NVIDIA, MSI, and Lucid speak about Hydra delay

[sammorris]

Expecting nvidia to support PhysX with an ATI card fitted is not exactly part of the issue, but there's a difference between not supporting something and actively stopping it. Back in the days when the nForce chipsets were popular for motherboards, you didn't see nvidia coding out Intel, VIA etc. chipset boards when nvidia graphics cards were installed.
Quite frankly, nvidia is just trying to avoid customer complaints IF and when PhysX turns out to cause problems with an ATI card through a coincidental conflict (I'm no internal graphics guru but it's my belief this is unlikely). If ATI + nvidia can work via Lucid, PhysX with an ATI card installed can sure as hell work, and of course they're trying to garner more sales. Instead what they've done is tick off enthusiasts who want to get the best from what the graphics world has to offer, from either company. Think hard about it, are you that likely to see anyone but enthusiasts put nvidia and ATI cards together in tandem solely for the purpose of running PhysX? nVidia certainly aren't going to advertise that the combination works, so I see no reason why it's something for them to be afraid of everybody doing.
If Lucid get their way though, this is all moot anyway.

[NVIDIA_tap]

@sammorris - Seems like you are already aware that NVIDIA did nothing to impede Lucid - so that is all good. But I thought I might say a word on you comment about NVIDIA as a company.

I don't know if you have met me or seen me, but I go to a lot of LAN events and public speaking events. I am a regular up at PDXLan, and most of the key guys in the gaming hardware business have had some contact with me over the years. I spend a lot of time with our customers here in the US, Asia and in Europe, and I spend the bulk of my freetime gaming (Borderlands is my current addiction). I am also one of the business guys at NVDIA.

We in general love gaming and most of us here are gamers. We spend a ton of money on technology that is only targeted at making PC gaming better. A sampling of the efforts include :
- Huge GPU development teams
- Huge GPU software teams
- Huge QA teams (you would be shocked how complex QA is these days)
- And a very extensive Game development infrastructure (We even have engineers that go and help game studios build better games).

All of this investment results in PC gaming innovation and really cool technologies like we saw when Batman AA came out.

What I find disheartening is that some folks may not see the magnitude of our commitment to gaming. You may not understand as an example that if NVIDIA had not invested Millions and Millions of dollars in the design of PhysX - we would not be arguing over out business practices - which BTW are not as self interested as some have implied.

Your comment related to Lucid and NVIDIA could not be more wrong. If Lucid makes gaming better on PCs and our customers get more excited about gaming because of them (and maybe buy more GPUS) - then we welcome thier contribution. I reamain skeptical and there are serious compatability concerns that I shared in my earlier interview....but If they can overcome that - the all good.

I may not be able to change your view of NVIDIA as a company...But think of us just humans doing the best we can.

[NVIDIA_tap]

@sammorris - This is in response to your 2nd post.

You post seems to imply that the presence of a lucid chip may somehow enable PhysX to run with a non-NVIDIA GPU acting as a render agent. While we have not seen Lucid silicon, I don't expect that to be the case. The Lucid solution as far as I understand it will not change our poicy that NVIDIA cards can only be used for PhysX in syste,s where the rendering is done on an NVIDIA GPU.

Also I think you may not yet appreciate how significant quality is to NVIDIA. We clearly sell products to a very well educated group of customers that have high expectations from us. We know that when our quality slips we see direct impact on our sales and on our reputation. In the case of PhysX we felt that the risk was high enough to take action to avoid future issues. It was not a trivial decision for us so hopefully that gives you some sense of how complex this interaction can be.

We are not taking any action on Lucid because we think the QA burden falls clearly on them. If an ODM or OEM uses ther silicon to build a board that does not work correctly, we belive it is clearly thier isue and will not impact stongly on NVIDIA. PhysX in contrast is not that way. Clearly AMD is not going to QA the AMD+NV physx solutions. So issues tere will inevitably put the onus on us.

[sammorris]

The Lucid situation does not change a thing in that regard, but, however rudimentary it may be, the no-PhysX coding is so simple (because it's so unnecessary) it's already been successfully hacked out of the drivers. How effectively is yet to be seen, but it's been done, simply because there is no reason (yet) for PhysX not to work when the primary renderer is not an nvidia GPU.
I fully intend to make use of a PhysX nvidia GPU on a Lucid platform when I fancy PhysX (and don't need the second card to render graphics, or hey, maybe even stick one in the third PCIe slot) and intend to employ this little work-around. nVidia I'm sure will update their drivers to stop this, but if you're in touch with the gaming community at all you'll know that the underground community that's out there will relish the challenge of doing it all over again.

I will reserve my comments about nvidia's quality, they aren't relevant here, but suffice to say I do not entirely agree.

[CAT-THE-FIFTH]

All this talk of multi cards setups and high end cards is very nice but what about an indication of when Nvidia intend to release its sub £200 DX11 cards??

Are we going to see them in stores in the next three months??

[sammorris]

I don't think even an nvidia rep can really explain that due to non-disclosal arrangements etc, but the pace of the market is almost always top-end single first, midrange single later on. Last I heard the GTX300 top-end card will be due anywhere between late January and late April, if we take the middle of that, early march, then it seems unlikely a midrange version of GTX300 will be on shelves until May at the earlies, but that's just a guess. Who knows, there could be an HD5890 by then.

