some info about upcoming refresh

from a guy who attended that meeting, will keep update: (google translated)

Just attended the meeting CLEVO foundry in Kunshan notebook
CLEVO the end of 2011 there will be an upgraded version of the machine X7200 launch 17-inch dual card 580M
In early 2012, there X7200 upgraded version of the 3D version of the machine but does not seem to the naked eye
The most fun is the face of battlefield requirements of 3 abnormal release code for NVIDIA's GTX660M N13E-GTX notebook graphics
Sub-memory version of the 2G and 4G, 4G memory one 3Dvantage scores P21000, 2G P20000
The low level version of the GTX650M performance in P17000

Scores than the current GTX580M high number P16200
The GT640M is said to belong to this fraction in the P13000 midrange graphics card
Another new video card, but AMD has rotten performance
HD7900M I see above is the announced P15500 P14200 than it is now HD6990M the higher
So rich enthusiast brothers or start GTX660M

Also ask the Taiwan engineers said that now P150HM P170HM P180HM can be upgraded to the latest graphics cards
The interface is still the same Oh now we look forward to it of course also good drops Oh GTX560M

___________________________________________________________________________________






note nvidia physx so x0.85=overall score..........but still beat 7900m

___________________________________________________________________________________
现在，在华硕的295.18的OEM驱动里，就有GTX670M的DEV号！

http://download.laptopvideo2go.com/nvidia/295series/29518_win7x64.exe

NVIDIA_DEV.1210.01 = "NVIDIA GeForce GTX 570M"
NVIDIA_DEV.1210.02 = "NVIDIA GeForce GTX 570M "
NVIDIA_DEV.1213.01 = "NVIDIA GeForce GTX 670M"
NVIDIA_DEV.1213.02 = "NVIDIA GeForce GTX 670M "

1210是GTX570M的设备号，1213是GTX670M的设备号！很奇怪的是，它竟然和GTX570M在一起出现了。。但再看这里，就更奇怪了。。

%NVIDIA_DEV.1210.01% = Section066, PCI\VEN_10DE&DEV_1210&SUBSYS_14871043
%NVIDIA_DEV.1210.02% = Section118, PCI\VEN_10DE&DEV_1210&SUBSYS_21041043
%NVIDIA_DEV.1213.01% = Section066, PCI\VEN_10DE&DEV_1213&SUBSYS_21191043
%NVIDIA_DEV.1213.02% = Section118, PCI\VEN_10DE&DEV_1213&SUBSYS_21201043

要知道华硕的295.18的OEM驱动里，全是华硕笔记本显卡用的！而华硕的G74系列从没有用过570M显卡！！

那为什么会出现570M呢？？再看下上面的规格图，你是否能明白点什么了？

也许GTX670M就是gtx570m马甲！

是的，现在的假设就是：GTX600M全是上一代GTX500M的28NM马甲！那么GTX660M的规格呢？

如果说GTX570M可以通过高频来一次马甲，那已经是GTX460M马甲的GTX560M怎么来马甲呢？

目前为止知道的还有GTX675M的设备号！1212！它是否是256BIT的570M？或是GTX580M的马甲？现在还不知道！

[DEV_1210] NVIDIA GeForce GTX 570M
[DEV_1211] NVIDIA GeForce GTX 580M
[DEV_1212] NVIDIA GeForce GTX 675M
[DEV_1213] NVIDIA GeForce GTX 670M

GTX680M的DEV号现在仍然未知。。。

 

so btw the vantage p is physx on, gpu score may be 18000....nearly same as gtx580m highest oc
it seems nvidia no longer use x80 for flagship......but 4gb memory!!!!!! i decided to skip amd 4g one which saw on leaked info
hope the x7200 refreash 3d version still have full oc ability, dont become next p170hm3 which locked oc tweaking

 

And what about the P270WM?

 

I believe the X7200 'refresh' as the source calls it is very well the P270WM--it just isn't labeled that specifically.

 

I thought so, but whether it will be with a new X79 Chip?

 

So MXM 3.0 has been confirmed. I'm glad I was wrong.

 

Does this mean that people with 6900-series AMD should be able to upgrade to the 7000-series ones?

