• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

nordichardware: GTX 590 in February

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Last edited:
Castiel said:
So i was just reading over at the EVGA forums.

EVGA is releasing a dual GF114 card and Nvidia is going to release a dual GF110 card. Looking like Nvidia will beat EVGA to the market but..

590 = Dual GF114, 2GB Memory
595 = Dual GF110, 3GB of Memory

HOPE this info is correct.

http://www.youtube.com/watch?v=WpLDuxRWgj0&feature=player_embedded
http://www.youtube.com/watch?v=Vf_7XpJEeOM

AMD... Nvidia has left you a Tweet, Voicemail and text

http://www.youtube.com/watch?v=mWLO4acMTCM
Click to expand...

Is that the memory for each GPU?
 
Skurge said:
Is that the memory for each GPU?
Click to expand...

I believe is for the Graphics Card.

On GTX590,

If they can manage to put 2x GF110 with 512 Cuda Cores then the core frequency would be close to 650-675MHz, so that means close to GTX570 SLI performance (i would say 90-95%).

In order to keep the 384-bit memory controller and 3G total memory they will clock the memory low, i would say close to 800-830MHz
 
AtenRa said:
I believe is for the Graphics Card.

On GTX590,

If they can manage to put 2x GF110 with 512 Cuda Cores then the core frequency would be close to 650-675MHz, so that means close to GTX570 SLI performance (i would say 90-95%).

In order to keep the 384-bit memory controller and 3G total memory they will clock the memory low, i would say close to 800-830MHz
Click to expand...

If they do this, I hope they don't use lower spec RAM. I hope it's just clocked lower by default. And this card really ought to be a 6GB card. I can't wait to see what kind of non-reference designs AIB's come up with. :biggrin:
 
Though I am certainly no "fanboy," I think it will do good to have more of a selection, or at least a choice for dual-gpu cards for this generation of DX11 cards. AMD makes a fince Dual-GPU solution and it's upcoming will assuredly be an improvement. However with the lack of competition, there is no price war, and it is is better to compare apples to apples when benchmarking.

As many of you have pointed out the upcoming dual Fermi card will certainly be a downclocked cousin of the 580. If you go back generations of card from nVidia you will see this has been the case. The GTX295 was not a dual 285. It was a hybrid between that and the 260. Slower clocks than the 285 and the memory size and band width of the 260. It was such a good combination that nVidia split it in half and called it a 275. Go back one generation and look at the 9800gx2. Slower core and memory clocks. all the way around. And before that was the 7900gx2. Not a popular card but the story was similar.

Now imagine for a momnet this generation. If you look at some of the differences in the 580 vs 570 vs 560, you may be able to piece together a good speculation as to it's specifications.

512 cores @ 600-650 mhz
Shader clock 1400-1500 mhz
Perhaps only 1280mb ram with a 320 bit memory interface running at 3700-3800mhz

I certainly think it will have 2 eight pin connectors. And it lill be less than 800 dollars.

It's just my thoughts and I could be way off but it seems the logical conclusion.

Jason
 
Zerocarestate, the Fermi cards run sp at twice the core clock. if the core is 600 or 650 then the sp would have to be 1200 or 1300.
 
ugaboga232 said:
lol How does this fit into PCI specs? Super low clocks?
Click to expand...

Cherry picked 580 cores, slightly lower clocks. I am sure they can make this happen. nVidia, to their credit, can seem to pull a cat out of the hat when they really need too. I think that this dual GPU will be a great performer, but I don't expect it to be a wise choice in regards to performance/fps.

I expect this thing to be no cheaper than $799, and even then I am skeptical. I would also expect limited quantities, but then again, with a more limited market as well. So, they should be able to provide ample stock for those who want one.
 
Skurge said:
Anyone know what PLX chip they will be using? Sapphire improved the performance of the 5970 toxic not just by higher clocks, but with a better PLX chip (I think that gave it the edge over the ARES). Not sure what the differences were, but it seemed more robust.
Click to expand...
Sapphire also used a better fan system and probably a better vapor chamber system than Asus as well. Sapphire has a lot of experience with those systems.
 
I'm still not sure how they'll fit 2 580's inside the 300w pcie spec. I figure they'll have to downclock it so much that it's going to defeat the purpose of using 580's instead of 570's.
 
-Slacker- said:
I'm still not sure how they'll fit 2 580's inside the 300w pcie spec. I figure they'll have to downclock it so much that it's going to defeat the purpose of using 580's instead of 570's.
Click to expand...

I also don't see it happening. The 580 uses the same amount of power as a 480 anyway.
 
-Slacker- said:
I'm still not sure how they'll fit 2 580's inside the 300w pcie spec. I figure they'll have to downclock it so much that it's going to defeat the purpose of using 580's instead of 570's.
Click to expand...

Everyone mentions this yet the ones who plan on buying the cards can care less.
 
Which proves my point. Both are over 300W spec if you combine them, but your probably going to see two GF110 cores used anyway. Plus the cores that will be used is probably going to have a much lower TDP than your regular run of the mill GTX580.
 
Cookie Monster said:
Which proves my point. Both are over 300W spec if you combine them, but your probably going to see two GF110 cores used anyway. Plus the cores that will be used is probably going to have a much lower TDP than your regular run of the mill GTX580.
Click to expand...

Actually the gtx 295 was just under 300w and they had to downclock the core, shaders and memory.

Since the 580 eats up 25% more than a 275, then it's reasonable to assume each 580 chip on the 590 will have to be downclocked even more to fit the specs.

To that end, I'd have to wonder if going with gtx 570 chips instead would be a better idea.

Of course, I don't know jack about GPU R&D, so I'm just guessing here, but, again, looking at all the past dual gpus, It seems fair to ask how in hell is nvidia going to pull this one off?
 
Castiel said:
Everyone mentions this yet the ones who plan on buying the cards can care less.
Click to expand...

I don't think it's people's concern over being green or frugal that cause them to mention this. Speculation about TDP is relevant because Nvidia and AMD need to stick under the 300W TDP limit. There have been custom cards that go above and beyond that spec but, AFAIK, the reference cards have always met that limit.

So if Nvidia and AMD continue to comply with the TDP spec, it will limit which gpus they can put on their new dual cards. That's what all the speculation is about. What can each company feasibly use from their current line-up? AMD's choice is fairly concrete at this point, somewhere in between a pair of 6950's and 6970's. Nvidia, who knows? Probably a pair of GF110 gpus but that still leaves a wide range of performance.
 
-Slacker- said:
, It seems fair to ask how in hell is nvidia going to pull this one off?
Click to expand...

With vodoo volts!

3dfx_voodoovolts_display_ensemble.jpg
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top