Say you have a web based service with 2000 paying subscribers, and the servers are creaking a bit under the load. Do you spend a month optimising code, or drop another server into the cluster?
Of course there's a tipping point, which is why the post you replied to says "it depends".
(Similarly, you don't spend hours striving to make a data structure one byte smaller -- unless that data structure is repeated millions of times at runtime.)
Those are good examples of titles for which I can't think of a direct clone (although I bet if you trawl through the Wii shovelware catalogue, you'll find a shoddy Animal Crossing clone).
But the key thing is that there's nothing stopping a developer from making a direct competitor to these games. The fact that they haven't done so simply suggest that nobody believes strongly enough in a market for such a game.
Unless you typed that post by moving a cursor over an on-screen keyboard and hitting an action button to select each letter, I suggest you get the fuck out of here and back to your console.
Come on, the video shows them playing Mario Kart 64. You want an analogue stick to play that.
There's no consumer benefit to not having games on your local machine.
I think there's a pretty strong consumer benefit in being able to turn on a machine and be playing an arbitrary high-end game seconds later. No buying physical media, no waiting for an 5GB download - get straight in and play, and if you don't like it, get out just as fast to try something else.
OnLive was big on demo levels, I'm sure Gaikai will be the same. If it works, this is surely a great way to try out demos. Then, once you're in, you don't have to do anything else except pay, to continue playing.
The article is illuminating in other ways. OnLive knows that if the broadband industry goes in a certain direction, they're screwed. But they feel that other factors will drive consumers to push the industry in the other direction.
In this case, they will, hardware requirements will have direct correlation with price of subscription and/or profits. Expect very aggresive optimisations.
It depends. One view is that hardware is cheaper than programming. How many man-hours are you willing to spend saving a few CPU cycles or a few MB of RAM?
On the other hand, when you scale up to cloud numbers, it's not a few MB, it's that number multiplied by however many thousand servers you have.
If you interview for the appropriate part of Google, they want you to know about seriously low level code omptimisation. If you can knock 1% off execution time, that maps directly to hardware savings for Google. If a service like Gaikai has sufficient users, they'll be the same.
The ISP's are already bitching about gigging the content providers as well as their customers over bandwidth use. Do you honestly think they're going to go for this sort of abuse?:-D
I do agree that the current standards of domestic broadband mean that widespread adoption of this can't work. Intuitively, it feels as if were I the only person on the street doing this, it'll work. If I'm contending with a few others, it's going to break down.
However, 8 years ago (which doesn't feel that long ago to me!) I was on a 56Kb modem, and we have to assume things will continue to progress.
There's a chicken and egg situation - high bandwidth applications won't work perfectly until ISPs deliver good enough networks. ISPs won't work perfectly until consumers start demanding that high bandwidth applications work well.
But that's changing. Mainstream consumers want YouTube to work acceptably in HD. They want to download HD movies to their Xbox quickly. As long as there's competition between ISPs, and consumers who want something better, networks should get better.
All well thought out, and we'll just have to wait and see. It shouldn't be long until we start hearing what the beta testers think.
One thing I'd like to mention is that the bandwidth rates they like to mention are indeed peaks, with much smaller use of bandwidth between them. Of course, if your ISP can't handle those bursts, there's a problem. But I saw claims from OnLive that two people could play on the same DSL connection reasonably successfully, because their bursts would be unlikely to coincide.
However, pissing off Nintendo would be really stupid. My theory is that either they've cleared it, or that Mario Kart won't be available to the public.
Wait a second here. That sounds like a monopoly to me... I realize that there are similar products competing for a particular game, but if there is only one way you can get access to the game, and one company controls the access, that's a monopoly.
You're stretching the definition of monopoly a bit there. Currently the only way you can play Killzone is to buy a copy from Sony and play it on a Sony PS3. Does that mean Sony has a monopoly on Killzone? I suppose so, but it's not something that worries regulators, because Sony does *not* have a monopoly on first person shooting games.
That model is bound to extend to the streaming services. Some titles will be a Gaikai exclusive, some will be an OnLive exclusive, some will be available on both services, and some on neither. Plus they're likely to have more competition if the concept works out.
I was mostly agreeing with you - a service like this does NOT break competition.
I do wonder whether in terms of profit ($$$ - revenue minus outgoings) a cheap iPhone game would beat a big budget PS3 game.
