GTX480 (Fermi) 1536MB/600W (42A) Required

A GTX 280 recommended a 550W psu an had a tdp of 237W so this is in line with expectations (approx. 250W tdp). That said it's still a silly figure - this chip seems more designed for 28nm then 40nm.
 
More power / heat = more potential for overclocking for those of us who might be swapping coolers or W/C. :D

There is logic behind this, just not going to go into it now. :D

My only concern is idle power. If this thing can downclock and consume a reasonable amount of power during windows operations (aero, video playback, idle) then it's golden AFAIA concerned. It will spend 95% of its life at idle, so this is paramount.
 
There were posts @ B3D that the idle power draw would make those worried about it happy. No details, just rumors.
As long as it's not so low it's unstable....
Kind of defeats the saving if the user must increase the idle clocks to have a useable 2D setting. Does look great in power useage graphs though.
 
Kowan said:
There were posts @ B3D that the idle power draw would make those worried about it happy. No details, just rumors.
As long as it's not so low it's unstable....
Kind of defeats the saving if the user must increase the idle clocks to have a useable 2D setting. Does look great in power useage graphs though.
Click to expand...

I'd expect no less then 60 hz or 120 hz synced to refresh (depending on monitor) aero performance along with video playback at idle clocks. I agree anything else would not be acceptable.
 
Kowan said:
There were posts @ B3D that the idle power draw would make those worried about it happy. No details, just rumors.
As long as it's not so low it's unstable....
Kind of defeats the saving if the user must increase the idle clocks to have a useable 2D setting. Does look great in power useage graphs though.
Click to expand...

I see what you did thar!
 
Agreed. Flickering screens because the power is set too low wouldn't be a great bullet mark for marketing.
 
fdiaz78 said:
I'm calling shenanigans on this until we get some official specs.
Click to expand...

You're calling shenanigans on a retail box? Dude, seriously. FFS, All the bad stuff about Fermi that you guys keep denying... it's true.

4802.jpg
 
no suprise, the card has over 3 billion transistors, the card is a monster and will perform like a monster.
 
DualOwn said:
no suprise, the card has over 3 billion transistors, the card is a monster and will perform like a monster.
Click to expand...

One can hope. I have a feeling the supposed price tag is going to be a probable Achilles heel for many potential customers, though.
 
deff said:
One can hope. I have a feeling the supposed price tag is going to be a probable Achilles heel for many potential customers, though.
Click to expand...

I'm a bit skeptical of the rumored 650$ price tag unless this thing is significantly more than 30% faster than a 5870.
 
vengence said:
I'm a bit skeptical of the rumored 650$ price tag unless this thing is significantly more than 30% faster than a 5870.
Click to expand...

I could very easily see Nvidia charging $600+ for it even if it is only 20% faster than the 5870. You want top of the line, you PAY for top of the line. Top of the line cards are rarely competitive in a price/performance analysis.
 
kllrnohj said:
I could very easily see Nvidia charging $600+ for it even if it is only 20% faster than the 5870. You want top of the line, you PAY for top of the line. Top of the line cards are rarely competitive in a price/performance analysis.
Click to expand...

That pretty much sums it up.
 
The only thing that concerns me from these box shots, is how it renews my worry that NVIDIA may very well force everybody to buy overpriced Fermi-based Quadro cards, if they want DisplayPort and 10bit output...
 
kllrnohj said:
I could very easily see Nvidia charging $600+ for it even if it is only 20% faster than the 5870. You want top of the line, you PAY for top of the line. Top of the line cards are rarely competitive in a price/performance analysis.
Click to expand...

Oh I agree, however there is a limit on how out of whack you can get. You aren't going to be able to sell it for more than the 5970 if the 5970 is shown to out benchmark it regardless of crossfire.
 
kllrnohj said:
I could very easily see Nvidia charging $600+ for it even if it is only 20% faster than the 5870. You want top of the line, you PAY for top of the line. Top of the line cards are rarely competitive in a price/performance analysis.
Click to expand...



it will more than that in dx11 benchmarks at high res with alot of AA.

maybe 20% in older games though.
 
deff said:
I doubt it be that large of a margin.
Click to expand...

