Gabe Newell: Valve will release its own console-like PC

If you release a box that has a set hardware (or combinations) you can do a lot more work to validate everything upfront, guaranteeing quality, speed and reliability.
I suspect MacOSX is that stable because its ecosystem is under tight scrunity. (Not sure about the vocabulary here.)
 
Because old drivers will not work on new hardware.

Think of it this way, steam keeps all drivers installed on the machine, starting with a baseline one that it ships with and any future ones that games need. When a game runs they request the driver they need (ie, been tested with) and that's what steam loads and then executes the game. That's how it would work on the current steam box. On steam box 2 you would do the same thing, except if they want to provide a console like experience then what they have to do is test the games on the new machine with its new baseline driver and see which ones work fine. For the ones that do, they are allowed to run on steam box 2 without update. Basically when a steam box 1 game is run on steam box 2, steam would see it's requesting an old driver on old hardware which it no longer has, so it checks it's compatibility list, sees that it's been tested and runs fine on the new baseline driver for steam box 2, and then automatically runs it with that. For games that don't run properly they would have to let the publishers know before steam box 2 got release so that the game could be patched to run properly on steam box 2 before it's release.

That's the only way they can give customers console like stability, and backward compatibility in the process. If they still go with a one driver for all system, then it will be no different than a pc and they have no way of guaranteeing that all a users games will keep working. They do this already with the VC runtime, games use the one they are tested with and a pc usually has a whole slew of these installed by games. They need to extend that idea to drivers if they want this machine to run like a console. Otherwise there is no way to predict if new drivers will cause problems on older games. They aren't going after users like me who will stick to a pc for maximum graphics fidelity, they are going after console users and I can't imagine the typical console gamer would accept that, where a driver update breaks their old games, that would be suicide. I don't think they can half ass this, if they are just going after a rebranded 'steam' pc then personally I think they are wasting their time.
 
Think of it this way, steam keeps all drivers installed on the machine, starting with a baseline one that it ships with and any future ones that games need. When a game runs they request the driver they need (ie, been tested with) and that's what steam loads and then executes the game. That's how it would work on the current steam box. On steam box 2 you would do the same thing, except if they want to provide a console like experience then what they have to do is test the games on the new machine with its new baseline driver and see which ones work fine. For the ones that do, they are allowed to run on steam box 2 without update. Basically when a steam box 1 game is run on steam box 2, steam would see it's requesting an old driver on old hardware which it no longer has, so it checks it's compatibility list, sees that it's been tested and runs fine on the new baseline driver for steam box 2, and then automatically runs it with that. For games that don't run properly they would have to let the publishers know before steam box 2 got release so that the game could be patched to run properly on steam box 2 before it's release.

That's the only way they can give customers console like stability, and backward compatibility in the process. If they still go with a one driver for all system, then it will be no different than a pc and they have no way of guaranteeing that all a users games will keep working. They do this already with the VC runtime, games use the one they are tested with and a pc usually has a whole slew of these installed by games. They need to extend that idea to drivers if they want this machine to run like a console. Otherwise there is no way to predict if new drivers will cause problems on older games. They aren't going after users like me who will stick to a pc for maximum graphics fidelity, they are going after console users and I can't imagine the typical console gamer would accept that, where a driver update breaks their old games, that would be suicide. I don't think they can half ass this, if they are just going after a rebranded 'steam' pc then personally I think they are wasting their time.


Or you can just keep backwards compatibility in your drivers. Wouldn't that be easier?
 
Or you can just keep backwards compatibility in your drivers. Wouldn't that be easier?

It depends.
For example if driver V1 has a bug, and one or more games are released that depend on the incorrect implementation (yes it happens all the time) then when you release V2 and fix the bug you break the game.
Or if you change undefined behavior A because it increases the speed of the driver, bty games are dependent on the original behavior.

There is a fair amount (I might claim the majority) of the code in a PC driver that does nothing but try and fix broken game code. In some cases it's for performance, in others it's because games are dependent on the incorrect/undefined behavior of older drivers.
That code eats CPU cycles, and is one of the reasons that consoles have less driver overhead.
 
