How will ATI respond to Fermi? (poll included)

[1h4x4s3x]

There's no point in lowering prices when the retail prices are inflated due to high demand/relatively low production capacity. Nvidia will face the same issues and thus competition will be severely limited.
Only dealers would benefit (even more) from a lower MSRP.

That's my take.

 

[evolucion8]

Hmmm I remember the xt1950xtx raping a 7900gtx? And the beating got worse with more shader heavy games.

I saw a review that I'm still looking that in newer games like Bioshock 2, Mass Effect 2 and others, the X1950XTX was twice faster than the 7900GTX, the nVidia card was averaging 6fps and ATi's card was averaging 13fps, both unplayable,

I just hope that there is no shortage in supply at launch for Fermi like the 58xx series had.
Thats a killer for all of us, we all will pay more for our upgrades wether it be ATI or Nvidia.

Many people like myself have seen NO reason to upgrade our gtx 260'/75/80/85's or 4850's/60's/70/or 90's for the past 5 months untill now with Metro 2033.

I sure wasn't paying no $300 for a 5850(400+ for a 5870) nor was I waiting for bing cash back or a rebate. I will just wait for lower prices or a new compeditor to lower prices.

So I'm pointing now and laughing.

I strongly agree with you. I see no reason to upgrade since there's no game that I wasn't able to max with a single HD 4870 except Metro 2033 and now I can max it with Crossfire. The only real upgrade would be a HD 5870 in CF and I don't want to spend that much, too expensive for my tastes.

So I'm pointing now and laughing hysterically.

 

[taltamir]

the correct answer is "tiny bump of under 5% in clockspeed"... no redesign needed

 

[1h4x4s3x]

ATI Radeon 5870 2GB 3X2 Eyefinity
http://enthusiast.hardocp.com/article/2010/03/24/ati_radeon_5870_2gb_3x2_eyefinity_gaming_experience

Damn nice screens :awe:

 

[Martimus]

They aren't relying on the HPC per-se, but they are trying to grow it.
If they sell 1000 more PC GPUs than they sell in the HPC market, doesn't it make sense to try and change the balance of that?
Lets assume the profit levels are 10x what they are for GPUs, and it is 1000:1.
If they want to increase profits, then which is easier, going from current HPC to 100x HPC levels to almost double profits, or going from current GPU sales to 2x that?

The easiest way to increase profit for NV is to get the high margin HPC market buying their products in much greater numbers. Since the market is probably going to grow at a much greater rate than the desktop market, and the margins are much higher, it makes sense to invest in it while maintaining a high level of product in the desktop market.

That is true, but the margins will drop as the market expands. The margins are only so big now because it is a niche market with little competition. As it becomes less of a niche, more competitors will come out and the profit margin will naturally drop.

 

[rahvin]

I thought the 5-10k rumor was only referring to the availability of the GTX 480. Charlie reported a yield of ~2% (1.7?), ~104 candidates per wafer, and 5000 risk wafers, so that seems to make sense. Assuming nVidia managed to salvage another 10-20% of the dies as 448sp models, than that means around 50-100k GTX 470s.

I'm just guessing but I don't remember Charlie ever saying what the 10K number referred to, whether it was all Fermi chips or just 480. I can't imagine that if they have a 2% yield for 480 where they already have 32 dead SP's that they are going to get another 10-20% yield for chips with another 10% dead SP's. It seem implausible to me, in fact with 2% yield on the A2 die with the best they can do having 6.25% dead SP's then stepping down to the number of SP's on 470 (don't remember off the top of my head) is going to be a significantly larger yield. Statistically it would be highly improbably given the nature of chip defects. The best ball park I could see would be at best a 5% yield on the 470's given a 2% yield for 480's. I actually think it will be much closer to the 2% yield for the 480's though because most of the working chips are getting pulled as 480's given the number of dead SP's on the chip.

Even if you are right at the 10K only is the 480 I can't imagine that there is more than 20K 470's. You also have to wonder how much of the working production is going to get binned for Tesla cards where they will actually make a LOT of money on the chip. Tesla cards sell for more than $1000 so I would imagine nVidia is going to bin every chip they can for Tesla and Quadro cards and only produce the number of consumer cards they need to actually say they produced it. Everything nVidia has done press wise indicates this thing is a no show with a limited production run and probably a quick follow up to save face.

 

[1h4x4s3x]

You also have to wonder how much of the working production is going to get binned for Tesla cards where they will actually make a LOT of money on the chip.

Nvidia doesn't really make money with Teslas at least not yet. Quadro is their cash cow.