1 billion more transistors over the 5000 series are there for something you know.

Fermi will shine in dx11 performance over the 5000 series and that's all what matters to me right now.
 
DualOwn said:
1 billion more transistors over the 5000 series are there for something you know.
Click to expand...

Yup, they are there for double precision FP work. Want to guess how many games use DP precision? Zero. None. Zilch. Nada.

Number of transistors doesn't tell you shit about performance, especially when dealing with Fermi which is so heavily focused on GPGPU as well.
 
Headfoot said:
LOL

All the bad stuff like the fact it will use more or less the same power as the GTX 280 and not more like everyone says? :eek:
Click to expand...

Power consumption is supposed to decrease not stay the same or increase.
The card has the same power draw as a 5970. How is that a good thing?
 
deff said:
I doubt it be that large of a margin.
Click to expand...

kllrnohj said:
Yup, they are there for double precision FP work. Want to guess how many games use DP precision? Zero. None. Zilch. Nada.

Number of transistors doesn't tell you shit about performance, especially when dealing with Fermi which is so heavily focused on GPGPU as well.
Click to expand...

expect to be owned when bechmarks come out.
 
okashira said:
More power / heat = more potential for overclocking for those of us who might be swapping coolers or W/C. :D

There is logic behind this, just not going to go into it now. :D

My only concern is idle power. If this thing can downclock and consume a reasonable amount of power during windows operations (aero, video playback, idle) then it's golden AFAIA concerned. It will spend 95% of its life at idle, so this is paramount.
Click to expand...

First thought that came to mind was, Wc time! :p

You're calling shenanigans on a retail box? Dude, seriously. FFS, All the bad stuff about Fermi that you guys keep denying... it's true.
Click to expand...

Its not hard to fake a box to be honest, but im also not denying this to be true. One thing that I also said was, What no back shot of the box? Showing a semi decent picture of the card?

Looking forward for some previews soon. Price is not an issue here if it performs well.
 
dnottis said:
$600 price tag. I'm leaning more towards 650-800.
Click to expand...

800$? lol You think they are going to charge 30% more than a 5970 and twice as much as a 5870? I dunno what you on, but it must be some good stuff.
 
DualOwn said:
1 billion more transistors over the 5000 series are there for something you know.

Fermi will shine in dx11 performance over the 5000 series and that's all what matters to me right now.
Click to expand...

I'm not saying the technology is impressive, I am saying, however, that more of something doesn't always equate to higher performance.

All anyone can do is speculate at this point.
 
kllrnohj said:
Yup, they are there for double precision FP work. Want to guess how many games use DP precision? Zero. None. Zilch. Nada.

Number of transistors doesn't tell you shit about performance, especially when dealing with Fermi which is so heavily focused on GPGPU as well.
Click to expand...

While overall you're right about it not painting the picture on performance, you're not giving the whole perspective.. DP is not accounting for all of the transistor difference.. to assume that is a bit naive.

There is also a major re-design of how it handles geometry which was covered by reviewers weeks ago. That had a significant impact on transistor count.
 
BlueStorm said:
While overall you're right about it not painting the picture on performance, you're not giving the whole perspective.. DP is not accounting for all of the transistor difference.. to assume that is a bit naive.

There is also a major re-design of how it handles geometry which was covered by reviewers weeks ago. That had a significant impact on transistor count.
Click to expand...

they already did the dx11 tessellation on Fermi, it was 100% faster then the 5870.
 
I'm wondering how they expect people to run two of these things to get triple-monitor support. Not only is it cost prohibitive, but the power draw is a bit nutty.

They really need a $250 card, designed specifically so that two in SLI equal the performance of an HD5870. That would be a reasonable solution to get triple-monitor with Nvidia hardware.

Oh well, at least the upcoming driver update will let me run triple-monitor with my old GTX260 (without the need for SoftTH), I just need to grab a second one for SLI (only $180) and I'm good to go.
 
You must log in or register to reply here.
Back
Top