Controllers and tablets. They need controllers. Maybe one with an integrated trackball, but I think that's unlikely. I imagine a split dual-stick controller that extends horizontally and can clamp around a whole host of smartphones (paying me lots of patent royalties!).

Controllers or not, I would sincerely hope they would continue to support mouse and keyboard. Tablet controls would seem unnecessary unless it actually had some validation for a particular game

Most PC games these days have some kind of official controller support, so I don't see any issue except, obviously Valve would need a "made just for our system" controller. I'm sure Logitech would be the first candidate and would gladly provide, but the controller is another addition to cost. As necessary as it would be to noobs, the idea of saving $15 bucks or so by eliminating it is attractive and indubitably the system would support general USB controllers but in the long run, Valve would be better off having everything in one box.

Software and hardware control obviously is a big issue here. With the world moving to a more open source on closed hardware nature, leaving the system open to new software is a must. Keeping the hardware closed isn't a huge issue. Like I addressed earlier, good hardware is profitable in a realistic budget. And the value of a closed, unupgradeable system isn't too much of an issue assuming the hardware is up to snuff for a decent period (two to three years) and the software provides enough value to the consumer while being migratory to a new system in the future.
 
I think the article mentioned it would come with some form of gamepad/controller, and I imagine it would work on the PC as well. I'd be very surprised if they didn't allow keyboard and mouse control, in fact I think it's more likely that I'm struck by lightning twice.
 
I think the article mentioned it would come with some form of gamepad/controller, and I imagine it would work on the PC as well. I'd be very surprised if they didn't allow keyboard and mouse control, in fact I think it's more likely that I'm struck by lightning twice.

You're absolutely right they would. Not all steam games are made for controllers. Some like RTS games NEED a mouse and keyboard. So not supporting KB/M would be silly for a SteamBox
 
It depends.
For example if driver V1 has a bug, and one or more games are released that depend on the incorrect implementation (yes it happens all the time) then when you release V2 and fix the bug you break the game.
Or if you change undefined behavior A because it increases the speed of the driver, bty games are dependent on the original behavior.

There is a fair amount (I might claim the majority) of the code in a PC driver that does nothing but try and fix broken game code. In some cases it's for performance, in others it's because games are dependent on the incorrect/undefined behavior of older drivers.
That code eats CPU cycles, and is one of the reasons that consoles have less driver overhead.

Is this common on android games? Of IOS games?
 
Isn't this completely the wrong time in the console cycle to introduce a 'Steam Box'?
Sony and Microsoft are just about to release consoles with the latest hardware, and are willing to sell at a loss, so how could valve compete with performance per £ where they, or whoever else builds the boxes, has to make a profit? I'm assuming they'll have to make a profit on the hardware because I can't see steam, in the early years at least, bringing in enough profit.

Surely a couple of years from now is the best time to release where hardware has moved on, costs dropped and with possibly appreciable graphic improvements.
 
If it sells a 100K units it's still an extra 100K PCs out there more likely to buy games using Steam ... any level of success is sufficient, if in the future they can do it more successfully then that remains an option. They don't have to retain a single platform for half a decade.
 
Is this common on android games? Of IOS games?

A friend of mine does a lot of IOS development and bug fixes in IOS6 apparently broke a lot of apps.
Apples policy appears to be it's the developers issue, and they just expect the developer to patch the app.
Since it's trivial for debs to do and most apps make money over long periods it seems to get done.
 
A friend of mine does a lot of IOS development and bug fixes in IOS6 apparently broke a lot of apps.
Apples policy appears to be it's the developers issue, and they just expect the developer to patch the app.
Since it's trivial for debs to do and most apps make money over long periods it seems to get done.

Seems like a valid business strategy then!
 
