to jump half a second later. It's a massive delay in real terms.

er...40ms is less than a 20th of a second, not half a second mate! 40ms will be fine. I just find it hard to believe that 40ms will be achievable with anything less than perfect network conditions (i.e. whenever it is running "in the wild" rather than on a LAN in their labs).

Originally Posted by ch424
No, they won't. They need one PC per concurrent player. I game for about 2 hours a week, so they only need a PC for me 1.2% of the time. For the other 98.8% of the time, the PC can be assigned to other subscribers. Say a decent gaming PC costs £500, they could charge me £1 per month and have their money back in under half a year. Obviously they have to pay for bandwidth, maintenance, a building etc. but the whole point is that you spread the cost across many subscribers.

but at what time of the day do you do those two hours? i very much doubt much is done at 5 in the morning, but that is the time you would probably find your concurrent pc isnt being used (or attempting to be used) by 10 other people.

plus, if my 8mb connection fails to provide me with a decent game of l4d at peak times i cant see how its going to be able to play crysis at 720p on a pc 1000 miles away. seems to me at least here in the uk our infrastructure is nowhere near good enough yet, i think someone else said they should have tried this in japan or similar fibre countries first, why start in europe?

the fact that all these major companies have invested in this shows it has potential, whether that potential can physically be realised is another matter. for one thing i definately dont think this is 'omg the end of the hardware industry!!!1111!1!1' there are always gonna be people (current pc gamers) who want to play their games any time they like, without 35-40ms lag and at a higher res than 1280x720. in fact we could find more people coming over to our side of the tracks once they've tried some of these titles and want to see the full extent of pc gaming.

time will tell, its certainly going to be interesting either way!

Originally Posted by will.
It might very well work, but it's going to need ideal conditions to do so, and my internet connection, even at 24mb/s is not going to be able to support what they are promising.

same, even if it works I can't see my Virgin powered conn being able to handle this :(

Originally Posted by mclean007
er...40ms is less than a 20th of a second, not half a second mate! 40ms will be fine. I just find it hard to believe that 40ms will be achievable with anything less than perfect network conditions (i.e. whenever it is running "in the wild" rather than on a LAN in their labs).

They tested it live from a server 50 miles away, not a LAN connection.

Originally Posted by Major
They tested it live from a server 50 miles away, not a LAN connection.

And according to what I've heard from journalists I trust, such as the Idle Thumbs group, there was still a noticeable feeling of delay in some of the games they had at GDC.

What if the servers go down? You can't play any games.

Originally Posted by Leitchy
I think it was mentioned somewhere that for SD resolution a 1.5mbps connection was required, and for HD resolutions, a 5mmbps connection was required!

If that's 5mpbs sustained, I can see ISPs fighting this. They might expect you to grab a sizeable chunk of the upstream feed for a short period (file download, for example) but in their shared model, pulling off a 5mbps chunk (per user) for a sustained period of time is going to result in limits or outright banning... or price increases.

Originally Posted by Major
They tested it live from a server 50 miles away, not a LAN connection.

And what kind of connection did they have between the client and that server? Not a domestic DSL line, I bet. Also, how many subscribers will live within 50 miles of the 2 servers they are rolling out at launch? My point was that the technology may work acceptably in controlled "best case scenario" conditions, but I would be surprised if the experience is as good in the real world.

Originally Posted by mclean007
er...40ms is less than a 20th of a second, not half a second mate! 40ms will be fine. I just find it hard to believe that 40ms will be achievable with anything less than perfect network conditions (i.e. whenever it is running "in the wild" rather than on a LAN in their labs).

They say in that article that 40ms is an expected average, maximum of 80ms, try my little latency experiment I find 60ms or less playable, and I'm not a particularly fussy or hardcore gamer, so I think they're going to need a better maximum round trip latency than 80ms if they want to dominate the market.

Originally Posted by mclean007
er...40ms is less than a 20th of a second, not half a second mate! 40ms will be fine. I just find it hard to believe that 40ms will be achievable with anything less than perfect network conditions (i.e. whenever it is running "in the wild" rather than on a LAN in their labs).

*facepalm*

Yep, my bad. Either way, I use Citirx on a daily basis (exactly the same concept, but for Office apps) and any kind of delay at all (let alone 40ms) is not only very noticeable but very annoying. The problem is when you do anything that is not based on *looking*, *acting*, *looking to see if it worked*, *acting again*. A lot of time gameplay is not like this, you pre-empt what your actions will do.

Take an FPS. You see an enemy is 15 degrees to your left, so you spin and shoot. You don't think about it, you just react and you get the kill. With a 40ms delay on your input affecting the screen's update, and given this latency will change over time (it may jump from 40 to 50 and back again in a second) you'll probably not only miss, but you won't even know if you aimed correctly until the game catches up. What's more, the command to "fire" may make it to the server quicker than the command to "turn". This is the Internet afterall, and you end up shooting too early. You end up doing everything half as fast because you're not 100% sure of what effect your inputs will have.

I really hope to be proven wrong, but it was a real shock to the system when I started using Citrix. Even just opening the file menu and pressing New on Excel suddenly felt awkward and difficult. You have to go to File, press open, wait, then move down to the New and click on it, and wait. This would totally kill a quick-reaction game.

Originally Posted by Shuriken
They say in that article that 40ms is an expected average, maximum of 80ms, try my little latency experiment I find 60ms or less playable, and I'm not a particularly fussy or hardcore gamer, so I think they're going to need a better maximum round trip latency than 80ms if they want to dominate the market.

Good test, but totally different in an FPS, any slight lag is annoying, it's smooth or nothing.

And with that block, 80ms was fine for me, and I used to be a hardcore FPSer.

I couldn't think of who had something like OnLive.

http://en.wikipedia.org/wiki/Sega_Channel

Let say it work IF and only IF you're the only one who is playing a game. What happen when 100, 500, 1000.... of people wanted to play the same game you're playing? That's remain to be seen to believe.

Originally Posted by ch424

Quote:

Originally Posted by Goty
they'll need the equivalent of one fully-fledged PC per subscriber. I just don't see how they'll manage to keep up with that once they get any real kind of subscription base.

No, they won't. They need one PC per concurrent player. I game for about 2 hours a week, so they only need a PC for me 1.2% of the time. For the other 98.8% of the time, the PC can be assigned to other subscribers. Say a decent gaming PC costs £500, they could charge me £1 per month and have their money back in under half a year. Obviously they have to pay for bandwidth, maintenance, a building etc. but the whole point is that you spread the cost across many subscribers.

not even, with virtualization, one server can probably run 4 or more people depending on the game, on top of what you said they'll be for concurrent users. That's way more efficient they can have 4x the amount of users on a £1700, and make more money that if it were 1:1, pc:active user.

Wait, Epic gave their stamp of approval?

I need OnLive!!!11

Because everyone knows Gears of Lag provides the best online gaming experience...ever.. :)

Been playing FIFA 09 on PS3, and that for me is laggy, even those both connections are smooth, the delay with SP vs MP is pretty big for me anyway.