NVIDIA Fermi: Architecture discussion

Honestly, seeing some reactions here, it's almost like some people have never witnessed a graphics card launch ever again. "OOOHH give us a release date!" "Release all the facts!!" "They are beaten without even releasing hardware"

...and so on and so forth.

Not a long ago, people were predicting doom and gloom for ATI. Now it's nVIDIA's turn to sink. Maybe in a year we'll be back to dissing ATI. And then nVIDIA again. And so on and so forth. But it's fun, I'll give you that; in more ways that you think actually :p
 
Honestly, seeing some reactions here, it's almost like some people have never witnessed a graphics card launch ever again. "OOOHH give us a release date!" "Release all the facts!!" "They are beaten without even releasing hardware"

...and so on and so forth.

Not a long ago, people were predicting doom and gloom for ATI. Now it's nVIDIA's turn to sink. Maybe in a year we'll be back to dissing ATI. And then nVIDIA again. And so on and so forth. But it's fun, I'll give you that; in more ways that you think actually :p

Very true, been happening since Ati got serious with 9700 I guess and actually became a competitor to take over from 3dFX.

I don't see the point much in a photo though, I'd rather see just see those leaked cherry picked powerpoint slides showing the performance difference. Now that is something to argue over :D
 
GT200 had a lot of other advantages compared to the RV770 so back then the Flops-rating was "meaningless":

GTX285 / 4870

Pixel: 20736 <-> 12000; 1,73:1
Texel: 51840 <-> 30000; 1,73:1

Therefore back then the Flops-rating advantage of the RV770 was not enough to overcome all the other disadvantages.

With Fermi the situation is most likely different.If the base-MHz/Shader-MHz ratio is roughly the same as with the GTX285 then the base frequency of the Fermi-GPUs is ~550MHz (this number comes also from Ailuros):

Fermi / 5870

Pixel: 26400 <-> 27200; 0,97:1 [according to Ailuros Fermi has 48 ROPs]
Texel: 70400 <-> 68000; 1,04:1 [based on the rumour that Fermi has 128 TMUs]

=> Fermi will, most likely have no other advantages compared to the RV870 except bandwidth. Therefore the difference in Flops-rating can tip the scale in RV870's favour.

This analysis doesn't seem to show anything useful with respect to gaming performance (at least without looking at other performance metrics too).

For example:

5870 / 4870

Pixel: 27200 <-> 12000; 2,27:1
Texel: 68000 <-> 30000; 2,27:1


So by this logic, the 5870 (having both a huge advantage over 4870 in the areas above AND in terms of pure FLOPS) should be walking all over 4870, right? We all know why this is not the case.....many other metrics need to be considered too.
 
Last edited by a moderator:
Possible.

I do want to know why the motherboard seems to be resting on nothing, just looks strange to me. A nice clear image would be nice, but that would be too simple I guess, better to post low quality images and have people speculate.

The mb seems to be at the edge of the table.
What really intrigues me is the right inwards side of the fan, its like as if theres a hole pierced in the pcb under the fan, odd?
 
This shows two things:

a) It will run DX11 benchmarks.
b) It's big.

So unless nVidia thought the community had any doubts of the above, why release this while conveniently leaving out the fps. Seems a bit... desperate is too strong a word but definitely awkward. I'd expect a similar stunt by PowerVR/SIS back from the dead or even LRB which do have something to prove.

Does nVidia believe it has something to prove?

Why would NVIDIA bother to show off [gaming] framerates right now? The drivers are not ready, so performance at this stage may be great but still not optimal. Also, no point in letting much out of the bag right now with respect to gaming performance considering the highly competitive nature of the business, and considering that consumers will not get cards in their hands for at least a couple months.

What NVIDIA needs to do is to surprise and even delight gamers. Think G80, NV40, etc. We'll see what happens soon I suppose.
 
Last edited by a moderator:
I'd rather see just see those leaked cherry picked powerpoint slides showing the performance difference. Now that is something to argue over :D
Yeah, but not with actual framerates, but with percentage differences! Now that rules. One game runs at 10fps with one card, 15fps with the other. OMG 50% difference :LOL:
 
We're in agreement then. Both Intel and Nvidia unveiled tidbits about their new architecture at their technical conference. Let's treat them the same then.
Fair enough. My point was Intel can waffle along with Larrabee for at least another year and not be financially affected by it. Nvidia not so much with Fermi.
The mb seems to be at the edge of the table.
It looks to be hardly on the table at all. So assuming it is dangling at the edge, pretty sloppy testing going on. I would never have a motherboard half off the edge of a bench, especially considering that is supposed to be Fermi running there, no doubt a rare card.
 