A friend of mine does a lot of IOS development and bug fixes in IOS6 apparently broke a lot of apps
yeah there were a couple Ive encounted, eg their pow() function gave wrong results.
where I was relying on the wrong shader behaviour, but theyre easy to fix.
unfortunately IOS6 has made some new bugs, which me & others have complained to apple about (Ive sent test app etc) but theyre pretty slack compared to other hardware/software guys

Isn't this completely the wrong time in the console cycle to introduce a 'Steam Box'?
yes thats true, a couple of years ago or in a couple years would be right, then again MSONY mightnt release new consoles until 2014 (as it looks like the wiiU aint gonna be a threat) and theyre making money of the current ones
 
yeah there were a couple Ive encounted, eg their pow() function gave wrong results.
where I was relying on the wrong shader behaviour, but theyre easy to fix.
unfortunately IOS6 has made some new bugs, which me & others have complained to apple about (Ive sent test app etc) but theyre pretty slack compared to other hardware/software guys
There seem to be plenty of issues with Tegra 2 on Android that no-one's fixing.
 
There seem to be plenty of issues with Tegra 2 on Android that no-one's fixing.
there prolly are but Ive never programmed for android. I meant windows with nvidia/ati

WRT apple even though all the hardwares similar
Ive seen errors thats not there with SGX535 but are with SGX543 etc

hopefully this valvecon will be exactly the same hardware in all (like a console) and not an upgrade every year or so (like iphones etc, not to mention ipads/ipods :()
 
Would it matter that much if that was the case? I can't remember the last time any decent game didn't run on amd/nvidia hardware with the exception of rage but that only worked like ass on amd because Carmack was being a twat.

Generally speaking stuff just works today. Not saying there aren't any bugs but generally speaking it's nothing game breaking. If it's a windows box driver updates will be there anyway so I wonder if sticking with the same hardware for a longer period of time with no/less bugs outweighs configurations that are updated on a more frequents basis.
 
hopefully this valvecon will be exactly the same hardware in all (like a console) and not an upgrade every year or so (like iphones etc, not to mention ipads/ipods :()

An A10 APU w/ A55 chipset based machine would be quick, cheap and easy to produce with off the shelf parts. It would require minimal research and development, while being capable of running just about any of the most intensive PC games at 30 FPS at 720p at minimum specs. It can run Valve titles at 1080p too.

Much like consoles though, it could be a standard for developers to aspire to, and not be willing to develop for higher end systems sans basic "upgraded" features like screen res and AA, so there is a possible issue there too.

However, there is great opportunity for AMD here if Steambox was APU based. If Valve wanted a dedicated GPU, Valve could custom fix the CrossfireX profiles for games released on their own box which could trickle into the PC space. It also provides AMD an opportunity to get developers to embrace GPGPU for PC gaming if the Steambox was APU + GPU.
 
Would it matter that much if that was the case? I can't remember the last time any decent game didn't run on amd/nvidia hardware
part of the reason is cause the ati/nvidia driver writers put execptions into their drivers working around how game X works (*) cause it will work better a different way on hardware X vs hardware Y/Z (which they tested it on when they built the game).
Having fixed hardware puts the onus on the developer to really optimize it more & makes it far more simple as well.

(*) you'll see lists of games where driver version XX.X.X improves crysis 3 framerates by 30% or whatever

I wonder if sticking with the same hardware for a longer period of time with no/less bugs outweighs configurations that are updated on a more frequents basis.
A/ it'll be cheaper for the end user. B/look at the PS360, try running that level of graphics on a PC of the same spec, impossible
 
Isn't this completely the wrong time in the console cycle to introduce a 'Steam Box'?
Sony and Microsoft are just about to release consoles with the latest hardware, and are willing to sell at a loss, so how could valve compete with performance per £ where they, or whoever else builds the boxes, has to make a profit? I'm assuming they'll have to make a profit on the hardware because I can't see steam, in the early years at least, bringing in enough profit.

Surely a couple of years from now is the best time to release where hardware has moved on, costs dropped and with possibly appreciable graphic improvements.

I don't believe they'd try to compete on price or volume. I'd expect them to launch more expensive and without any expectations of unseating MS or Sony. I agree with MfA that any volume helps there business, as long as they don't sell the units at a loss.
 
Back
Top