Geforce GTX 485?

[Monkey_Business]

Just like the GTX 280 to 285, will they do this with the Fermi cards? If they do, what improvements will be made? 32nm die shrink?

[roast]

Probably a die shrink, and probably nothing else. May leave more headroom for Overclocking, and lower temps though.

[Hockster]

http://lmgtfy.com/?q=dual+GPU+fermi

[roast]

http://lmgtfy.com/?q=dual+GPU+fermi

:roll:

I expected that.

[epicfail]

http://lmgtfy.com/?q=dual+GPU+fermi

lmao i did the same thing in his, other topic.

[ShiBDiB]

http://lmgtfy.com/?q=dual+GPU+fermi

:slap: that would be a 495 he's asking about a 485... two completely different things.. way to fail at being a douche tho


Expect a 485, it'll likely be a die shrink just like the 280->285 was. Not a huge performance increase but slightly better temps which means slightly better OC abilities

[btarunr]

There will be a GeForce GTX 485. It will have 512 SPs, but same 40 nm core.

[ShiBDiB]

There will be a GeForce GTX 485. It will have 512 SPs, but same 40 nm core.

No shrink? They should call it a 490 then imo.. when most people see 485 theyll think that like the 285 its a die shrink

[btarunr]

No shrink? They should call it a 490 then imo.. when most people see 485 theyll think that like the 285 its a die shrink

They probably won't. GTX 465 which launches on 3rd June is based on the 40 nm core as well. So that should tell people that xx5 doesn't mean die-shrink in this series.

[mdsx1950]

I hope they launch a dual fermi card. GTX 495 or something that beats the 5970 by a long shot that has lower temps and power consumption than the GTX480 or atleast about the same.

[Monkey_Business]

I hope they launch a dual fermi card. GTX 495 or something that beats the 5970 by a long shot that has lower temps and power consumption than the GTX480 or atleast about the same.

If a dual GPU Fermi launches, it will most likely be Two GTX 470 cores.

[btarunr]

If a dual GPU Fermi launches, it will most likely be Two GTX 470 cores.

Yes, and HD 5850 Crossfire beats GTX 470 SLI. So down goes NV's dream of beating HD 5970 with GF100.

[TVman]

http://lmgtfy.com/?q=GTX+485 :pimp:

[the54thvoid]

I hope they launch a dual fermi card. GTX 495 or something that beats the 5970 by a long shot that has lower temps and power consumption than the GTX480 or atleast about the same.

Well that would be affordable.......

Bring on the tropical cards Southern and Northern Islands - Fermi is sooooo last year!

[boulard83]

There will be a GeForce GTX 485. It will have 512 SPs, but same 40 nm core.

+1

[mdsx1950]

If a dual GPU Fermi launches, it will most likely be Two GTX 470 cores.

Yes, and HD 5850 Crossfire beats GTX 470 SLI. So down goes NV's dream of beating HD 5970 with GF100.

Most probably. But who knows. nVidia might have made a card with two GTX480s and is keeping it in the dark till ATi puts out the 6xxx series so they can release the flagship card to beat ATi's 6xxx flagship card. But well, thats just opinion lol.

Well that would be affordable.......

Bring on the tropical cards Southern and Northern Islands - Fermi is sooooo last year!

It would maximum be priced at $1000. No biggie :p

[qubit]

There will be a GeForce GTX 485. It will have 512 SPs, but same 40 nm core.

...and on a 512-bit bus?

This will be the card that interests me. I'll only consider buying it once the price comes down from the stratosphere though.

The current top card is already cut down from the full GPU it should have been, so I don't want it. All that power, heat & noise really doesn't help, either.

Note that this comment is coming from someone who loves his GTX 285 and would really like to own a decent successor from nvidia.

[Benetanegia]

Yes, and HD 5850 Crossfire beats GTX 470 SLI. So down goes NV's dream of beating HD 5970 with GF100.

That is not correct. Not at all.

http://www.guru3d.com/article/geforce-gtx-470-2-3-way-sli-review/15

In fact, 470 SLI is faster than 5870 Crossfire in many cases.

[Monkey_Business]

How much faster will the HD 6xxx "Northern Island" cards be? As opposed to the HD 5xxx.

