R300 the fastest for DoomIII, John Carmack Speaks Again
- Thread starter multigl2
- Start date
- Status
What if the reason that there are no credible leaks about the NV30 is that NVidia is keeping its features very tight under wraps in order to blindside the competitors when it comes out? Remember NVidia's statement about how they love that their competitors are targeting to beat the GF3 architecture? Why would NVidia want competitors to know what the performance/features of their next-gen part is, 6-12 months ahead of its release? That gives competitors time to prepare a PR offensive against it, or worse, design more ambitious hardware targeting its level.
What if the NV30 really is a sea change in performance and features? Do you really think they would leak those spec sheet PDFs to all the review sites again?
Perhaps they will wait until the R300 is released, and on the day that ATI announces the R300 officially and the ATI's NDAs expire, NVidia will release or "leak" info about the NV30 and they will attempt to use the NV28 to buy time.
The NV28 might be a stop gap measure, like the GF3 Ti was, to reduce cost of the GF4 and boost performance, in case they can't deliver the NV30 this fall and have to delay until Q1 2003. In that case, the NV28 will go up against the R300 for a few months while the NV30 is finished.
What if the NV30 really is a sea change in performance and features? Do you really think they would leak those spec sheet PDFs to all the review sites again?
Perhaps they will wait until the R300 is released, and on the day that ATI announces the R300 officially and the ATI's NDAs expire, NVidia will release or "leak" info about the NV30 and they will attempt to use the NV28 to buy time.
The NV28 might be a stop gap measure, like the GF3 Ti was, to reduce cost of the GF4 and boost performance, in case they can't deliver the NV30 this fall and have to delay until Q1 2003. In that case, the NV28 will go up against the R300 for a few months while the NV30 is finished.
OpenGL guy said:Uh, what are your sources for this information? Do you work at ATI? Have you signed an NDA with ATI? Have you seen the R300? I'll answer the last three for you: No, no and no.
I see now. JC hasn't worked closely enough with nvidia, obviously, that's why ATI "won" the demonstration at E3. Sounds logical to me!
I find this incredibly amusing.
I think Kristof is claiming to be a psychic.
Go on believing what you want, it'll make my job that much easier.
P.S. Your "DoomIII drivers" comment reminds me of that wacko a few months back claiming that ATI had some "extra-special-benchmark-cheating-drivers released only a few select websites".
SIGH... you sure seem to have a black and white view of the world, if something is said it must either be Pro-NVIDIA and Anti-ATI or Pro-ATI and Anti-NVIDIA.
Do I work for ATI : NO, do I work in the 3D Hardware Industry : YES. Have I talked to ATI DevRel people in the last couple of months : YES. Do I know how this kind of demos is handled by ANY company with some common sense : YES.
I am not saying ATI did something bad, on the contrary they did what any company (sh)/(w)ould have done. They are offered a huge marketing opportunity, not doing what I have suggested would have been silly.
All you seem to care about is twisting what I say so it sounds like ATI did something bad and I must be pro-NVIDIA.
JC has a huge impact on the 3D Hardware industry, any company that cares works very closely with him. All I suggested is that given this opportunity ATI must have worked over time and aimed to fully please JC no matter what.
Think about it :
- JC wants to demo DoomIII, he needs a graphics card that runs it at a good speed and without crashing.
- ATI wants to hype their hardware, they want to make sure the thing runs fast and does not crash.
- NVIDIA also wants to hype but has no real new stellar hardware ready.
JC and ATI decide to hook up, ATI pulls out all resources to make sure both JC and ATI do not become the laughing stock of the industry when the demo crashes more than anything else. Sounds all very logical to me and you do not need to be psychic to realise this, just use some common sense. And I still don't see you providing any common sense reasoning why there was "not" a special DoomIII driver build, unless ATI suddenly has the magical gift to pull perfect drivers out of its a$$ I am damn sure that a special set was created just for this... its the logical things for a Hardware company to do in such a case especially given the timing (R300 will not appear "soon" in stores near you) 8)
Or maybe you care to enlighten me on how JC and ATI came to demo-ing DoomIII on not-yet-announced R300 hardware ? Do you believe that ATI just provided JC with some early in-progress drivers that happened to run fine without any extra effort from either side ?
K~
Typedef Enum
Regular
With the clarification made by JC...and factoring in pure speculation...I'm still of the opinion that nVidia attempted to get an NV30 sample to E3...and they tried pretty hard. But in the end, they weren't able to deliver.
I agree with Demo about NV30...but I will also say that it's very difficult to keep things under wraps when you're the top dawg, and there are a good number of non-nVidia outfits that potentially have access to NV30 information...
With the possible exception of Matrox, things have a funny way of leaking when a chip is close to being final...Particularly if you're talking about nVidia...we can look back in history and see it happen time/again.
But there could be some degree of truth with NV30...It's a real question mark, due to the 3dfx/Gigapixel acquisition(s). We know it will definitely be a first for this fusion of talent, and I think you can pretty much say that anything is possible at this point in time.