The point is that you can compete on aspects other than game quality. Price is an important factor. Some people won't spend $60 on the best racing game on the market, and would happily spend $30 on a game with poorer reviews.
I think we're agreeing that there's room in a world of streaming games for healthy competition which drives prices down and quality up.
No game customization, modding. This also affects the community around games.
There's not really a technical barrier to this. If there's a consumer demand for moddable games via a streaming service, the game that provides it will succeed in the market.
Examples of ways it could work:
- Fully featured editing tools accessible through the streaming service
- Editing tools to run locally, upload the results to the service where others can play them.
- etc.
MJPEG is literally a JPEG encoded frame, followed by another JPEG encoded frame, etcetera. That's why MJPEG files are so big.
The codecs used to stream video use keyframes and deltas. So you start with a full frame, then for the next frame you transmit as little information as possible, such that the client can approximate the next frame. So that could be as simple as "no change" (for a still image), or "copy the rectangle with these coordinates to these new coordinates" (for a panning scene). Every now and again, things are reset, by including a new full frame. Try encoding a video with the compression rates cranked right up some time. On playback, the nature of those deltas becomes pretty obvious.
The processing involved in working out those deltas is complex and processor intensive. The processing involved in applying the delta and displaying the next frame is pretty simple. That's why you can go and have a cup of tea while iMovie exports your 2 minute video clip, yet a mobile phone can play the result instantly.
An encoder that can look ahead several frames can make better guesses about what kind of delta is going to be most useful, and in most applications (producing a DVD, broadcasting digitial TV etc.) since there's no reason not to work that way, that's what's done. It makes for better quality in less bandwidth.
When latency is unacceptable, the encoder can't work with "future" frames. So it must calculate the delta using only the current frame, the previous one and any other context it's kept. It just means more hard work for the encoder, and possibly a poorer quality to bandwidth ratio. You need a codec designed for that application, and that's what's being done.
In a previous/. discussion, someone with experience in live TV broadcast said that 1ms encoders were in common usage in the TV industry. OnLive boast about their 1ms hardware encoders -- suggesting that it's not an easy feat but they did it. Gaikai claim that with their approach they can get sufficiently low lag without custom hardware.
Anyone saying "it can't be done" is essentially accusing them of blatant lying. I can't see why lying like that would benefit them (unless they anticipate a buyout before the lie is exposed...)
Yes, the unique nature of each games means that they are not totally interchangable, but within a genre the best quality game usually is the one that gets the most sales.
All other things being equal, yes. But price and availability come into play. I wouldn't be surprised if "Cro-Mag Rally" ($1.99 on the iPhone store) sold more units than FUEL.
Streaming videos or games will not work as long as these caps are there. And seeing how my ISP also delivers us video on demand (which doesn't count towards my download limit) I really can't see them eager to change this.
I suspect that services like Gaikai and OnLive will be eager to partner with ISPs - so you won't be too many hops from their servers. In a deal like that, I'd expect the streaming game packets not to count towards the limits. I guess we'll see how it pans out.
Akamai is another company that puts servers at ISP so that they're close to the client. They do it in order to cache web content and make sites feel faster. A lot of ISPs actually pay Akamai to put their servers on their sites, because the biggest cost to the ISP is not transferring data between itself and its customer, but transferring data from the ISP to and from the wider Internet. So you can see that ISPs might be keen to host OnLive/Gaikai/etc. servers.
Slashdot Mirror
OK, how do I know, at point of purchase, whether they reach full brightness in an acceptable time?
I've bought branded CFLs in the last few months that leave the room feeling dim for a couple of minutes.
I have only CFL's in my house.
Do you spend the first two minutes after entering a room and turning on the lights, blundering around and bumping into things?
Oh well, when my incandescents blow, I'll be in the same boat, thanks to EU law.
Guess what... most folks still used Office instead.
Not in my department. How on earth did "most folks" get an Office license from the IBM beancounters?
Not long ago, IBM's standard word processor was Lotus WordPro.
I have a load of .LWP files lying around from my IBM days, that I can't read...
It goes to show that a company like IBM can function using a "minority" office suite.
Say you have a web based service with 2000 paying subscribers, and the servers are creaking a bit under the load. Do you spend a month optimising code, or drop another server into the cluster?
Of course there's a tipping point, which is why the post you replied to says "it depends".
