D9P/G94: 9600 GT on 19th of February

Yes, and we've all seen what happened in the past when nV tried to dictate DX compliancy when they had the biggest marketshare. MS 'chose' the underdog...

Nvidia has been on both sides of this fence already, I can only think of one time however when they've been hurt by it, and that was with a much worse performing GPU. The most important aspect of this is that Nvidia is simply much much better at communicating with developers and that will be key. ATi could have 30x the features but if they provide no support in its use it won't matter at all.
 
Thats not enough, not microsoft and they API madeing the differences, DEV's madeing it.
MS just made the API, but when noone interested (like now with dx10.1) than not even MS can do anything.
BTW. i not read any interview where someone from MS say how important DX10.1,silence from MS side tell me that not even MS give any credit to DX10.1.

Only thing what can help in the situations like this when ATi finding 2-3 DEV and work very closely with they, but this can be hard because ATi has significant lower market share and with even great support can be hard to made similar program like NV TWIMTBP.

That´s not true. If we think like that Intel managed everything.
Devs and companys choose how they whant to face the market by their strategies and not others stratagies.
You have a very limited vision of the world.

DX_10.1 will come with SP1 of Vista and securely very DEVS want to make their game more revelant with markting about Vista SP1 + DX_10.1.
But only time will tell.
One thing is true, is that ATI put DX_10.1 in their chips very easly and Nvidia didn´t.

For you is bad, for others is not. If you think that DX_10.1 is patetic, I think that more patetic is Nvidia don´t supporting it when ATI support it.
 
I think CJ is talking about DX9 and the NV3X - R300 situation.

Bingo. ;)

Nvidia has been on both sides of this fence already, I can only think of one time however when they've been hurt by it, and that was with a much worse performing GPU. The most important aspect of this is that Nvidia is simply much much better at communicating with developers and that will be key. ATi could have 30x the features but if they provide no support in its use it won't matter at all.

Yes, they've been on both sides of the fence. And yes, they had a much worse performing GPU. But one of the reasons that particular GPU series performed much worse was because NV got locked out by MS when they were finalizing the DX9 specs. So NV had to partly "guess" at what was really going to be in the final specs. Yes, of course they knew most of it, but not all.... What you saw in NV30 was a GPU that sometimes exceeded the DX9 specs but in the end was still underperforming, while ATI made a GPU (R300) which fit like a glove when it came to DX9 games.

So eventhough NV has a lot of power... they just underestimated MS. Having good relationships with developers is nice, but it's not the end all be all.
 
Ye it's pathetic, but it doesn't win the war for ATI if developers don't support it.

Look at SM2.0 .. While ATI had it, Nvidia stuck with SM1.4 until they could release SM3.0. Yes it sucked and Nvidia did hold back development of SM2.0 with developers(not all of course) but when SM3.0 came out, it was just so much better than SM2.0.

So while DX10.1 might be held back(you can only think so until Nvidia DX10.1 cards come out), is it really a train crash. IMO, no. DX10 currently sucks performance wise compared to DX9, so until the newer cards come out(irrelevant if it supports DX10 or DX10.1) and it makes game performance much better in DX10/10.1 the adoption of DX10/10.1 will remain low.

If the new cards are indeed 15%-30% faster than the Ultra, that would be really good imo. If driver support is good, I will change to Vista(DX10/10.1) without any issues.

ATM, XP is my choice of horse.

US
 
So eventhough NV has a lot of power... they just underestimated MS. Having good relationships with developers is nice, but it's not the end all be all.

If we count how many DX9 games released until NV3x reached EOL it was really a win for MS and they DX9 API? for me its not.

Today NV has much bigger power in they hand than the API maker MS.
 
No, they don't. Not corporate power, and certainly not market penetration. Windows is more or less everywhere, nVidia is still trying to get there.

But that's irrelevant. There's no gung-ho scenario where MS and nV go into this gang war with each trying to sway devs and whatever else non-sense. nV won't fuck with MS nor will MS fuck with nV, it's counterproductive to both parties, on one hand, and DX10.1 is hardly a reason for this. Can anyone please enlighten me WRT why is an incremental DX update that brings some tweaks/adds a few features without changing nothing significant, basically, has so many panties tightening up between so many butt-cheeks?Because ATi showed a demo of ping-pong balls being flown around with a blow-drier?

Yes, it's a nice thing to have printed all caps on a box:DX10.1 SUPPORT!It'll also probably get some extra sales. Will it significantly change the outlook of the market?Unlikely, not as long as nV has stellar dev-rel and ATi has...erm...had Humus who now is gone, which makes the above statement become the question:"Does ATi have dev-rel?". Not as long as consoles are what they are.

OTOH, you can have your basic engine in DX10 and add DX10.1 features on top of it easily, unlike how it was going from DX9 to DX10(whoa, huge surprise there, no?:) ). So checkbox features will possibly be there to enjoy. The question is wheter or not performance with em' enabled will be there, and if things until now are indicative of anything it seems it won't, at least with current architectures, so nV won't need to employ censorship, end-users and reviewers will gladly disable them and forget about them:D.
 
Unlikely, not as long as nV has stellar dev-rel and ATi has...erm...had Humus who now is gone, which makes the above statement become the question:"Does ATi have dev-rel?". Not as long as consoles are what they are.
This is a very blinkered view, and quite horribly false. I think some confuse marketing schemes with Dev Rel; they are quite different.
 
This is a very blinkered view, and quite horribly false. I think some confuse marketing schemes with Dev Rel; they are quite different.