[epicfail]

How much faster will the HD 6xxx "Northern Island" cards be? As opposed to the HD 5xxx.

its not released yet no one knows.......:banghead::banghead::banghead::banghea d::banghead::banghead:

[roast]

You have several threads in which you seem to be discussing similar issues.
Use the search option to find the information you need, or stick to one topic.

[ERazer]

How much faster will the HD 6xxx "Northern Island" cards be? As opposed to the HD 5xxx.

dude there already thread about this or speculation about "nothern island" :slap: and im not gonna search it for u :nutkick:

[ShiBDiB]

Ignore him dude - he's american :toast:

Now wheres those moderators at to stop this 'Nationality' nonsense :slap:

haha its nothing to do with nationality, he's jus one of those posters who posts absolutely nothing helpful in a thread.

but anyway back on topic. bt do u have a link with the 485 specs?

[LAN_deRf_HA]

I hope they launch a dual fermi card. GTX 495 or something that beats the 5970 by a long shot that has lower temps and power consumption than the GTX480 or atleast about the same.

Never going to happen, only chance would be a die shrink and nobody is going below 40nm this year.

If a dual GPU Fermi launches, it will most likely be Two GTX 470 cores.

It's going to be two 465 cores.

[phanbuey]

I kind of want to wait until the refreshes.. but at the same time the 470 is looking pretty sweet. It would suck to buy it though only to have the 475 come out a few months later...

EVGA step up it is!

[roast]

haha its nothing to do with nationality, he's jus one of those posters who posts absolutely nothing helpful in a thread.

but anyway back on topic. bt do u have a link with the 485 specs?

Obviously it is, if you brought it into it.... disapointed in you shibby... :shadedshu
Lets leave nationality out of it.
I was just referring to the OP's other thread, where theres more discussion about various GPU's.

[Tatty_One]

This thread belongs in Graphics Cards > NVidia..... moved. Have a nice day! :cool:

[the54thvoid]

There's no way there will be an affordable 485 with 512 cores on the 40nm process. The whole reason the 480 is 480 cores is because the 512 version was unmanufacturable. This was primarily designed as a compute heavy card for the HPC market and the desktop part was to be a derivative. You can go research that to find that is the truth.

It took NV, arguably the most 'powerful' gfx manufacturer 6 months to get Fermi fixed and to the market and that was with the 'crippled' 480 core version.

Why do people speculate about a 485 version when it took half a year to realise they couldn't make the 512 core version without making a card insanely hot and insanely power hungry. We have a 480 core Fermi for a very real and practical reason - the 512 core part could not be made, at least not in significantly working numbers.

It needs a die shrink or a redesign to be able to make a 512 core variant. Do we honestly believe NV think, "ooh, we broke fermi, lets try our hardest to make the 512 core dream version work!" all the while ATI work on their next new'ish' architecture (southern/northern islands).
Dont be silly. NV got burnt in this round, ATI stole the PR and Fermi came out hot and hungry. NV will want to ensure that ATI dont trump them again and focussing on trying to make Fermi work like it was meant to is not a step forward.

Just my 2 cents.

one last point.....it would be impossible to make the 512 version a market contender. It would consume more power and be hotter and be more expensive if it stays on TSMC's 40 nm process. Die shrink or just please die.

New technology please.

[phanbuey]

There's no way there will be an affordable 485 with 512 cores on the 40nm process. The whole reason the 480 is 480 cores is because the 512 version was unmanufacturable. This was primarily designed as a compute heavy card for the HPC market and the desktop part was to be a derivative. You can go research that to find that is the truth.

It took NV, arguably the most 'powerful' gfx manufacturer 6 months to get Fermi fixed and to the market and that was with the 'crippled' 480 core version.

Why do people speculate about a 485 version when it took half a year to realise they couldn't make the 512 core version without making a card insanely hot and insanely power hungry. We have a 480 core Fermi for a very real and practical reason - the 512 core part could not be made, at least not in significantly working numbers.

It needs a die shrink or a redesign to be able to make a 512 core variant. Do we honestly believe NV think, "ooh, we broke fermi, lets try our hardest to make the 512 core dream version work!" all the while ATI work on their next new'ish' architecture (southern/northern islands).
Dont be silly. NV got burnt in this round, ATI stole the PR and Fermi came out hot and hungry. NV will want to ensure that ATI dont trump them again and focussing on trying to make Fermi work like it was meant to is not a step forward.