I would never have a motherboard half off the edge of a bench, especially considering that is supposed to be Fermi running there, no doubt a rare card.
You've obviously never been to a computer lab before :p And as far as rare cards are concerned, they are rare for you and me, but not so much for the companies. They don't have plenty of them, but there always are working prototypes months before the release of the card. I remember when I was working for AMD, and went to England's HQ for the first time, I got to see a 3870X2 4-5 months before it was released. I was like OMGOMG, whereas the others were like "meeeeeh, when will the Amerikanski send us some proper drivers for this thing?" :LOL:
 
Fair enough. My point was Intel can waffle along with Larrabee for at least another year and not be financially affected by it. Nvidia not so much with Fermi.

Relax man. NVIDIA posted a press release on their website which stated that Fermi-based GeForce cards are expected to be available for purchase sometime during Q1 '10 (calendar year). Even though turnaround time is two or three months more than NVIDIA wanted, this still seems almost infinitely better than Intel.
 
It looks to be hardly on the table at all. So assuming it is dangling at the edge, pretty sloppy testing going on. I would never have a motherboard half off the edge of a bench, especially considering that is supposed to be Fermi running there, no doubt a rare card.

Fermi, graphics at the edge of reality™ :LOL:
 
The mb seems to be at the edge of the table.
What really intrigues me is the right inwards side of the fan, its like as if theres a hole pierced in the pcb under the fan, odd?

If you look at the Tesla picture. It has something similar.

http://www.computerbase.de/bildstrecke/27497/3/

Nvidia has done this with its 295 and 9800GX2 cards as well because it allows airflow in from the back.
 
ATI kinda did the same thing with Juniper (6 months ago) and nobody was asking if it worked. Sure it was playing Wolfenstein and there were demos with EVERGREEN BROADWAY EG stamped on deviceID, but you could argue it's easily faked.

Precisely. It was not possible to verify that that was a real card DX11 card. No DX11 demo...nothing...and IIRC they even used the same demo for the HD 4800 launch. yet there was little mention of it being fake. Only in one of TechReport's podcasts did Scott mention that possibility.

Tchock said:
Let's stick to the more tangible stuff and crucify people that deserve to be crucified first. :LOL:

Agreed! :)
 
Maybe. I don't know. I do not know the reason for everything Nvidia does. Unfortunately everyone seems to think I do. If something I say does come from Nvidia. I am pretty clear on outlining that. TBH I am surprised at the reactions here. Most forums ((with the exception hardocp/B3D that I have read. Have mostly been positive. Most people are just happy to see a picture of the card.

I can comment on the HardOCP forum's "reaction"...Ever since AMD has launched the HD 5800 series, that it's practically impossible to start a discussion about NVIDIA products there, without that particular thread being flooded by the same old AMD die hard fanatics, that just ruin it to the point of it being closed. I was (and sort of still am) a regular there, but I don't post as much, because of that fact. And also the fact that despite reports, after offenses and even harrasement by some of these fanatical members, the mods do practically nothing and those people roam free to do what they want without consequence.
The video cards sub-forum there is pretty much just flamebait, disinformation and doom and gloom nowadays.
 
Perhaps we need to stem this tide? Yes?

I suggest to the mods here that a spin off about how Nvidia should/should not conduct themselves is in order. That way the speculation as to where/when Fermi is can be surgically removed from discussions about how Nvidia goes about their business.

Btw if that team has a sense of humour they'd take the Nvidia GPU out and replace it with an AMD model (which they should have) and post some more pictures. :)
 
Why would NVIDIA bother to show off [gaming] framerates right now? The drivers are not ready, so performance at this stage may be great but still not optimal. Also, no point in letting much out of the bag right now with respect to gaming performance considering the highly competitive nature of the business, and considering that consumers will not get cards in their hands for at least a couple months.

That's my point exactly. This picture (and really I have no problems with the picture itself but the context in which it was presented and how) has nothing of that. The reasons why it has nothing are not important and/or entirely understandable, like you said: drivers not ready, etc. etc.

Mod hat on: I recommend people dial down the conspiracy theories, mindless-speculation and leading questions. If you have unshakable proof of photochopping or whatever fine, but this isn't a thread for (over)analysing pictures some guy took with his cellphone.
 
Back
Top