TPU: NVIDIA Readies Non-Ti GeForce GTX 560 To Ward Off HD 6790 Threat

[GodisanAtheist]

http://www.techpowerup.com/143238/N...Force-GTX-560-To-Ward-Off-HD-6790-Threat.html

NVIDIA is readying a new SKU based on the GF114 GPU: the GeForce GTX 560, to help strengthen the company's competitiveness against AMD's upcoming Radeon HD 6790, and probably higher SKUs in the HD 6800 series. Next week, AMD will unveil its Radeon HD 6790 SKU, which will compete with the GTX 550 Ti, and probably seat itself in the market somewhere between the GTX 550 Ti and GTX 560 Ti. It is this gap that NVIDIA is looking to fill.

-Looks like we're in for more than a few months of "war of the gaps" before 28nm hits the shelves. Nothing like both competitors making sure there is a sku at every 5 dollar interval.

Oh yeah and the naming. God the naming.

 

[bryanW1995]

Sure beats the alternative of zero competition and extremely high prices for little to no performance gain over the previous generation like we saw a couple of years ago. But yeah, the naming just plain sucks... I can't wait to seen an EVGA gtx 560 ++ SSC.

 

[Dark Shroud]

It's also a good way to use up lower performance chips.

 

[badb0y]

Wow, so an unecessary GTX 550 killer warrants an unecessary HD 6790 killer.

*high fives AMD and nVidia*

 

[bryanW1995]

get used to it guys. when 28nm slips into spring 2013 b/c tsmc drops the ball again, we're gonna start seeing a LOT of this kind of junk...

 

[Arkadrel]

http://www.techpowerup.com/143238/N...Force-GTX-560-To-Ward-Off-HD-6790-Threat.html



-Looks like we're in for more than a few months of "war of the gaps" before 28nm hits the shelves. Nothing like both competitors making sure there is a sku at every 5 dollar interval.

Oh yeah and the naming. God the naming.

+1 interwebs 2 u my friend.

 

[notty22]

I can't complain about the names. From either company your going to be confused unless you do a little homework.
gts 450, gtx 550ti, gtx 460se,gtx 460, gtx 560, gtx 560ti
is no more confusing than
5770,6790,6850,5850,6870,5870

edit;: I had to fix my 6's I got confused !

 

[Lonyo]

I can't complain about the names. From either company your going to be confused unless you do a little homework.
gts 450, gtx 550, gtx 460se,gtx 460, gtx 560, gtx 560ti
is no more confusing than
5770,6790,6850,5850,6870,5870

edit;: I had to fix my 6's I got confused !

Um, it is more confusing.
What's better, 560 or 560Ti?
460 1GB SE, 460 768MB or 460 1GB, and that's ignoring all the overclocked ones with their own suffixes.

Yes, 4xx vs 5xx and 5xxx vs 6xxx is equally confusing, but AMDs naming system within a generation is much simpler. Higher number = higher performance. Nothing to get confused about.

 

[GodisanAtheist]

I can't complain about the names. From either company your going to be confused unless you do a little homework.
gts 450, gtx 550ti, gtx 460se,gtx 460, gtx 560, gtx 560ti
is no more confusing than
5770,6790,6850,5850,6870,5870

edit;: I had to fix my 6's I got confused !

-There is no denying that the naming system from both companies is head scratchy, but because Nvidia uses only 3 numbers in their naming scheme they're forced to resort to suffixes (they really aren't, but things get cramped a la GTX 460/465/470) while AMD just keeps upping the numbers.

Comparing cards across generations is and always will be messy, but as far as comparing cards WITHIN one generation, I don't think its unfair to say AMD has the upper hand here (which isn't an endorsement of their recent naming gymnastics by any means either).

 

[notty22]

Um, it is more confusing.
What's better, 560 or 560Ti?
460 1GB SE, 460 768MB or 460 1GB, and that's ignoring all the overclocked ones with their own suffixes.

Yes, 4xx vs 5xx and 5xxx vs 6xxx is equally confusing, but AMDs naming system within a generation is much simpler. Higher number = higher performance. Nothing to get confused about.

How do you think a new buyer compares AMD and Nvidia products to begin with , after price comparisons. They have to do a little homework.
A important obvious Nvidia specification to compare within its offerings is cuda cores.

 

[cusideabelincoln]

If this GTX 560 is slower than a GTX 460, then I do not see the point. Would be another pointless card. The 560 non-Ti simply needs to just replace the 460. Releasing anything slower would be redundant since they have the 460-768MB, 460SE, and 550 Ti already occupying those other slots.

And I'm not a fan with leaving a suffix off of this card when they are already using them. Should call it the 560 SE.

 

[Skurge]

Is this going to be a 460 rebadge?

 

[Lonyo]

Is this going to be a 460 rebadge?

From the specs and story, it's a GTX460 with higher clocks, effectively.
So it will be slower than some of the factory overclocked GTX460s, but faster than a stock 460 and some factory overclocked 460s.

 

[notty22]

Is this going to be a 460 rebadge?

GF 114 vs GF 104

The GF 114 has tweaks that make it faster, better memory controller. Its why the 560ti gives the 6950 a run for its money.
But you knew all that

 

[Lonyo]

GF 114 vs GF 104

The GF 114 has tweaks that make it faster, better memory controller. Its why the 560ti gives the 6950 a run for its money.
But you knew all that

Actually, according to AT, there are no architectural differences between GF104 and GF114. The only difference is in transistor use to improve power and clocking.

http://www.anandtech.com/show/4135/nvidias-geforce-gtx-560-ti-upsetting-the-250-market/2

GF114 does not contain any architectural changes

But you knew that.