Just my 2 cents.

one last point.....it would be impossible to make the 512 version a market contender. It would consume more power and be hotter and be more expensive if it stays on TSMC's 40 nm process. Die shrink or just please die.

New technology please.


You're assuming a static process and no improvement of the design in terms of manufacturing. Sure, 6 months ago an affordable 485 was impossible... but theyve been working on it for yet another 6 months - on improving the 45nm mfg process, and on modifying the Fermi design to gain better yields.

Look at the difference between Intel's C0 and D0 chips - you CAN improve a chip drastically without a full blown shrink.

Yes it will be power hungry, and yes it will be expensive - but they will release a 45nm revision of fermi with all 512 cores IMO

[Monkey_Business]

You're assuming a static process and no improvement of the design in terms of manufacturing. Sure, 6 months ago an affordable 485 was impossible... but theyve been working on it for yet another 6 months - on improving the 45nm mfg process, and on modifying the Fermi design to gain better yields.

Look at the difference between Intel's C0 and D0 chips - you CAN improve a chip drastically without a full blown shrink.

Yes it will be power hungry, and yes it will be expensive - but they will release a 45nm revision of fermi with all 512 cores IMO

Maybe then the Fermi's won't run @ 70/95 *C.

[Binge]

You're assuming a static process and no improvement of the design in terms of manufacturing. Sure, 6 months ago an affordable 485 was impossible... but theyve been working on it for yet another 6 months - on improving the 45nm mfg process, and on modifying the Fermi design to gain better yields.

Look at the difference between Intel's C0 and D0 chips - you CAN improve a chip drastically without a full blown shrink.

Yes it will be power hungry, and yes it will be expensive - but they will release a 45nm revision of fermi with all 512 cores IMO

Maybe then the Fermi's won't run @ 70/95 *C.

45nm would be going backwards. It's already 40nm. The C0 to D0 did not change the thermals of the 920 at all.

[phanbuey]

45nm would be going backwards. It's already 40nm. The C0 to D0 did not change the thermals of the 920 at all.

right I meant 40nm... but they did increase the clock of the chip, and the d0 could handle higher clocks at a lower voltage.

[qubit]

This thread belongs in Graphics Cards > NVidia..... moved. Have a nice day! :cool:

It broke the links in my email alerts - gave invalid thread and I thought the thread had been pulled for some reason. Perhaps there's a bug in the forum software? I've unsubbed and resubbed so I'm ok now.

Perhaps W1zzard should know about this glitch?

[Binge]

right I meant 40nm... but they did increase the clock of the chip, and the d0 could handle higher clocks at a lower voltage.

That's not true either. The d0 revision had more chips that could reach a higher clock with lower voltages, but like the C0 I had it could do all the way up to 4.6GHz with the same voltages it took a D0 to get 4.6GHz. The only difference was most D0 could hit 4.0GHz at lower voltages. Still performance was the same. C0 was 2.66GHz, D0 was 2.66GHz same clocks.

[phanbuey]

That's not true either. The d0 revision had more chips that could reach a higher clock with lower voltages, but like the C0 I had it could do all the way up to 4.6GHz with the same voltages it took a D0 to get 4.6GHz. The only difference was most D0 could hit 4.0GHz at lower voltages. Still performance was the same.

http://www.tomshardware.com/reviews/i7-c0-d0,2386.html

http://www.tomshardware.com/reviews/i7-c0-d0,2386-3.html

http://www.techarp.com/showarticle.aspx?artno=655&pgno=1

Yah of course you can find an exception to the rule... I had a non-g0 Q6600 that hit XXXmhz. Thats great, but in general the D0 > C0 and E0 > C0 and G0 > B3 in terms of chip behavior with respect to clocks and voltages. They were all the same performing chips, but the latter tended to have better thermals and hit better clocks. Energy efficiency did improve with D0 as it ran with a lower voltage - intel kept the tdp the same, but they did generally have an energy efficiency advantage. If someone asked you if you wanted a tray of d0's or c0's which one would you pick and why?

Thats what they will do with the 485. Might not change the thermals a ton, but it will be able to run a higher clock with less juice, and it will affect thermals to a degree because the fermi is leaky right now if the next stepping decreases leakage even a little, the thermals will improve.