Nvidia D8E (9800/8900) pushed till 2008? [FUD WARNING]

[LordEC911]

We already have enough low end parts and they never start with releasing low end. High-end comes out first. Furthermore, if they launch in the same time frame as Intel (Harpertown in November) AMD will be between a rock and a hard place with no answer on both fronts and their shares would go down.

With no answer?
Where have you been? (haha I know, I made a funny)

Anyways, Barecelona and Phenom before the end of the year and on the GPU side, RV670 and R680 in Jan.
That is a decent "answer."

[virtualrain]

1. NVIDIA promised double precision floating point in the end of 2007 to the developers.
2. NVIDIA needs new high-end part with lower power consumption (read single slot cards) so it can revive SLI and Quad-SLI interest.
3. NVIDIA needs DX10.1 compliant chips and GDDR4.
4. NVIDIA would f*ck up ATI completely if they manage to release another high-end chip

Based on that, I believe that there will be new high-end part this year.

We already have enough low end parts and they never start with releasing low end. High-end comes out first. Furthermore, if they launch in the same time frame as Intel (Harpertown in November) AMD will be between a rock and a hard place with no answer on both fronts and their shares would go down.

Interesting view...

Personally, I don't think points 1 to 3 are driving them... as a product manager, I would be factoring these things into my roadmap but I would not let these influence my strategic thinking and product launch timelines.

Point #4, however, deserves some discussion in my opinion... however, in my opinion, the way to kill ATI is not with another high-end product... nVidia already has them beat in that department. The way to kill ATI is to offer better performing mid-range and low-end products at prices that force ATI to bleed to death... exactly the way Intel nearly killed AMD this year with aggressive price cuts to mid and low-end products.

I suspect the revenue from high-end cards is probably 10-15% of overall GPU revenues for both ATI and nVidia... which is part of the reason ATI could really care less about owning the top spot. If you really want to kick ATI in the nuts, you need to steal their bread and butter which is the OEM and main-stream consumer space.

By launching better mid-range and low end cards that are cheaper to produce, which is exactly what nVidia is doing, they will have a much bigger impact on ATI.

[LordEC911]

Point #4, however, deserves some discussion in my opinion... however, in my opinion, the way to kill ATI is not with another high-end product... nVidia already has them beat in that department. The way to kill ATI is to offer better performing mid-range and low-end products at prices that force ATI to bleed to death... exactly the way Intel nearly killed AMD this year with aggressive price cuts to mid and low-end products.

But there is, literally, no down side to releasing a new highend this year.
Just means the Ultra needs to be phased out, meaning a drop to ~$500-$550 and the GTX stays at ~$450-$500.
Hold onto the G90GTS till the Ultra/GTX are nearly gone, a $600-$650 GTX isn't exactly going to "hurt" them. (though everyone seems to think that...)

[virtualrain]

If nVidia launches another high-end part without any competition, I suspect the initial price would make the Ultra's launch price look like a bargain. There is virtually no incentive for them to lower the price on any GTX level product.

[quadnad]

The way to kill ATI is to offer better performing mid-range and low-end products at prices that force ATI to bleed to death... exactly the way Intel nearly killed AMD this year with aggressive price cuts to mid and low-end products.

While I wouldn't say "kill" ATi, you're certainly correct about putting a financial squeeze on them. The bread and butter of these graphics companies is the low to mid range mainstream parts.

BUT, in order to generate interest for those parts, they release the high end parts first. That was the reasoning behind them releasing their new high end in November; the following spring they would release a slew of low to mid range parts that act off of the success of the high end part (just like the 8600).

[Obscured]

In November 2006 ATI was fully competitive so nVidia quickly released it's new top G80. All things differ today and nVidia is in no need for hurry - ATI can't beat even "simple" G80. That's why nVidia has no reason for releasing their new top now. On the other hand there is a great gap between DX10-capable mid-end and high-end cards today. (a gap between 8600GTS and 8800GTS) - 8800GTS costs too much for mainstream; but 8600GTS can't show clear perfomance advantage over AMD RV570. This gap must be filled up.

[Decami]

Ok

We have "real" info straight from Nvidia's mouth saying Q4 2007(who cares when they said it). +many other confirmations of this from other sources. (of course, since it came from nvidia publicly)

next

We have information from a place that we are for shore is wrong (mostly) 80% (or more) of the time about the the news they post, saying that Nvidia has delayed their high end. Though seems like a good marketing decision, would also be going back on their previous info regardless of when they said it, and so far, I havent seen anything about this anywhere else, that doesn't say, "according to fud."

thats what we have......
.......I say we stick to thinking Q4 until something else arises.

[EternityZX9]

The one thing you guys have to keep in mind is that just because nVidia has a superior product, that doesn't mean it will stop them from sticking to a release schedule. Why do I say that? Because after the 6800 came out less than a year later the 7800 came out, when nVidia already had the performance lead. As much as I think it's possible they might delay the G90 until next year, it's still up to nVidia what they do and at this point in time that could be anything - probably why we see a new story come out each week from one of the internet rumor mills.

[Decami]

I also remember a couple people saying that Nvidia always releases low end first and never high end first. If you haven't been paying attention, that hasn't happened in a long time (if ever). GPU's from Nvidia are usually always made high end releasing first and then they begin releasing low end and making decisions for the names and type based on the spin of the silicon. Then sometimes release an even better version of the high end, hence the 8800 Ultra.

[Decami]

If nVidia launches another high-end part without any competition, I suspect the initial price would make the Ultra's launch price look like a bargain. There is virtually no incentive for them to lower the price on any GTX level product.

Well it really depends. Would it be smarter to lower the previous to make way for a little lower price on the release card, to apply it to a larger percentage? May equal to more gain? Or would it be smarter to keep the previous cards the same price and just have an extremely higher price on the release card that applies to a smaller percentage of the consumers? Which would equal more gain? Well thats a decision for Nvidia to make. But the past tells the truth. Which is usually a release of a high price which gradually (which is super slow for nvidia usually) lowers along with the previous cards.

[LordEC911]

I also remember a couple people saying that Nvidia always releases low end first and never high end first. If you haven't been paying attention, that hasn't happened in a long time (if ever). GPU's from Nvidia are usually always made high end releasing first and then they begin releasing low end and making decisions for the names and type based on the spin of the silicon. Then sometimes release an even better version of the high end, hence the 8800 Ultra.

I believe some of them were referring to the shrink.
Nvidia almost always shrinks their mid-low end cards first and then follows with a highend card on the same PP.

[virtualrain]

I believe some of them were referring to the shrink.
Nvidia almost always shrinks their mid-low end cards first and then follows with a highend card on the same PP.

Exactly... nVidia has typically done a GTX/GTS followed by mainstream and low-end followed by a die-shrink refresh of the mainstream along with an ultra variation (and in the case of the G70 a 7950GX2 as well) at the top to soak all the margin they can from the current gen before launching a next gen part.

Remember there was 2+ years between the 7800GTX launch and the 8800GTX launch... again, because ATI couldn't catch up with the 7800GTX for a long time.

We've now got news of a G80 version of the 7950GX2 product due in Q1 so I don't see a new next gen part coming until Q2 or later based on that news.

[LordEC911]

Remember there was 2+ years between the 7800GTX launch and the 8800GTX launch... again, because ATI couldn't catch up with the 7800GTX for a long time.

R520?

[Decami]

I believe some of them were referring to the shrink.
Nvidia almost always shrinks their mid-low end cards first and then follows with a highend card on the same PP.

Ahhh. Yeah, then in that case, yeah.