• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

NVIDIA GeForce GF100 Architecture

I'm saying there is conflicting information from different sources--wait and see what happens.

No, there's just one source in contradiction that is clearly based on the info presented on GTC* that is speculating on what was going to happen, because Nvidia at the time only presented the Fermi that the CUDA/OpenGL/Compute will see. CUDA/OpenGL/Compute run on an abstraction level that directly bypasses all the graphics and fixed function units and hence everybody thought Fermi wouldn't have them (so the GPGPU modded into GPU idea came from here). Well, it has them, and oh boy, sure it has them, it has 4 "full" pipelined ones.

*And I know it's based in the GTC info, because he is saying exactly the same that all the publications posted when they returned from GTC. It's almost a copy/paste in fact.
 
Ben, whats true in this?
 

The very thing you've quoted does ask the question regarding price. I later elaborated on that statement by saying that I that I was hoping someone in the know could provide the information since the video card is do out in a few months. But if you don't know I don't hold it against you. The question wasn't directed towards you. However, it's apparent that my question has really upset you through your repeated arguing. However, we will see how they are priced soon enough though.
 
Like I said, I responded what you asked and I'm not upset at all at the question. I only have a problem with how you react to the answer you were asking for. Erocker understood the answer perfectly, but you just never accept anything from me without arguing and pretending I have something against you or your comments. Once again, ask the right question and you will receive the right answer.

Right questions:

- Anyone knows something about the price?
- Is there any info on the price?
...

Answer to the right question:

- No idea.
- No idea, probably more expensive than Cypress.
...

Wrong question:

- What's going to be the price based on the die size this die that...

Answer to the wrong question;

- Street price will not be based on die size. + I was kind enough to explain why it's not based on die size alone.
 
Last edited:
That's a tad bit too petty. Anyway back at the ranch, I think the following would shed some light on the GF100:
-final clock frequency
-power draw
-heat, cooling and noise
along with what's already been discussed, possible SKUs (448 vs 512), as well as final price. Some things have been said about it in the presentation but there are other points of interest as well.
 
Last edited:
Wow, this is like my 2nd post?

I see a lot of petty arguements here.

FACT:
This is a lifting of the NDA but done very differently. There is no product yet. The boards are for PR only. In essence, what we have here is most definitely careful marketing to disrupt confidence in the ATI 5xxx series and deflect it to the 'monster on the horizon'.

Yes there is substantial info here but why show only certain things, such as a 'portion' of the DX11 bench? Why not the score over the entire engine? Is Charlie right in his assertion at semi-accurate that as tesselation is shared then when the GPU is processing other articles it will slow down. (i.e. an optimum benchmark example has been given).

Also, no power info. The card is power hungry. Charlie was wrong about why Fermi and Oak ridge aren't happening (it's political and green) but the power consumption of HPC's is an issue here. For ref, CD stated Fermi was too hungry for power to be used - not true according to sources at BSN*.

This card is late for some very worrying reasons. It was slated to hit the christmas market way back in Sept 09 but it has slipped worryingly far back. That directly infers either a defect in some process or it had to be reworked to better ATI's offering (or both).

To discuss now if the card is better than the HD 5xxx series is a moot point. Is it more powerful? - almost definitely. Is it faster than a HD 5870? - almost definitely true. Is it faster than a HD 5970? Unknown but probably by a small margin.
But what we also have to consider is it's heat output (it's a massive die), it's cooling solution (for noise comparisons) and most importantly, it's price. It's going to be more than a HD 5970 or be sold at a loss, surely.

And please folks, dont forget, this is an Nvidia marketing release. It's 100% biased in it's delivery. There will be many elements of GF 100 that aren't palatable but you'll not hear about it from NV. NO company undersells itself.

So, I say, accept that if it makes it to the light if day it'll be:
Very Powerful, Very hot, noisy and prohibitively expensive. And dont forget to use 3D vision on more than one screen you need more than one 120Hz monitor. Aston Martin make great cars.... but they're in chronic debt because not enough people buy them.

Watch this space.
 
Wow, this is like my 2nd post?

I see a lot of petty arguements here.

FACT:
This is a lifting of the NDA but done very differently. There is no product yet. The boards are for PR only. In essence, what we have here is most definitely careful marketing to disrupt confidence in the ATI 5xxx series and deflect it to the 'monster on the horizon'.

Yes there is substantial info here but why show only certain things, such as a 'portion' of the DX11 bench? Why not the score over the entire engine? Is Charlie right in his assertion at semi-accurate that as tesselation is shared then when the GPU is processing other articles it will slow down. (i.e. an optimum benchmark example has been given).

