nVidia's 28nm Mobile Lineup Leaked

According to Semiaccurate, these will be Fermi die shrinks (not Kepler), which isn't a bad thing at all if the 3dmark Vantage projections are accurate:

Nvidia’s 28nm mobile lineup leaked | SemiAccurate

 

20000 for the 680m gtx sounds interesting. pretty big leap from the 580m gtx at 13000. but then again its a ways off

 

Why no 3DMark 11?

Oh well, guess they cant leak it all.
Hopefully the 28nm cards will be accepted by the current gen Sager / Clevo notebooks.

 

That lineup looks false.

Nvidia love to hack off a memory bus for their slightly lower card (580m vs 570m) to give a 192bit card, and I doubt they are going to drop it either as a bus for their other cards.

 

Even if it is accurate it doesn't look like much of an improvement besides the card predicted to release in almost a year. I would certainly hope there would be at least a 20% increase in power from the 580m by that time.

 

lol the last one is 685m and the second last one is 680m, duh.

 

but theirs no 585m coming out what makes you think the next gen will have one. duh

someone said its not that impressive and hopes it will be at least 20 percent faster than the 580m gtx. well since when did the 580m gtx hit 16000 stock. this is news to the world.

btw the top score was estimated at 20 000 which is a 60 percent increase over the current gen.

 

gaminacu

congrats on purchasing the m18x im sure that beast will last 5 years on medium settings, until the new consoles come out

when you get your laptop can you overclock it andrun crysis very high benchmark and post a thread on it?

 

Yea, I'm going to start OCing it once I get it.

 

fake garbage is fake.

 

They "leaked" the AMD 28nm cards too.
Probably fake

 

OMG that's even worse.

A lowend and midend chip consuming the same power and performing the same between a pro and XT? Yeah right.

 

Never Say Never

/OT

I agree, these are probably not legit.

Isn't that site known to be a little less than semi-accurate?

 

But who would buy a more expensive equally performing and power consuming chip?

 

the high end chips consumed less power.

 

But it wont, a design targeted at a certain power consumption will always perform better than a larger design clocked down.

Think GTX480m vs GTX485m.

 

28nm, Die shrink.

Meaning less power consumtion, more performance/watt.

 

Those slides are probably being leaked by someone within TSMC.

 

In order to do what, increase share prices?

 

Is this anywhere else on the web?

case(s) in point, semi-accurate seems to be offline right now.

 

At least the Vantage scores seem more realistic than Nvidia's.

 

Chelsea is a mainstream part that's rumored to be faster than Whistler(HD6700M) and Heathrow is a high-end part. Thames is the codename for the low-end part that's missing from S|A's list.

Cutting down a high-end card to top off the mainstream line, or having a hypothetical HD7830M that is only slightly better than an HD7770M are both believable scenarios. There's also the possibility still that Heathrow and Chelsea aren't manufactured at the same Foundry...in which case having two sources produce overlapping cards, for what's usually a very popular market segment, is good for supply.

To be honest though, most of the information in S|A's leaks isn't all that revealing and not worth worrying about it's validity. Bus width, memory, and TDP are all relatively easy to predict for mobile parts....but predicting cores and shaders is much harder and there's no info given for those counts.

Nvidia's chip names aren't exactly a mystery either, as they've followed the same pattern for generations, and as far as 28nm being a die shrink of Fermi, S|A wasn't the first with that reveal.

Nvidia's scores aren't necessarily Vantage scores. The column on the leaked Nvidia chart is titled 3DMark and there's no letter in front of the scores indicating what level settings the benchmarks were run at. On the AMD chart the column is titled Vantage and the scores are preceded by a P, making it more obvious which benchmark those scores belong to.

 

Can't say it's a big surprise, Nvidia wrung every last drop out of the previous chips when they were struggling to make Fermi work, if Kepler is having difficulty, it makes sense for them to fall back to what works now to have 600 series product to release in 2012 with Ivy Bridge notebooks.

Could be a great thing too, if a 680M is faster than a 580M but has the 75W TDP of a 560M. Nice upgrade.