• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

** ALL HAIL THE NEW HEAVEN KING - ATI 7970 3GB!! (TRI-FIRE TESTED)**

Nobody can question those scores, but I a little disappointed with AMD's modest (20%) gains over NVidia's old GTX580.

When you consider that the 7970 GPU and memory are 18.5% and 37% faster respectively, plus the 7970 contains 33% more transistors than the 580, it should be much faster. It seems that most of the new 28nm process gains have been achieved by clock speed increments rather than by inovative design. Perhaps all NVidia have to do to beat AMD (cheaply) is shrink Fermi to 28nm and increase the clocks to the same speeds. That way they will have a much smaller die providing the same performance, running on a very cheap 2 year old technology.

I can see Kepler (with double the core power of Fermi, plus higher clocks) crushing this card, but whether they scale as well with 2 or 3 cards in SLI is more questionable.
 
Last edited:
555BUK said:
Nobody can question those scores, but I a little disappointed with AMD's modest (20%) gains over NVidia's old GTX580.

When you consider that the 7970 GPU and memory are 18.5% and 37% faster respectively, plus the 7970 contains 33% more transistors than the 580, it should be much faster. It seems that most of the new 28nm process gains have been achieved by clock speed increments rather than by inovative design. Perhaps all NVidia have to do to beat AMD (cheaply) is shrink Fermi to 28nm and increase the clocks to the same speeds. That way they will have a much smaller die providing the same performance, running on a very cheap 2 year old technology.

I can see Kepler crushing this card, but whether they scale as well with 2 or 3 cards is more questionable.
Click to expand...

Kepler wont crush it whenever it might come out.
People seems to have little understanding about the engineering obstacles at 28nm. Even if they get a 30% from 580 they would do a good job and match the 7970.

I expect around 10% more from driver development the next few iterations due to gcn is brand new for them.
 
555BUK said:
Nobody can question those scores, but I a little disappointed with AMD's modest (20%) gains over NVidia's old GTX580.

When you consider that the 7970 GPU and memory are 18.5% and 37% faster respectively, plus the 7970 contains 33% more transistors than the 580, it should be much faster. It seems that most of the new 28nm process gains have been achieved by clock speed increments rather than by inovative design. Perhaps all NVidia have to do to beat AMD (cheaply) is shrink Fermi to 28nm and increase the clocks to the same speeds. That way they will have a much smaller die providing the same performance, running on a very cheap 2 year old technology.

I can see Kepler (with double the core power of Fermi, plus higher clocks) crushing this card, but whether they scale as well with 2 or 3 cards in SLI is more questionable.
Click to expand...


It's 20% faster in a optimised benchmark, real world performance ie games etc its on average 8 to 10% faster.
 
flopper said:
Kepler wont crush it whenever it might come out.
People seems to have little understanding about the engineering obstacles at 28nm. Even if they get a 30% from 580 they would do a good job and match the 7970.
Click to expand...

Well then, please, do enlighten us...

As far as I can see, the R+D effort, as well as a large proportion of the extra transistors, have gone into enhancing the compute capabilities of GCN. These adaptations inevitably cost transistors and die-space - much as they did with Fermi. As far as I can see, the only major question mark surrounding the 28nm process is the out of character increase in transistor density (+74% vs +105% expected- see here). This is more likely due to redundancy or design issues, rather than the TSMC 28nm manufacturing process. But still, even if Kepler can only achieve a ~ +80% increase in transistor density, I can't see it falling behind the 7970.
 
A die shrunk GTX580 running at 7970 speeds would likely be equal to the 7970. Even if Kepler is extremely non-optimised, it should still contain double the core power (stream processors, ROP's, texture units) of the GTX580 and run at higher clocks. Given NVidia's past history of >60% performance gains over previous gen cards (8800 to 280 to 480), I think Kepler will be closer to 7990 performance than it is to the 7970.

It is possible that NVidia will release a crippled Kepler at launch (like the GTX 480) to increase yeilds and allow for a later respun top-dog card (like the GTX 580). Whatever happens, Kepler will probably be significantly faster than the 7970, hopefully forcing prices of 7900's down to reasonable levels.
 
Last edited:
Just found this, sorry if it's already been posted:

Untitled-5.png


http://wccftech.com/nvidia-kepler- [...] h-q2-2012/
 
SamLB said:
Just found this, sorry if it's already been posted:

Untitled-5.png


http://wccftech.com/nvidia-kepler- [...] h-q2-2012/
Click to expand...
What does this spreadsheet and Jordan's boobs have in common?

GK100 is widely rumoured to have 1024 cores and 512bit memory.
GK104 is widely rumoured to have 512 to 768 cores and 256 to 384bit memory.

...but we will only know for certain once the reviews hit in 3-6 months time. One thing that is for sure is that NVidia will have not set clock speeds yet, so the above spreadheet is fake.
 
8800/9800 (128 cores / 754M transistors) > GTX280 (240 cores / 1.4bn transistors) > GTX580 (512 cores / 3bn transistors)

1024 cores would continue the trend for GK100, but transistor count would be 5-6bn. NVidia do have history with huge cores on therir top cards, but perhaps they will be conservative this time around and concentrate on performance per watt. NVidia should be able to fit approx 5.5bn 28nm transistors on the same die size as the GTX 580.
 
Last edited:
tellmewhy99 said:
Sapphire 7970 - RM2050 = 410 pounds without tax
PowerColor 7970 - RM1950 = 390 pounds without tax

confirmed in my country now
i would rather take gtx 580 now lol
Click to expand...

If it does turn out to be over £450 it will be a sad joke, the initial lot will sell obviously but after that they will just sit on shelves. I suspect a price cut down to sub £400 very soon after launch.
 
That document is fake because:-
i). MemBus and MemCapacities are incorrect for GTX680 & GTS640. The numbers simply will not work together.
ii). NVidia will not know clock speeds until they have working samples, so why leak the unknown?.
iii). Pricing cannot be established so long before yields are known (simple demand and supply rules will apply at launch).
iv). Document is unprofessional, mis-alligned, and clearly a botched excel jobby.
v). Since when has NVidia launched cards with such orderly naming and specification? It leaves no room for consumer confusion (also no room for a dual GPU 6xx card).
 
Last edited:
555BUK said:
/snip
Click to expand...
Hang on a minute! Are you saying everything on the internet isn't 100% true?

Personally, I love the platitudes all the Nvidia fanboys are trotting out.

In reality people will buy the fastest card(s) for the money, up and down the price range. Why is the badge an issue?
 
You must log in or register to reply here.
Back
Top Bottom