[NVIDIA_tap]

I do expect the community to work to re-enable PhysX. And I as a long term engineer myself I know that it is not overly complex to do so. I say go for it! The reason I am pretty open to this kind of thing is easy to explain. If someone is willing to do a hack to turn something on then they pretty much know that it is unsupported. If things don't work after that then I don't see that negatively impacting NVDIA.

All I can say is that overtime I expect it to be less and less likely that the hacks will work as our technology becomes more integrated between graphics and physx. Games will start taking advanatge of new features and it is likely that someday soon hacked solutions just won't work. In the meantime though I say have at it.

One point I do want to make though is that I don't condone hacks that take money out of the pockets of folks doing work to innovate in the gaming areana. As I have matured I realize that it's all connected. If folks pirate material (games, music, windows, whatever) then that reduces the inventive to invest. And we do want that to happen- especially for PC gaming.

[NVIDIA_tap]

@CAT-THE-FIFTH - Sadly I will get whacked if I leak any infor about future card plans. All I can say is OMG - it is so fricken awesome.

[sammorris]

I'm happy with the whole PhysX being disabled thing, as long as nvidia leave it at that, so it can be re-enabled without too much fuss (though to be fair, a hidden away, but official advanced config menu would be preferable). I just have my fears that as more and more drivers come out, nvidia will end up working harder to stop people from unlocking PhysX on other hardware. That's as good a test of nvidia's corporate attitude as any, in my opinion. Leave the enthusiasts alone and it's all good, make life even harder for them though, and it's the past all over again.

Once again, Piracy is a matter for another thread, but in all honesty, I think there's a bigger threat to PC gaming revenue than that.

[CAT-THE-FIFTH]

@CAT-THE-FIFTH - Sadly I will get whacked if I leak any infor about future card plans. All I can say is OMG - it is so fricken awesome.

But will it run Crysis.......2??

[NVIDIA_tap]

@sammorris - I can't agree to stop improving our security on enabling technologies. We use the same mechanisms on a bunch of different enabalements. We will from time to time change the way protections work...which will send the community back to the keyboards to figure it out again. Isn't that the cycle?

It is not badly intentioned- just a fact of life.

But back to Hydra...any other questions?

[NVIDIA_tap]

But will it run Crysis.......2??

You must know that Crysis is still one of the most GPU intensive games out there. And yes...it will run Crysis. of course the devil is in the details.

Lets say today as an example....a GTX 260 will run Crysis at 22 fps @19x12 with everything on. A GTX 285 SLI setup will hit 40fps at the same settings. So our next big thing will get #$* FPS with the same config. As you can see we kick some serious triangle butt.

#$* - redacted to save my job.

[sammorris]

Cat: Ugh, don't go there. I played Crysis on medium with low shaders and no AA on my HD3870. Even now, with 9x the graphics power, I still can't play it on Very High with AA, nowhere near it, and as for Warhead, whoever it said it ran better than the original never tried Enthusiast mode. My CPU isn't even fast enough for that.

Tap: As good enough as a response as I know anyone in your position can give. It is good to see nvidia at least do have some well-meaning guys out there that are willing to speak openly when the flak comes out.
As far as Hydra is concerned, I'm not sure there's a great deal else to discuss other than waiting for some press-time with Lucid again. I don't need any more convincing that nvidia haven't a part to play in the delay (though that was more by results than by discussion). I only hope that 'improving security on enabling technologies' leaves the Hydra platform well enough alone. As a best-chaser, the Hydra is bigger news to me than most.

[sammorris]

One thing I was going to ask, perhaps it's not something you can comment on but are we ever likely to see the GTX275/285/295 again, or is that it?
Since My 'ideal' BigBang setup would see the first GPU as an HD5970 (Here's of course hoping that the Hydra system works with two GPUs integrated together - from a logical software standpoint, I don't see why that wouldn't work with X2 cards, can't speak for nvidia's offerings as I think they still have an SLI enable/disable tickbox whereas the ATI cards are somewhat stealthed) in the 16x bandwidth slot, and two smaller cards in the 8x ones, currently thinking maybe 5870 and a GTX260, but should a cheap GTX285 turn up, it would come in handy for those games that really do only run on nvidia hardware.

[CAT-THE-FIFTH]

You must know that Crysis is still one of the most GPU intensive games out there. And yes...it will run Crysis. of course the devil is in the details.

Lets say today as an example....a GTX 260 will run Crysis at 22 fps @19x12 with everything on. A GTX 285 SLI setup will hit 40fps at the same settings. So our next big thing will get #$* FPS with the same config. As you can see we kick some serious triangle butt.

#$* - redacted to save my job.

I do hope Crytek do a better job of making Crysis 2 run better. The last two levels of Crysis ran like a blooming arthritic dog with back pains when it was first released.

[sammorris]

I remember noting that the same settings that got me 55-70fps in the first level on high shaders, got me 35-40fps on the last level with low shaders, which dropped to 18-19 when the smaller of the two final bosses fired its laser.