Nvidia GTX 680MX?

If that really happens amd can close up shop. lol

 

A mobile GPU with Super-Q and able to get hold of 255 threads, most utilization problems related to CPU speed could very well be solved... The current GTX 680M could only do 63 same as the old Fermi chips...

 

For a little bit more you could get an Ares II with two overclocked 7970 cores and built in liquid coolings system....

 

Close up the shop? If you know nvidia that thing will be insane pricey compered to AMD so there is nothing to be afraid.

 

If they can improve process and get TDP under control, it likely will be the case. That would require an 18% reduction in TDP to achieve 100W, which could be likely. Although from what I've seen so far I don't see the 680MX as a significant jump over the 680m that it would warrant an upgrade for current 680m owners.

 

Both you and me knows that 680M is far from 100W TDP.
580M was 100W, temperature wise the 680M is much lower. Same can be said about power consumption. Notebookcheck is full of themselves and I don`t trust them here. Nvidia havent disclosed any TDP information about 680M either.

Nvidia says:

I don`t think Nvidia have to sacrifice anything if they want to unleash a true 100W GPU like the 580M. 780M imo will be 1536 cores @ 720MHz and the memory bandwidth the same if not slightly higher.

The uncertainty here is how much Nvidia can squeeze out of the Kepler architecture. If its the same as Fermi, where we saw 15% better performance and 12W less TDP with the jump from GTX 480 to 580, you can start imagine how good the 780M really can become

 

On average it does consume less but TDP is talking about what the heatsink has to handle even at peak loads.

The reason the MX has such a big jump in TDP is the high voltage ram, so I think 780M will be a fully unlocked core with low voltage ram still. Hopefully there will be some faster low voltage GDDR5 by then (say 1000mhz) which would help a lot.

 

We don`t know what TDP limit the heatsinks in the various notebooks have.
They may very well be designed over 100W. Same for the chassis.
If anything, one should assume that its over 100W considering that our huge overclocks have never had too high temperatures or caused any failures. Even with the SVL/Saltius Vbios its around the 580M temperatures. And the peak TDP at these clocks are way higher than the peaks you would see @ stock clocks with the 680M. I`m positive that the 680M is lower than 100W. Maybe 85W.

The MX cards have different VRAM? Whats the benefit with high voltage VRAM vs low voltage? Yeah I`m beginning to think we will see 1000MHz for the GDDR5 since it must not choke the extra cores down on the 780M. We all have seen how stable it is at those clocks anyways all the way up to 1200MHz.

 

Low voltage ram goes up to about 1250-1300mhz.

Normal voltage ram (same chips but with a higher voltage) goes up to 1500-1600mhz.

 

And AMD's crappy driver support as a one time exclusive bonus. lol

Ohh wait! They include that with all their products. xD

 

While it's true the enduro situation is not great, the rest of their drivers are very much like nvidia's.

 

I thought their game support sucked too

Sent from my SPH-L900 using Tapatalk 2

 

Their single gpu drivers have improved but CFX support is still crap from what I've read.

 

so.. as per wikipedia (still subject to change) specs showing 680mx becomes the next 780m. I think I just lost a s*** load of respect for nvidia, giving us something half a year old that Apple already has their hands on while charging us premium, possibly the same TDP as well, load of bs man..

look up 680mx and 780m Comparison of Nvidia graphics processing units - Wikipedia, the free encyclopedia

 

Until the 780M releases we won't know. Pointless to point blame here and there until we know for sure. That means being a little patient. If it is the same chip we probably don't have long to wait.

Besides Nvidia have always charged a premium for their products. Not a new issue there. Their products generally work out of the box, work well and are supported well. They know many will pay for this. Also many never will.

 

wikipedia could be wrong and has been many times in the past, wont be surprised. rather its good to have it wrong so we can see some real deal. I was real stoked and wanted to go for nvidia for once this year around instead of AMD but if 780m = 680mx, not happening.

aint letting apple/mac camp look down on us

 

Lol well don't count on AMD getting up on their high horse if Nvidia don't. AMD doesn't have as deep pockets as Nvidia I am pretty sure so they won't push out new tech to compete with...last gen tech.

From what I have seen this gen. For the ultimate high end the 7970M was VERY good performance for value. The 680M especially with OC is also fantastic but really expensive. Some can justify it I can't.

The real winner from nvidia I think is the 675MX. With a lower tdp than the older 6970M it smashes everything below the 2 top cards with an OC. With the lowly 920XM it scored 6300 points in 3dmark11. With a modern CPU it could hit 7000 points! That is not very far behind the 7970M at all!!

 

Link to 6300 3Dmark sscore? Didn't think it would get anywhere close to that.

 

I told you it overclocks like a beast.

 

NVIDIA GeForce GTX 675MX video card benchmark result - Intel Core i7-920XM Processor,Alienware score: P6332 3DMarks

Here we are! Running in an M17x R2. Reborn's machine

 

Damn, you guys weren't kidding! Reborn needs to go sli with those suckers; he'd break 10k!

 

He would need a 330W psu for that though. Does that work on the r2?

 

Oops, forgot about that lol. Doesn't the 675mx have a lower tdp than the 680m? I guess it would still be too much for a 240W. He could go with daisy-chained 240W power supplies

 

When you overvolt a gpu it can consume drastically more. The higher clocks you set the higher voltage you need thus increasing consumption. The 675mx will draw more power than a stock 680m to reach it's performance level. This is why nvidia moved toward having more cores at lower clocks with Kepler. This will most certainly continue with Maxwell. You can expect even more cores at even lower clocks.