Could be. But if marketing schemes=getting your card to work ok with games upon release instead of a few months later through driver tinkering, they may have their uses. Let's see:

-World in Conflict-marketing scheme, DX10 still monsterously slow by comparison, quite bugged upon release
-Hellgate:London-marketing scheme, DX10 still slow, bugs with AA on ATi cards only recently fixed by dev., support for CSAA but none for CFAA.
-The Witcher:bugs upon release with drivers then available, some bugs still remain with Transparency AA
-Crysis:required hotfixed drivers by ATi after release, whilst nV had drivers prior to release, upon release and post-release
-Bioshock:required hotfixed drivers. still significant performance drop with DX10, even with added features disabled, contrary to behaviour observed on nV HW.
-Gears of War:hugely slow upon release in DX10 with or without AA. Later drivers seem to have fixed it to some extent.
-UT3:poor performance with CF, some odd slow-downs even in single-card configurations
-Company of Heroes:slow as a dog in DX10 mode, although it seemed to be quite good in DX9 on ATi HW. Seems to have improved with later drivers and patches, still not up to speed with nV.
-Lost Planet:still bugged in DX10 mode, had a number of issues from release until present day, slower in DX10 when compared to DX9 even with added features disabled, unlike nV HW who actually is a tad faster in the same scenario
-NWN2 and NWN2:MOTB-slow as a dog with maximum settings on ATi HW since release and up to present time, for no obvious reason, even after numerous patches/drivers.

So, that marketing scheme everyone is ragging upon and depicting as the spawn of Satan may have a thing or two down right, at least for the consumer. Because the above list doesn't suggest to me that ATi cards are the primary development target, but that they're at best an afterthought, and that after release you guys have to start tinkering with the drivers to make the best out of an unoptimal solution. Perhaps I'm retarded and am seeing things and all is peachy.
 
I'm farily sure you'll find that the majority of those titles NVIDIA have also been releasing beta drivers for as well as us releasing "Hotfix" drivers; timing may be a little different, but thats likely down to Q/A efforts I suspect.
 
but when SM3.0 came out, it was just so much better than SM2.0.

Exactly how much better was SM3.0 compared to SM2.0?

I vaguely remember some NV slides where they compared Far Cry in "SM2.0" vs SM3.0 mode to show how good and superior SM3.0 was... but didn't those turn out to be SM1.1 vs SM3.0 instead? And couldn't SM2.0(b) hardware produce the exact same image (and iirc did it even faster than competing hardware at the time)?

Anyway, I just hope NV adopts DX10.1 sooner than later. You do realize that the changes in DirectX were put in there because not only MS wanted it, but also because the IHVs and most of all the ISVs asked for it. I've already seen various developers state that they're very happy with the new stuff that DX10.1 brings and that they'll definitely be using this... even on this very own forum...
 
[conspiracy]I find it weird that MS release their DX updates only when Nvidia release their cards. Last time MS did that for ATI was when DX9 was released. Actually after DX9a was released, MS still went on to support SM2.0b in DX9b for Nvidia 5900 series cards. DX9.0c was only released when Nvidia released their SM3.0 cards. DX10, Vista was released when Nvidia released their G80 card, and MS have yet to officially release DX10.1(vista SP1) which will happen sometime early this year or .. that's right when Nvidia release their DX10.1 card.[/conspiracy]

That said, maybe it's just a coincindence thing.

US
 
Ahhh, the falacy of "DX9.0c = SM3.0" still exists I see.

DX9.0c brought HLSL profiles for SM3.0 and SM2.0b (R4xx series). An HLSL update was brought in for SM2.0a for NV3x prior to that as well.
 
Exactly how much better was SM3.0 compared to SM2.0?

Remember the pics of AOE3 just before it came out? It was awesome. Nvidia 6xxx series cards had still to be released and DX9.0c also. So how much better was SM3.0 to SM2.0 .. alot I thought.

Of course I still preferred my X800(SM2.0) until the 8800.

ATI only caught up after they released the X1xx series.

US
 
Ahh .. so it was DX9.0a .. sorry then and thx for rectifying my mistake Dave.

Is there a timline anywhere of which DX(8,9,10) was released when etc.?

US
 
No obvious reason to you, but the root cause was given to the developer some time ago.

The point of my post(s) wasn't, as you or Dave might mistakingly believe, to rag upon your work. I respect what you do and your efforts. I simply don't think that must equate to constant praise, no matter what, and to the belief that all of ATi's choices are the right/good/proper/whatever ones.

That being said, my argument was that nowadays a consistent and intense effort of "getting in bed" with developers seems to pay off big time, and that having good HW is only part of the equation. And that ATi as a whole should try harder in this department. Was the way of expressing it exaggerated?Yes it was, but it was tongue-in-cheek(ish) if anything. I don't doubt that you communicate with devs, what I was saying(hoping) was that in the future there's going to be a serious push on that front, something like what your much hated competitor does by sending coders on site, for example. Because for them it seems to be an investment that has very nice ROI.

This has been a longtime issue with ATi, the presence of features and the non-obvious exploiting of these features. PS1.4. 3Dc. The huge dynamic branching muscle the R5xx series had that wasn't exploited during its stint in GPU primetime. Don't you think it's hurting you guys when upon release of most major new titles everyone is screaming "bloody-murder" because there are some significant bugs?Yes, hotfixes come, but that initial ugly reaction is far more vocal then the "it's fixed now" one. Don't you think it's hurting you guys when most triple A new releases sport a big nV:TWIMTBP splash screen on start-up?

Maybe this part of the thread should be chopped up and sent somewhere else(if there's interest) as it has about jack to do with the 9600GT.
 
Back
Top