(Similarly, you don't spend hours striving to make a data structure one byte smaller -- unless that data structure is repeated millions of times at runtime.)
Those are good examples of titles for which I can't think of a direct clone (although I bet if you trawl through the Wii shovelware catalogue, you'll find a shoddy Animal Crossing clone).
But the key thing is that there's nothing stopping a developer from making a direct competitor to these games. The fact that they haven't done so simply suggest that nobody believes strongly enough in a market for such a game.
Unless you typed that post by moving a cursor over an on-screen keyboard and hitting an action button to select each letter, I suggest you get the fuck out of here and back to your console.
Come on, the video shows them playing Mario Kart 64. You want an analogue stick to play that.
There's no consumer benefit to not having games on your local machine.
I think there's a pretty strong consumer benefit in being able to turn on a machine and be playing an arbitrary high-end game seconds later. No buying physical media, no waiting for an 5GB download - get straight in and play, and if you don't like it, get out just as fast to try something else.
OnLive was big on demo levels, I'm sure Gaikai will be the same. If it works, this is surely a great way to try out demos. Then, once you're in, you don't have to do anything else except pay, to continue playing.
I smell an a la carte pricing model!
Ya pays yer money, ya takes yer choice.
Sounds like market failure, and I hope for your sake it's temporary.
It's begging for a competitor to steam in in swoop up a lot of custom. Perhaps some sort of wireless broadband...
http://multiplayerblog.mtv.com/2009/04/10/sony-ms-onlive-weigh-in-on-tiered-internet-pricing/ .. in which the OnLive CEO suggests that HD would use 950MB/hour (= 2.1Mb/s), SD would use 350MB/hour (= 0.8Mb/s)
The article is illuminating in other ways. OnLive knows that if the broadband industry goes in a certain direction, they're screwed. But they feel that other factors will drive consumers to push the industry in the other direction.
In this case, they will, hardware requirements will have direct correlation with price of subscription and/or profits. Expect very aggresive optimisations.
It depends. One view is that hardware is cheaper than programming. How many man-hours are you willing to spend saving a few CPU cycles or a few MB of RAM?
On the other hand, when you scale up to cloud numbers, it's not a few MB, it's that number multiplied by however many thousand servers you have.
If you interview for the appropriate part of Google, they want you to know about seriously low level code omptimisation. If you can knock 1% off execution time, that maps directly to hardware savings for Google. If a service like Gaikai has sufficient users, they'll be the same.
The ISP's are already bitching about gigging the content providers as well as their customers over bandwidth use. Do you honestly think they're going to go for this sort of abuse? :-D
I do agree that the current standards of domestic broadband mean that widespread adoption of this can't work. Intuitively, it feels as if were I the only person on the street doing this, it'll work. If I'm contending with a few others, it's going to break down.
However, 8 years ago (which doesn't feel that long ago to me!) I was on a 56Kb modem, and we have to assume things will continue to progress.
There's a chicken and egg situation - high bandwidth applications won't work perfectly until ISPs deliver good enough networks. ISPs won't work perfectly until consumers start demanding that high bandwidth applications work well.
But that's changing. Mainstream consumers want YouTube to work acceptably in HD. They want to download HD movies to their Xbox quickly. As long as there's competition between ISPs, and consumers who want something better, networks should get better.
All well thought out, and we'll just have to wait and see. It shouldn't be long until we start hearing what the beta testers think.
One thing I'd like to mention is that the bandwidth rates they like to mention are indeed peaks, with much smaller use of bandwidth between them. Of course, if your ISP can't handle those bursts, there's a problem. But I saw claims from OnLive that two people could play on the same DSL connection reasonably successfully, because their bursts would be unlikely to coincide.
That would affect your users/Mb/s calculations.
Interesting analogy, and you do have a point.
However, pissing off Nintendo would be really stupid. My theory is that either they've cleared it, or that Mario Kart won't be available to the public.
Wait a second here. That sounds like a monopoly to me... I realize that there are similar products competing for a particular game, but if there is only one way you can get access to the game, and one company controls the access, that's a monopoly.
You're stretching the definition of monopoly a bit there. Currently the only way you can play Killzone is to buy a copy from Sony and play it on a Sony PS3. Does that mean Sony has a monopoly on Killzone? I suppose so, but it's not something that worries regulators, because Sony does *not* have a monopoly on first person shooting games.
