GTX485 is coming

Qbah

Diamond Member
Oct 18, 2005
3,754
10
81
That neoseeker picture is sooooo badly photoshop'd it's not funny...

As for the kitguru link - sounds reasonable except for the time. Fermi just launched, no way they're releasing an improved one this fast.
 

faxon

Platinum Member
May 23, 2008
2,109
1
81
Will GTX 485 offer bigger holes for your enjoyment?
you can sign me up for one! :biggrin:

and yea, that 485 photo is obviously photoshopped. one of the heatpipes up top got cloned, but the artist forgot just where those heatpipes are all supposed to come out, there's absolutely no reason why they would route the heatpipe the way the artist routed it, it doesnt even line up with the mounting point for the rest of them haha
 

T2k

Golden Member
Feb 24, 2004
1,665
5
81
" 20MHz more on the core clock, "

WoOoW... really? :D

OTOH everyone always said they will try to fix their problems later (vias etc) and release a 512-core original version somehow... of course, if it won't hit before August then it's pretty much a wasted effort as AMD is already working on something big for the Fall-Winter season.
 

extra

Golden Member
Dec 18, 1999
1,947
7
81
"Will GTX 485 offer bigger holes for your enjoyment?"...

Uhhmmmmm. Yeah. I don't even know what to make of that caption =p. lol
 

MentalIlness

Platinum Member
Nov 22, 2009
2,383
11
76
I call shens. Look at the bottom part of the grill in the GTX485. Some of the lines doesn't line up. Like in photoshop or something.
 

Meghan54

Lifer
Oct 18, 2009
11,684
5,225
136
I call shens. Look at the bottom part of the grill in the GTX485. Some of the lines doesn't line up. Like in photoshop or something.


The card you're looking at, the one on the right that's being called a 485, is actually an EVGA GTX 470 Superclocked+ card.

http://www.newegg.com/Product/ImageG...20Video%20Card


Click on the second image in top row. Compare to the KitGuru picture. Strange how they're exactly the same, isn't it?
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
" 20MHz more on the core clock, "

WoOoW... really? :D

OTOH everyone always said they will try to fix their problems later (vias etc) and release a 512-core original version somehow... of course, if it won't hit before August then it's pretty much a wasted effort as AMD is already working on something big for the Fall-Winter season.

Yep, the original fullspeed Fermi was supposed to be 512 CUDA core @ 750 Mhz.

I'm sure they will have this spec running sometime by the end of the year.
 

zerocool84

Lifer
Nov 11, 2004
36,041
472
126
I call shens. Look at the bottom part of the grill in the GTX485. Some of the lines doesn't line up. Like in photoshop or something.

Yea it's definitely chopped. Lines don't match up like you said not to mention they didn't even try to make the heatpipes up top look believable.
 

Sylvanas

Diamond Member
Jan 20, 2004
3,752
0
0
Well, It's pretty much a given that we will see a 512core part at some stage- I hope the sooner the better. I would not be surprised if they got a little more clockspeed out of it but 20mhz is pretty trivial.
 

Cookie Monster

Diamond Member
May 7, 2005
5,161
32
86
It be better for them to fix the power problem first "if" this new GTX485 is based on a new silicion revision. If the power can be fixed, then obviously noise/heat problem would also be fixed.

So it will be interesting on which of the short comings of GF100 were fixed unless the card itself is based on cherry picked parts (of the A3 chips) that run at a lower voltage for instance.

It be better for competition if the card was a full 512SP part with the fixed GDDR5 controller (that would result in a bandwidth of 240GB/s if the GDDR5s were clocked at their stock clock frequency of 5000MHz). So most performance improvements over a GTX480 would come from those fixes and extra logic units than the mere 20MHz increase.
 

Skurge

Diamond Member
Aug 17, 2009
5,195
1
71
It be better for them to fix the power problem first "if" this new GTX485 is based on a new silicion revision. If the power can be fixed, then obviously noise/heat problem would also be fixed.

So it will be interesting on which of the short comings of GF100 were fixed unless the card itself is based on cherry picked parts (of the A3 chips) that run at a lower voltage for instance.

It be better for competition if the card was a full 512SP part with the fixed GDDR5 controller (that would result in a bandwidth of 240GB/s if the GDDR5s were clocked at their stock clock frequency of 5000MHz). So most performance improvements over a GTX480 would come from those fixes and extra logic units than the mere 20MHz increase.

Can you tell me more about this memory controller?
 

Cookie Monster

Diamond Member
May 7, 2005
5,161
32
86
Can you tell me more about this memory controller?

In other words, nVIDIA didn't design the GDDR5 controller properly hence the reason why the GDDR5 memory clocks are clocked so low than their rated specs. Apparently when memory clock frequency increases, there is too much noise introduced to the controller. This isnt desirable hence the reduction in clocks. In contrast AMD has had 2 years experience with GDDR5 controllers, so they haven't had any real problems although its only with the cypress generation that they have been able to downclock the GDDR5 memory.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
In other words, nVIDIA didn't design the GDDR5 controller properly hence the reason why the GDDR5 memory clocks are clocked so low than their rated specs. Apparently when memory clock frequency increases, there is too much noise introduced to the controller. This isnt desirable hence the reduction in clocks. In contrast AMD has had 2 years experience with GDDR5 controllers, so they haven't had any real problems although its only with the cypress generation that they have been able to downclock the GDDR5 memory.
you do know that gddr5 comes in different speeds dont you? just because AMD uses some gddr5 thats rated for 5000mhz doesnt Nvidia is using the same memory.
 

HurleyBird

Platinum Member
Apr 22, 2003
2,800
1,528
136
so it is the memory controller or at least some other problem on Fermi thats the problem? good lord Fermi has some issues.

Well, to be fair AMD's first gen memory controllers had issues to. And their current memory controller may have some minor issues, seeing as they run 5000MHz rated memory at 'only' 4800MHz.

Even then, it makes sense given that Fermi has a large bus and is power/heat limited that they would trade off memory speed for core speed at a specific TPD.