I agree with Demo about NV30...but I will also say that it's very difficult to keep things under wraps when you're the top dawg, and there are a good number of non-nVidia outfits that potentially have access to NV30 information...
With the possible exception of Matrox, things have a funny way of leaking when a chip is close to being final...Particularly if you're talking about nVidia...we can look back in history and see it happen time/again.
But there could be some degree of truth with NV30...It's a real question mark, due to the 3dfx/Gigapixel acquisition(s). We know it will definitely be a first for this fusion of talent, and I think you can pretty much say that anything is possible at this point in time.
Joe DeFuria said:
We can be certain it wasn't up to snuff by whenever Carmack's cut off time was. That might have been the night before, a week before, two weeks before, or three weeks before, etc.
He may have never been shown it as it wasn't demo-able at the time. It might have just returned from the fab. It may not work outside of a test bench environment, or without a debugger attached. It might work reasonably well, but have glaring output deficiencies that are easily correctable with a metal rev. There's lots of reasons that it might not be demoable now, but have no impact on shipping in 3-6 months.
On a related note, why wasn't Doom demoed with a Parhelia? And why aren't we assuming that the Matrox part won't come out for another 6-9 months?
Russ
p.s. I'm curious about the 615/620. There's only two small blurbs, all relating to AmdZone's comments on the 23rd of May. Everything else out there suggests its still on target.
*sigh* Even after all this, we still don't really know anything. 
You are searching for an silver lining on the horizont, don't You?DemoCoder said:
very unlikely with development cycles approaching 2years. The R300 is nearly finished; how should they (completely) alter the chip to have Nvidia's NV30 features on board.
They didn't do this before, why should they do it now? Normaly they do this 2-3months in advance but not 6-9months in advance.
You can be sure that they will do this if necessary. This behavior will be an clear sign of weakness ( but all Nvidia-fan's will think otherwise )
Yep; but very expensive too, because it will rely on >400MHz DDR-SDRAM to get an decent performance boost.
Joe DeFuria
Legend
Russ,
It doesn't matter. At "whatever point", R300 was able to run Doom3 "better than any other part", and at the same time, NV30 wasn't. (Though based on some reporter's comments, the decision was made at "the very last minute"...I'll try and find the link). Again though, there is no indication from Carmack that nVidia made a legitimate attempt to have Doom3 running on NV30. All indications are that they simply tried the tweaked GeForce4.
No disagreement. There are also lots of reasons for concern that if a part is not demoable now, it won't be ready for a "quality" launch in 3 months.
Ummmm...because Doom "demos best" on the R300, and not Parhelia?
Again, no one is assuming anything. We're speculating. But to answer your question, Matrox has
1) Publically and officially announced the Parhelia architecture, and made public statements about the estimate of availability of products.
2) Matrox has publically demoed Parhelia alpha / beta products running a variety of software.
nVidia has done neither of those with respect to NV30. The two situations (Parhelia and NV30) are as different as night and day.
It doesn't matter. At "whatever point", R300 was able to run Doom3 "better than any other part", and at the same time, NV30 wasn't. (Though based on some reporter's comments, the decision was made at "the very last minute"...I'll try and find the link). Again though, there is no indication from Carmack that nVidia made a legitimate attempt to have Doom3 running on NV30. All indications are that they simply tried the tweaked GeForce4.
No disagreement. There are also lots of reasons for concern that if a part is not demoable now, it won't be ready for a "quality" launch in 3 months.
Ummmm...because Doom "demos best" on the R300, and not Parhelia?
Again, no one is assuming anything. We're speculating. But to answer your question, Matrox has
1) Publically and officially announced the Parhelia architecture, and made public statements about the estimate of availability of products.
2) Matrox has publically demoed Parhelia alpha / beta products running a variety of software.
nVidia has done neither of those with respect to NV30. The two situations (Parhelia and NV30) are as different as night and day.
Johnny Rotten
Regular
Nice.
Anyways, while we're all speculating I'm going to say that we'll see an nvidia paper launch during their regular time (Sept/October) and we'll actually see shelf product November/December. I simply cant see nvidia surrendering fall/winter 2002 to ATI. They're too aggressive.
Joe DeFuria
Legend
Johnny,
Yes, if R300 is available earlier than NV30, "paper launch" might be in nVidia's cards, although I think that would back-fire for them in this environment. Such a move would be designed to get people to "wait" for NV30 vs. buying the competitor's R300 "now."
However, a paper launch would also make people think twice about buying the currently available high-end nVidia parts (GeForce4 ti) now. This would put nVidia's OEMs in a tough spot...they would likely continue their migration to ATI. nVidia hyping their next-gen when the OEMs are trying to sell the current gen? In this business, nothing can take the place of shipping products. Ironically, nVidia is the one that proved that approach "works" time and time again.
Of course, ATI could also play that game...when NV30 does ship, ATI could start the "hype" on the successor to the R300.....
Yes, if R300 is available earlier than NV30, "paper launch" might be in nVidia's cards, although I think that would back-fire for them in this environment. Such a move would be designed to get people to "wait" for NV30 vs. buying the competitor's R300 "now."