 

[bryanW1995]

Huh, gtx 460 isn't exactly lacking in the "clocking" department, and it's power usage is already very reasonable relative to its performance. If the standard gtx 560 is an improvement upon that then it certainly won't be a bad thing.

 

[Arkadrel]

we need 28nm....

So we can all get 6970's (389mm^2) that are about 240mm^2 big (on 28nm), and cost what LESS than a 6850, and performs like a 580, and uses less power.


^ just guestimateing for fun, but above probably isnt intirely unreasonable.
Going from 40nm -> 28nm is gonna be a big jump.

 

[notty22]

Actually, according to AT, there are no architectural differences between GF104 and GF114. The only difference is in transistor use to improve power and clocking.

http://www.anandtech.com/show/4135/nvidias-geforce-gtx-560-ti-upsetting-the-250-market/2


But you knew that.

I think you need to read the link you posted or maybe try to comprehend what is being said.

The GF104/GF110 Refresher: Different Architecture & Different Transistors

This makes it NOT a rebadge, WTH are you trying to say ?

In my post, I said 'tweaks'

and that it uses the same secret-sauce (read: electrical enhancements) that made GF110, an evolved clone of the GF100, totally rock with power consumption figures. 384 CUDA cores apart, there's a 256-bit wide GDDR5 memory interface, 32 ROPs, branched geometry processing, and the immediate fruition of the electrical enhancements, clock speeds: 822 MHz core, 1640 MHz CUDA cores, and 1000 MHz (4.00 GHz effective) memory.

 

[ViRGE]

I think you need to read the link you posted or maybe try to comprehend what is being said.

The GF104/GF110 Refresher: Different Architecture & Different Transistors

This makes it NOT a rebadge, WTH are you trying to say ?

The article seems to be referencing the architectural differences between GF110 and GF114. GF104 is indeed architecturally identical to GF104. It's not a rebadge because it is a different GPU due to the change in transistors, but it's otherwise the same.

 

[Lonyo]

I think you need to read the link you posted or maybe try to comprehend what is being said.

The GF104/GF110 Refresher: Different Architecture & Different Transistors

This makes it NOT a rebadge, WTH are you trying to say ?

It's a GTX460 with higher clocks... EFFECTIVELY.
Note the word "effectively".
Yes, it has a different GPU as the base, tuned to run faster, but the architecture is the same. GF104 is the same as GF114 from an architectural standpoint. The change that the transistors have been tweaked to allow better power efficiency/higher clocks/probably better yields.

The net result is that a GF104 and a GF114 at the same clocks with the same functional units are end user identical in performance, but differ in power consumption.
That means that a GF114 based GTX560 at some specific clocks is effectively the same as a GF104 GTX460 at the same clocks in everything but power characteristics, making them effectively the same from a performance standpoint.

Just like the G92 and G92b were effectively the same, although one was 65nm and the other 55nm, they were exactly the same outside that and its implications such as power use.

It is basically a rebadge. There are no architectural differences, it has the same number of functional units.
Was the 8800 -> 9800 a rebadge? Some 9800s used a different GPU. Still a rebadge.

 

[Idontcare]

Going from 40nm -> 28nm is gonna be a big jump.

HK/MG comes in. Could make things very interesting for us enthusiasts.

 

[tviceman]

It's a GTX460 with higher clocks... EFFECTIVELY.
Note the word "effectively".
Yes, it has a different GPU as the base, tuned to run faster, but the architecture is the same. GF104 is the same as GF114 from an architectural standpoint. The change that the transistors have been tweaked to allow better power efficiency/higher clocks/probably better yields.

The net result is that a GF104 and a GF114 at the same clocks with the same functional units are end user identical in performance, but differ in power consumption.
That means that a GF114 based GTX560 at some specific clocks is effectively the same as a GF104 GTX460 at the same clocks in everything but power characteristics, making them effectively the same from a performance standpoint.

Just like the G92 and G92b were effectively the same, although one was 65nm and the other 55nm, they were exactly the same outside that and its implications such as power use.

It is basically a rebadge. There are no architectural differences, it has the same number of functional units.
Was the 8800 -> 9800 a rebadge? Some 9800s used a different GPU. Still a rebadge.

I think by definition, a rebadge is taking an exact same chip and naming it something else. GF114 is not the exact same chip as GF104, and therefore is not a rebadge. The 9800GTX+ is EXACTLY the same as a GTS250. The same exact transistors, the same exact clock speeds, the same exact power consumption. That is a rebadge. GF114 is NOT exactly the same as GF104. The gtx560 will have different transistors, different clock speeds, and a different performance/watt ratio. Therefore it is NOT a rebadge.

By your definition, barts is just a rebadged cypress. Yes it has different GPU as it's base, tuned to run more efficient, but the underlying architecture is the same. Soooo?????????

 

[formulav8]

Figures nvidia will use a name to confuse customers yet again They should call it a 500, 555, 559, 550, or whatever. They thrive off of customer confusion. :thumbsdown: Them viewing consumers as a herd of cattle gets old...

 

[formulav8]

.

By your definition, barts is just a rebadged cypress. Yes it has different GPU as it's base, tuned to run more efficient, but the underlying architecture is the same. Soooo?????????

While I didn't care much from AMD's barts naming it is at least a very different core from the 5770's. More shaders, updated uvd, improved tessellation, core tweakes, ect... So no barts isn't near as bad as nvidia's rebadge. nVidia's is very much a rebadge to simply confuse/ripoff customers. :thumbsdown: Blah blah blah

 

[notty22]

One reason I'm not a big fan of this model, like the 6790, its actually going to help keep the prices up/stable on their big brothers, the 6870 /50 and gtx 560ti, you will see the new less powerful sku's fill the lower price segments the other models might have fallen in to.
But more choices often feeds price wars/bargains.