Also, no power info. The card is power hungry. Charlie was wrong about why Fermi and Oak ridge aren't happening (it's political and green) but the power consumption of HPC's is an issue here. For ref, CD stated Fermi was too hungry for power to be used - not true according to sources at BSN*.

This card is late for some very worrying reasons. It was slated to hit the christmas market way back in Sept 09 but it has slipped worryingly far back. That directly infers either a defect in some process or it had to be reworked to better ATI's offering (or both).

To discuss now if the card is better than the HD 5xxx series is a moot point. Is it more powerful? - almost definitely. Is it faster than a HD 5870? - almost definitely true. Is it faster than a HD 5970? Unknown but probably by a small margin.
But what we also have to consider is it's heat output (it's a massive die), it's cooling solution (for noise comparisons) and most importantly, it's price. It's going to be more than a HD 5970 or be sold at a loss, surely.

And please folks, dont forget, this is an Nvidia marketing release. It's 100% biased in it's delivery. There will be many elements of GF 100 that aren't palatable but you'll not hear about it from NV. NO company undersells itself.

So, I say, accept that if it makes it to the light if day it'll be:
Very Powerful, Very hot, noisy and prohibitively expensive. And dont forget to use 3D vision on more than one screen you need more than one 120Hz monitor. Aston Martin make great cars.... but they're in chronic debt because not enough people buy them.

Watch this space.
Nicely put;)

But they are still going to argue AND then once it's out they will War with each other:shadedshu

I just hope it's better than the Cypress for the Consumers sake an inferior card would lead to price gouging :banghead:
 
Wow, this is like my 2nd post?

I see a lot of petty arguements here.

FACT:
This is a lifting of the NDA but done very differently. There is no product yet. The boards are for PR only. In essence, what we have here is most definitely careful marketing to disrupt confidence in the ATI 5xxx series and deflect it to the 'monster on the horizon'.

Yes there is substantial info here but why show only certain things, such as a 'portion' of the DX11 bench? Why not the score over the entire engine? Is Charlie right in his assertion at semi-accurate that as tesselation is shared then when the GPU is processing other articles it will slow down. (i.e. an optimum benchmark example has been given).

Also, no power info. The card is power hungry. Charlie was wrong about why Fermi and Oak ridge aren't happening (it's political and green) but the power consumption of HPC's is an issue here. For ref, CD stated Fermi was too hungry for power to be used - not true according to sources at BSN*.

This card is late for some very worrying reasons. It was slated to hit the christmas market way back in Sept 09 but it has slipped worryingly far back. That directly infers either a defect in some process or it had to be reworked to better ATI's offering (or both).

To discuss now if the card is better than the HD 5xxx series is a moot point. Is it more powerful? - almost definitely. Is it faster than a HD 5870? - almost definitely true. Is it faster than a HD 5970? Unknown but probably by a small margin.
But what we also have to consider is it's heat output (it's a massive die), it's cooling solution (for noise comparisons) and most importantly, it's price. It's going to be more than a HD 5970 or be sold at a loss, surely.

And please folks, dont forget, this is an Nvidia marketing release. It's 100% biased in it's delivery. There will be many elements of GF 100 that aren't palatable but you'll not hear about it from NV. NO company undersells itself.

So, I say, accept that if it makes it to the light if day it'll be:
Very Powerful, Very hot, noisy and prohibitively expensive. And dont forget to use 3D vision on more than one screen you need more than one 120Hz monitor. Aston Martin make great cars.... but they're in chronic debt because not enough people buy them.

Watch this space.


Thasnks for ur observation, as ben didnt see my question
 
The "G" Man to nVidia: "Prepare for unforeseen obstacles!"
 
Lets wait and see what happens, til then we should keep our fingers crossed for real world benchmarks.
 
I bet you guys click that "System Specs" drop down arrow whenever you see an opimist / pessimist nvidia GF100 post :D
 
I bet you guys click that "System Specs" drop down arrow whenever you see an opimist / pessimist nvidia GF100 post :D

Then when you click on mine, see a HD4870X2 and then see me advocating waiting for Fermi, what do you think ? :D
 
Then when you click on mine, see a HD4870X2 and then see me advocating waiting for Fermi, what do you think ? :D

i then see the 9600 and 8600GT's and assume your one of the few people on these forums that actually dont play f@nb0|and understand the argument for both sides
 
fermi looks really good, better than tesla
but if the price too high, i might stick with ATI
 
Back
Top