Gabe Newell: DirectX 10 for Vista was a mistake

[Deadly Furby]

Gabe says Vista DX10 is a bad idea

Untie, rewind
By Wily Ferret: Monday 27 August 2007, 09:52

GABE NEWELL, the popular director of Valve Software, has berated Microsoft for limiting the uptake of DirectX 10 to Windows Vista owners.

Speaking to Heise, he said that the Vole had made "a terrible mistake" in excluding Windows XP owners from next generation content.

The latest Steam survey, which prompted Newell's comments, show that only two per cent of Steam gamers have a DX10 graphics card and Vista.

Newell also highlighted the cross-platform problem - the fact that neither the Xbox 360 nor the Playstation 3 use DirectX 10 features. This would limit DX10 support given the financial requirement for console versions of games, he explained.

At this point, it's fair to say that nobody except for Microsoft thinks that bundling DX10 solely with Vista has been a good thing for the PC gaming industry - the general consensus amongst both developers and hardware manufacturers seem to be that it fractures an already-diverse PC gaming market even more, making consoles look even more attractive. One only has to look at Bioshock - the PC version has staggering requirements, barely any image quality upgrade under DX10/Vista, as well as installation issues - whereas the Xbox 360 version plugs in and works. Food for thought.

[-=bingo-bango=-]

yup

[mervaka]

its rare they do get things like this wrong, but in this case its one of the dumbest things ms ever did tbh...

[TurboWolf]

Tbh.

The guy talks shit, 2 years ago he was dooms daying duel cores and programming for it, he does make good games though

[Defrag]

He didn't doomsday multi-core and parallelism. He said that going multi-core meant things have to change a lot and that people get ahead of themselves when it comes to the ramifications of multi-core's impact.

A lot of the stuff he said pertained to the fact that throwing a load of processors at a problem won't make it any faster if the architecture is tough to use and shitty to develop for. Turned out he was right about that.

Look at the PS3 right now -- developers are having a much harder time to develop for it compared to the PC & 360.

When I was at the game in Scotland development fair thingy (they let student scummers in, too), several developers expressed disappointment with the PS3's architecture & immature tools. This included a company that worked on simultaneous 360 & PS3 ports of the same game. The PS3 version was significantly harder to make. Look at UE3, too. The word on the street is that the developers are having a lot of trouble making it run satisfactorily on PS3 and that Sony are having to airdrop in a crack team of developers to help them optimise it.

He perhaps overstated the 'throw away your existing code' stuff (middleware & specialised compilers now exist to help with this) but he was right about how tough it is for software engineers to design efficient software for multi-core CPUs compared to single-core.

[Ihmhi]

Okay, so... PS3 is hard to develop for. Why is this again? What are the benefits from all of this difficult developing?

I mean, look at mapping in TFC versus FF. HL, it was not too hard to get the map code working. FF, it is a bit more challenging thanks to Lua, but ultimately you can do so much more with it. So where is the benefit to the PS3's developing difficulties?

[summit]

There is theoretically more power behind the PS3, but unless the game happens to be an exclusive, there's a great chance the extra time won't be taken to take advantage of it. Given the ease of development for the 360, it's often the lead platform, while PS3 receives what it needs to keep up - nothing more.

I think theoretically these things sound so great, but realistically they end off being horrible. We all know DX10/Vista will eventually be the standard by which we all have to adhere, but it won't be for a while.

It's also unfortunate that the new DX cannot run on XP (different architectures make this impossible, apparently) I'm sure while MS don't mind people forking out extra for the OS, it doesn't help the situation any better.

[Hawk Eye]

I thought the Microsoft Software company wasn't allowed to build software only for the Microsoft Operation System company... or did my mind just blank and their anti-monopoly court ruling is no longer in effect? [1999]

[Ihmhi]

Aha, so the higher end architecture allows for much better game design, but much more time must be invested in it. Looks like Sony sacrificed current profits towards hoping that future development will pick up.

[Defrag]

What he (summit) said. Back in the day, it was common to use loads of techniques and little technical tricks to squeeze every last ounce out of a format. It was sometimes a necessity if you wanted your game to run properly or even be possible.

Nowadays given the choice between increased performance and ease of development, most developers will choose ease of development. Why? 'cause if it's easy to develop for a platform then it's not like you're back to the days of scrimping everywhere and having to devise elaborate solutions or esoteric tweaking to get the job done. Today's PCs & consoles are very fast.

To give you an idea of how much you can get done with a console that's not running close to its limits: A lot of early titles for the PS2 didn't make use of the vector units (one of its main computational parts!) I.e. back then they still got stuff done because, for their games, the PS2 sans vector units was still powerful enough to do the job. As time passed, more and more people got to grips with them and made even more computationally demanding games.