That model is bound to extend to the streaming services. Some titles will be a Gaikai exclusive, some will be an OnLive exclusive, some will be available on both services, and some on neither. Plus they're likely to have more competition if the concept works out.
I was mostly agreeing with you - a service like this does NOT break competition.
I do wonder whether in terms of profit ($$$ - revenue minus outgoings) a cheap iPhone game would beat a big budget PS3 game.
The point is that you can compete on aspects other than game quality. Price is an important factor. Some people won't spend $60 on the best racing game on the market, and would happily spend $30 on a game with poorer reviews.
I think we're agreeing that there's room in a world of streaming games for healthy competition which drives prices down and quality up.
No game customization, modding. This also affects the community around games.
There's not really a technical barrier to this. If there's a consumer demand for moddable games via a streaming service, the game that provides it will succeed in the market.
Examples of ways it could work:
- Fully featured editing tools accessible through the streaming service
- Editing tools to run locally, upload the results to the service where others can play them.
- etc.
Do they have competition?
I was nodding my head until you said MJPEG.
MJPEG is literally a JPEG encoded frame, followed by another JPEG encoded frame, etcetera. That's why MJPEG files are so big.
The codecs used to stream video use keyframes and deltas. So you start with a full frame, then for the next frame you transmit as little information as possible, such that the client can approximate the next frame. So that could be as simple as "no change" (for a still image), or "copy the rectangle with these coordinates to these new coordinates" (for a panning scene). Every now and again, things are reset, by including a new full frame. Try encoding a video with the compression rates cranked right up some time. On playback, the nature of those deltas becomes pretty obvious.
The processing involved in working out those deltas is complex and processor intensive. The processing involved in applying the delta and displaying the next frame is pretty simple. That's why you can go and have a cup of tea while iMovie exports your 2 minute video clip, yet a mobile phone can play the result instantly.
An encoder that can look ahead several frames can make better guesses about what kind of delta is going to be most useful, and in most applications (producing a DVD, broadcasting digitial TV etc.) since there's no reason not to work that way, that's what's done. It makes for better quality in less bandwidth.
When latency is unacceptable, the encoder can't work with "future" frames. So it must calculate the delta using only the current frame, the previous one and any other context it's kept. It just means more hard work for the encoder, and possibly a poorer quality to bandwidth ratio. You need a codec designed for that application, and that's what's being done.
In a previous /. discussion, someone with experience in live TV broadcast said that 1ms encoders were in common usage in the TV industry.
OnLive boast about their 1ms hardware encoders -- suggesting that it's not an easy feat but they did it.
Gaikai claim that with their approach they can get sufficiently low lag without custom hardware.
Anyone saying "it can't be done" is essentially accusing them of blatant lying. I can't see why lying like that would benefit them (unless they anticipate a buyout before the lie is exposed...)
Yes, the unique nature of each games means that they are not totally interchangable, but within a genre the best quality game usually is the one that gets the most sales.
All other things being equal, yes. But price and availability come into play. I wouldn't be surprised if "Cro-Mag Rally" ($1.99 on the iPhone store) sold more units than FUEL.
If this works it will still suck outside of their area of coverage
If you're outside the area of coverage, you probably won't even be allowed to sign up.
Fairly obviously - you get what they give you.
There's no technical barrier to them offering you addons and mods, so it's the service provider's choice as to whether or not they do.
Likewise rebinding the keyboard layout. That's going to be a feature of the game, not the platform.
Don't be silly. If they're allowing the public to play Mario Kart, they're doing it legally, so a license fee is being paid somehow.
Streaming videos or games will not work as long as these caps are there. And seeing how my ISP also delivers us video on demand (which doesn't count towards my download limit) I really can't see them eager to change this.
I suspect that services like Gaikai and OnLive will be eager to partner with ISPs - so you won't be too many hops from their servers. In a deal like that, I'd expect the streaming game packets not to count towards the limits. I guess we'll see how it pans out.
Akamai is another company that puts servers at ISP so that they're close to the client. They do it in order to cache web content and make sites feel faster. A lot of ISPs actually pay Akamai to put their servers on their sites, because the biggest cost to the ISP is not transferring data between itself and its customer, but transferring data from the ISP to and from the wider Internet. So you can see that ISPs might be keen to host OnLive/Gaikai/etc. servers.