NVIDIA GF100 & Friends speculation
- Thread starter Arty
- Start date
What we need now is a declaration that whomever sells 10M units of GPUs first in a DX generation is the winner and that ATI have done just that. 
Actually the funny part about this industry is that the graphics cards are so open it often simply comes down to performance/price and brand name is only acting as a tie breaker betweem two equally compelling offerings.
Actually the funny part about this industry is that the graphics cards are so open it often simply comes down to performance/price and brand name is only acting as a tie breaker betweem two equally compelling offerings.
Is a powerpoint slide, other wise known as GTX 4x0 equally compelling to 58x0? :smile:
thatdude90210
Regular
Hypothetical question.
Let's say the German blogs reporting at Cebit are correct, that there are only 5000 total Fermi cards at launch world wide. Let's also say that PC perspective is correct in their podcast (~47minutes in), that "the manufacturability is pathetic," which echos another "unmanufacturable" claim.
If, and that's a big if, those conditions are true, could launching now actually backfire?
In other words, would it have been better to hold back until they have yield under control and then launch, or launch now and risk having their product launch be some sort of catalyst that helps whip up the desire to upgrade.
Let's say the German blogs reporting at Cebit are correct, that there are only 5000 total Fermi cards at launch world wide. Let's also say that PC perspective is correct in their podcast (~47minutes in), that "the manufacturability is pathetic," which echos another "unmanufacturable" claim.
If, and that's a big if, those conditions are true, could launching now actually backfire?
In other words, would it have been better to hold back until they have yield under control and then launch, or launch now and risk having their product launch be some sort of catalyst that helps whip up the desire to upgrade.
Assuming all these "slow, limited, hot" claims are true, then their best hope is is the B1 spin. That will come in June at the earliest, and there's no point letting AMD control the entire market. Besides the wafers have already been paid for. So even if the are getting 2 480's per wafer, it's better to make some money than none at all. They can always spin the physics+cuda thing and tell nvidiots to wait for B1.
If it is indeed as good as nv pr would like us to believe, then why wait for anything?
Either way, launching now seems better.
not only that but even really hardcore fan boys can only wait so long for new products. With limited avaliblity at least nvidia hardcore fans and some not so hardcore see that they have a competeing product even if its impossible to get
Aside from that I'm sure once they get their halo card out tehy will release all the defective fermis in how many configs as they can to recoup any money. So they will hope having the single fastest gpu will get them sales with the cut down stuff.
It could work as sme may have olde 8800s or 9800s or what have you that want dx 11 and they will use the old card for cuda and phsyx.
FenderBender
Newcomer
And it's actually pretty easy to spin any lack of availability.. "You're right, it's so hard to find the cards, isn't it? They're so popular, the new GPUs are a terrific success!"
You just need enough to make the first few days stock broadly available so it can't be called a paper launch. 5000 cards is enough to avoid that paper-launch label, but not all the grumbling.
You just need enough to make the first few days stock broadly available so it can't be called a paper launch. 5000 cards is enough to avoid that paper-launch label, but not all the grumbling.
They have a big disadvantage here. The install base of G8x and G9x is pretty large vs their contemporaries. And gpu accelerated physics is likely to be slow on 4xxx, and non-existent on anything before that.
OT: BTW, what is the present (ie today's) state of amd's gpu accelerated physics efforts?
Yup. I don't know ati's state of gpu acceleration physics. Did something happen today ? Sunday is my GF's day of the weekend to pick what we do (Friday is friends , Sat is my day ) And she wanted to lay in bed all day and cuddle and talk. I tried confusing her by talking about gpus but she wasn't falling for it..
Now with nvidia get a gtx 460 or 440and hook up a 8800 to it and your good to go.
I have no idea what ati is going to do. Though moving foward the 5x00 line looks really good in terms of heat. IF they actual had a go at this and got support (which amd sucks at getting for their exclusive features) I'd perfer to put a 5870 with a 5750 or 5670 together than a 8800 or with a gtx 480.
But right now nvidia has alot more support than ati
Eyeinfinty
DX 11 , lotro , Stalker , Dirt 2 , BF BC 2 , Metro , Battle Forge
Dx 10.1 Just cause 2 and others also.
But more importantly what about people with older gtx 8800 or 9800 or even me with a 4850.
Many of us want to upgrade for better performance. What part would a 4850 or 8800/9800 owner buy from nvidia if they had to buy today or in the last 6 months. Then explain why they would buy that nvidia card over the 5x00 line up.
My answer to you comes in the form of another question: And do you really think that if these performance rumors are true, that NVIDIA will launch the overall slower GTX 470, more expensive than the HD 5870 ?
You know, common sense is free...If the performance rumors are real, then there's no way that the GTX 470 will be more expensive than the HD 5870.
Let's list them...
- Batman: AA (with crap fps with PhysX On)
- Nothing else worth mentioneing
I'd say nVidia is nowhere with phisics on GPU. Exactly where AMD also is...
There's no need of phisics on GPU when it works so well on CPU, as Havok keeps remembering us...
nV's cards will launch at approximately 20 to 25% higher prices than their direct competitor parts.
GTX285 is slower than HD5850,but more expensive.....
NathansFortune
Regular
Indeed. With only 5000 available they will also sell out at full price, or even above full price. With availability so low it doesn't really make sense for Nv to compete on pricing. Hopefully with the B1 stepping they will get decent enough yields and remove some redundancy and make the chip smaller/cheaper/faster. That though is wishful thinking at best for B1...
Is that the new spin now ?
Heh, let's those fluid, hair or cloth physics with Havok on a CPU. Better yet, let's see fluid, hair or cloth physics on a CPU that doesn't cost an arm and leg.
Similar threads
