NVIDIA Kepler speculation thread

I already did
- not sure this is 100% proof though!

 

Don't fairy tales usually start with "Once upon a time...."

 

Yes , but they don't always end with "lived happily ever after" , not even often !

 

If bandwidth is a problem, 1920x1080 might be expected to do pretty well.

 

Interesting - with Tahiti, drop memory bandwidth by 33%, and fps drops by 13%....
- that is to say, a '256-bit' equivalent Tahiti gains 15% from 50% more bw
- well approximately anyway.. as they were simulating the 256-bit bus by reducing the bw on the 384-bit one... but a good ball-park perhaps...

 

Yeah, seems like single monitor users would be "OK" with 256-bit memory, but I imagine the falloff is a little steeper with multi-monitor.

 

What are the odds of GK104 actually being branded the "680"? If it is we could be looking at a very similar situation to the 9800GTX.

- Marginally faster over prior high-end and slower under certain conditions.
- Lower bandwidth.
- New process
- $299

Things are very different now than they were back then though. nVidia had no single card competition and the 9800 GX2 was already out too. It's hard to believe GK104 could produce enough of a performance increase over the 580 to earn the 680 name. Even the "broken" GTX 480 was 50-60% faster than the 280 + DX11.

 

I agree, Nvidia will be competition to itself - since GTX580-3GB version sales for $549, and why would anybody buy it for $549 - since you can get HD7970 for $549.

But AMD might need to cut price for HD7970 to counter GK104.

 

If those are the performance, than gk104 based vga will not cost 300$....

 

The whole problem with that is the following.

If it's faster than tahiti. Why price it at 299 USD when you can achieve the same effect at 399 USD? That's still hugely undercutting your competition. Assuming die size is similar and that 1 GB of GDDR5 isn't 100 USD or more. You'd still be eating into your margins.

At 299 USD, you'd be hugely eating into your margins for no reason.

I don't think either Nvidia or AMD want to start another price war, and that would be the only reasoning I can logically think of for charging 299 USD for a card that might outperform AMD's 7970 (not to mention your own current halo card).

And pricing it to allow room for a higher performing part doesn't make sense either. As Nvidia has already set a precedent for charging 649 USD (GTX 280) and 999 USD (7800 GTX Ultra/512). Although granted the latter card was only produced and shipped to reviewers to spoil AMD's product launch and for no other reason.

So, IMO, either the rumored price is wrong or the rumored performance is wrong.

Regards,
SB

 

1. GTX580 is EOL for 4 months, once GK114 hits the stores
2. GK114 is the performance card from NV, 299$ matches this
3. GK110 is the top dog and needs some room above GK114

I would consider the option that AMD is currently taking way too much money for their 79X0

 

BTW, what is "EOL" to you, as you can still buy GF114/GF110 from NVIDIA today and will be able to do so in the near future.

 

NV has not ordered new GF110 wafers since Dec. 2011 afaik. They still have some GPUs remaining, but I dare say that the stock is limited, as the price did not drop.

 

I can't comment on the specifics but it is not like what I've seen recently.

 

When is GK104 scheduled for release, roughly?

 

It's hard to imagine sales of the GTX580/570/560-448 being too brisk after the arrival of the HD7970/7950
- and even less so after the GK104 arrives
- but they may still need the GF110 for Tesla and Quadro, I guess...

Hey, look at that, I can edit my posts now!

 

Yeah, it is definitely hard to imagine. I'm wondering if they'll be showing Kepler off at PDXLAN 19... their site says, as of this morning....

"*** SPECIAL PDXLAN 19 ALERT!!! ***
Monday, 13 February 2012

Monday, 13 February 2012 NVIDIA and Gearbox are bringing something really special to PDXLAN 19.

Attendees are in for an exclusive treat that will blow your mind!!


NVIDIA and Gearbox are bringing something really special to PDXLAN 19.

Attendees are in for an exclusive treat that will blow your mind!!"

 

an announcement that there will be an announcement about when the product will be announced?

 

whitetiger,

It makes little to no difference to me whether you consider the GF114 or the GF110 a more balanced design. The case in point was an indirect parallel between performance and high end SKUs in the Fermi family and how there could be any comparable sitings in GK104 or in extension to that between performance and high end Kepler. For the record there's no such thing as enough bandwidth on any GPU out there.

 

That seems like a rather dubious statement. While bandwidth is certainly important, for graphics workloads you really only need enough not to cap the framerate in current titles below 120Hz at prevalent settings. Or, you could simply say that having more bandwidth than the ROPs can make use of is a rather inefficient use of resources.