However, a paper launch would also make people think twice about buying the currently available high-end nVidia parts (GeForce4 ti) now. This would put nVidia's OEMs in a tough spot...they would likely continue their migration to ATI. nVidia hyping their next-gen when the OEMs are trying to sell the current gen? In this business, nothing can take the place of shipping products. Ironically, nVidia is the one that proved that approach "works" time and time again.
Of course, ATI could also play that game...when NV30 does ship, ATI could start the "hype" on the successor to the R300.....
Paper launches have never meant much in the past, from what I can tell. I just hope that nVidia doesn't have massive problems with yields come September, so we don't end up with another debacle like the release of the original GeForce3 (low yields of the original GF3=very high prices for just a few released early...).
Of course, nVidia has always had some breathing room in the past. They don't anymore...
Of course, nVidia has always had some breathing room in the past. They don't anymore...
I doubt it would be, however OpenGL2.0 may play a part in this. If OpenGL2 is available by fall then ATi could at least support one higher level shading language; however, MS may not want to be perceived as being behind so its possible that openGL2 may push MS a little as well.
Johnny Rotten
Regular
Joe DeFuria said:
I doubt it cause any more hesitation in buying a GeForce4 Ti than an actual R300 based card on the shelves would.
At this point I think Nvidias OEM partners are pretty much used to being bent over like this. I cant imagine being an nvidia oem being all that much fun to be honest. Aggressive, rapid product cycles are great for the consumer but are hell for ihvs.
Yes, but imo paper launches are only stopgaps. A paper launch can buy say, maybe 6 to 8 (at the outside) weeks (using myself as an example here). When I'm itchy to buy though and you start asking me to wait in excess of 2 months, well....
Of course if ATI's successor to the R300 *does* fall within 4-6 weeks of the NV30, then it may actually be a valid tactic. I'm not expecting that to be the case though.
Joe DeFuria
Legend
GeeForcer,
I don't recall the exact circumstances around the R100 launch (is that the original Radeon?), but the situation was different with the R200 ATI, for two reasons:
1) R200 was not "paper launched" at the same time that GeForce3 Ti was launched. R200 was "paper launched" in August '01. GeForce3 Ti was launched in October. Both products actually shipped in a similar time frame, within a month of each other.
2) More importantly, at the time, ATI was NOT selling parts to OEMs. The R200 was the first part that it was selling to OEMs for 3rd party boards. So, in fact, ATI "pre-hyping" the R200 was a "good thing" as far as the vendors were concerned, and having a good relationship with ATI. It wasn't cannibalizing sales of "older" ATI parts they were trying to sell. ATI themselves may have taken a hit with fewer "older radeon" sales, but the vendors don't care about that.
I don't recall the exact circumstances around the R100 launch (is that the original Radeon?), but the situation was different with the R200 ATI, for two reasons:
1) R200 was not "paper launched" at the same time that GeForce3 Ti was launched. R200 was "paper launched" in August '01. GeForce3 Ti was launched in October. Both products actually shipped in a similar time frame, within a month of each other.
2) More importantly, at the time, ATI was NOT selling parts to OEMs. The R200 was the first part that it was selling to OEMs for 3rd party boards. So, in fact, ATI "pre-hyping" the R200 was a "good thing" as far as the vendors were concerned, and having a good relationship with ATI. It wasn't cannibalizing sales of "older" ATI parts they were trying to sell. ATI themselves may have taken a hit with fewer "older radeon" sales, but the vendors don't care about that.
Typedef Enum
Regular
So many questions...So few answers
Basically, we're about where we were 2 weeks ago or so. Still hypothesizing
Basically, we're about where we were 2 weeks ago or so. Still hypothesizing
Joe DeFuria
Legend
Yes, I agree. All "solutions" geared toward addressing a lack of a directly competing product would be a "stop gap."
IMO, if such a thing were to occur (NV30 being 3+ months later than R300), I think a better stop-gap solution for nVidia would be to significantly lower the price of GeForce4 Ti components....say GeForce4ti 4600 128 MB equivalent selling for under $200 MSRP.
This would keep the IHVs happy, as they are able to lower their price, and potentially keep the volume of chips flowing. The disadvantage to this approach compared to the "paper launch of high-end part", is taking a mind-share hit for no longer being the "leader", and taking a profit hit because of reduced selling price. But I think those are less important than maintaining OEM relationships for the long haul.
In either case, any stop-gap measure will eventually break-down if the root of the problem is not addressed: shipping competitive products. So, if NV30 is a bit late, nVidia must concentrate on getting back-on-track with their follow-up.
Typedef Enum said:
I don't know about you (and nvnews.net), but it's new to me that:
1) ATI has their next gen up and running in such a good state that Id can rely on it for a very important demo on a crucially important project. I'm impressed by ATI.
2) NV30 was nowhere as ready as R300 for E3. End of discussion.
3) nVidia is working on a higher speed GF4 which can only be on a 0.13um process. This implies a GF4 Ultra for this fall - or a massive shift of the GF4 line to 0.13um process to make it more viable in the mainstream segment.
- Status