So... if you have the choice between developing for a platform with a new & hard to grasp architecture, immature tools etc. and a platform that will get the job done and has a fair old amount of grunt, plus better support and an architecture that is more familiar ... people will generally plump for the latter. If you have a console that can theoretically perform 10 gajillion calculations, but it needs a lot of experience & effort to get anywhere near that even half of the time, but you can have 6 gajillion on a competing console nearly all of the time (without the hoop-jumping), it's a no-brainer. 6 gajillion is nearly always plenty (or at least it is right now!)

Once the PS3 matures a bit and people get the hang of it, it might come into its own and be able to do things the 360 can't handle. Right now, Sony's design decisions aren't looking too clever, though. I'm not sure why Sony went the route it did, but... they did. The fact that UE3 isn't running satisfactorily on the PS3 (while it's fine on the 360) again suggests that Sony underestimated the reality of what'd be like to develop for the PS3.

I'd normally treat a lot of what is being said with scepticism, but I've heard numerous tales of multiple format games being delayed by the PS3 development, plus all the UE3 talk and that games talk thing, too. Has anyone heard anyone singing the PS3's praises yet?

[Sidd]

Folding@Home

They love the ps3

[Leandros]

Gabe also generally thinks DX10 is a mistake.

GI: Do you think you’ll ramp source up to take advantage of DX10 or do you think you’ll stay with DX9.

Newell: Right now with the flexibility of DX9, we can take advantage of DX10 hardware functionality through DX9. There are far more customers with DX10 hardware running on Windows XP than Windows Vista. If you’re going to try to take advantage of that hardware, your customers are telling you, make sure it works on DX9 API. For example, the tesselator in ATI’s hardware is super valuable and we can get to it through DX9, so that’s probably a better investment than going after a DX10 API.

http://www.gameinformer.com/News/Sto...430.htm?Page=3

[Jetpac]

In the end games will move on to DX 10.
and after that they will most likely move to DX11.
and im guessing after that probably DX12.. i know i know CRAZY assumption.

Moves take time.
People need time to adapt.

give it a little bit and i guarantee there will be more people on vista.
granted DX10 vista only may have been a mistake, but however since noone is making games that really utilise DX10 Properly who gives a shit at the moment. and by the time they do start using DX10 to its potential Vista will be well on its way to having more users and there will be more people with dx10 cards.

[Defrag]

Originally Posted by Sidd

Folding@Home

They love the ps3

OK, but ... games stuff.

[Dospac]

The GPU in the 360 is faster hands down. The PS3 will never catch up in the fancy high res gfx dept. It is a good BR player and those folding guys LOVE it..

[gizmofo]

Originally Posted by Jetpac

give it a little bit and i guarantee there will be more people on vista.

Not many more. MS has already announced Windows 7 (Note the sex appeal of the name), and many big names are skeptical about vista. I think Vista may just turn out to be a building block on the next OS. They tried out some new stuff, tested out how the market shifted, and found out the (bad) results.
GL MS.

It sounded like dooms daying to me lol As for the pants gfx of the PS3 compared with the 360, GT5 (prologue?) looks kickass

[Jetpac]

Originally Posted by gizmofo

Not many more. MS has already announced Windows 7 (Note the sex appeal of the name), and many big names are skeptical about vista. I think Vista may just turn out to be a building block on the next OS. They tried out some new stuff, tested out how the market shifted, and found out the (bad) results.
GL MS.

Vista is a very good OS.
however it has receieved bad press because of original incompatibilities with some hard and sotware.

so that has turned alot of people off before they have even tried it. I was one of those people, i tried it and i have converted over to windows vista with no problems whatso ever.

its a pitty gossipmongering on the internet really damaged it.

[caesium]

Originally Posted by Dospac

The GPU in the 360 is faster hands down. The PS3 will never catch up in the fancy high res gfx dept. It is a good BR player and those folding guys LOVE it..

Not really. The 360 is pretty much a conventional pc, and so developers can easily make good use of its hardware right now (this is the major advantage of the 360). It has a top end gfx card (for its time), and a reasonable amount of cpu power. The ps3 is not a conventional pc, and so as ppl have said it is much harder to make good use of it's hardware. It has a much slower gfx card, but makes up for this with a lot of highly customisable cpu power which developers can use alongside the gpu to do exactly what they want in a game.

Folding@Home aside lol, what this all means is that if you simply port a pc/360 game to the ps3 then it will not make good use of its hardware at all, and will almost certainly look worse and run slower than on the 360. However, if a developer learns to make use of the ps3's hardware then they can do a lot more and have considerably more power at their disposal than on the 360. The problem is that they effectively have to make use of it themselves. So imo the big question for the ps3 is: in this deadline/profit based world of today's gaming industry, can and will enough developers learn to make good enough use of the ps3's hardware in order to produce consistently btr gfx than the 360 and hence justify the ps3's higher pricetag to more gamers?

The first batch of games that developers have had the time/know-how to even begin to optomise to a satisfactory degree for the ps3 are coming out this autumn, and judging by the few that I've seen so far (and comparisons to the 360 versions) the answer is a definite yes.