 

This quote.....

Seems to point to yes.

But interface aside, it will also come down to whether a BIOS upgrade is needed, and whether Clevo throws us that bone.

 

It certainly uses the X79 platform because it was seen on the spec sheet at CompuTex. The only question that remains is which features will be in the final product, and we won't know that until launch day of the P270WM.

 

I guess an unnecessary but expensive main board swap would be needed. We would be told only the new revison MB could support those new GFX cards( just because clevo plays tricks in bios and vbios making old rev MB not support)

That is exactly what they did to D901C/D900F users who sought upgrading to 8800MGTX/GTX480M.

 

Lol, if the new cards are good enough, I might consider doing that. But if the cost of doing so is close to half of simply buying a new laptop (with the same specs), I'm going to have a quandary on my hands...

Mr. Mysterious

 

I expect the usual prices. Meaning ~$800 for Nvidia and ~$500 for AMD.

 

Now that I think about it...$500 for a 3x increase in GPU power is....pretty darn good

Mr. Mysterious

 

I do not expect any BIG jumps in GPU performance this next generation...I kinda hope I am wrong though.
-

 

thats right gtx660m and 7900m no doubt will be only 28mm shrink but same gpu architecture........but 660m stock=560ti, and oc can catch up 580m sli/6990m cf

 

Isn't one of AMD's supposed 28nm chip holding like 1400 shaders at just 120W? If we see that go mobile, the jump will definitely qualify as a little bit more than BIG.

 

Possibly slight performance increase but with much better power consumption? Seems to be the trend with the CPU market these days. GPU following suit?

 

No, both will be new architecture. 660M will probably be Kepler, which was based on the original FERMI, not the new ones. 28nm will allow Nvidia to do with without the power/heat issues they had before.

AMD will be using the VIPW4 architecture which was only seen on the HD6970 desktop. Or... if we are lucky, the Next Core Gen architecture which will make it very similar to a FERMI. Either way, it will be a new architecture over what is seen for mobile right now.

 

is 660M optimus?

 

very interesting , I hope the cards which come out are as good as these scores! probably will be upgrading again next holidays! any news about p170m refresh????

 

No, both will continue use current architecture for sure, and 6xxm=28nm die shrink.....sigh..........

 

Either that or a giant leap in performance? I mean at some point, they should be able to discover or uncover something new regarding the power output and power comsumption ratio. Cant wait tbh! Who knows what will be released

 

Some 6xx m even remain 40nm(e.g. 630m is a rebranded 540m.)

 

I'm confused, from what I read above, it looks like the GTX 650M > GTX 580M. But other comments say GTX 600 series will not be a performance increase, just a die shrink to 28nm. I know I read something about Kepler being some 3x boost in performance or in less power use. I'm not sure, it was an image from a presentation posted on Engadget. So, is the 600 series just a die shrink, a performance increase, or both? One of the things I'm holding out for (not just Optimus, Backlit KB, and Ivy Bridge, but also nexgen GPUs which I assume from Nvidia was going to be Kepler as the 600 series). Can someone set my facts straight please... I'm so confused.

 

Trust me guys, the biggest leaps in performance are not architectural changes, but from die shrinks.

http://forum.notebookreview.com/sager-clevo/620757-gpu-arhitecture-performace-280m-vs-570m.html

 

yes die shrink and it improve the performance per watt, which is important in notebook platform......
architectural changes do more in desktop...like 9800gx2 vs gtx280, they performs almost the same but gtx280 is "single gpu", thus can build 3way sli beat 9800gx2 which can only 2 in sli(quad sli)

 

Exactly! (10 char)

 

Like what Black and DG said, a die shrink is a huge part in the power game. The mobile cards cannot drain infinite amounts of power from a PSU like desktop cards can, and it's limited by the MXM3.0b architecture too, I believe. It's an optimization game. How well can we make this run using X base architecture (Fermi), Y maximum power drain (100 or 120W or so, I forget the actual limit), while keeping maximum usage heat under Z temps (call this 80) in a hot-ish environment like florida/caribbean/mexico/etc. They did their limits with the G92b core of the 285M, and tried GF100 fermi. Then they tried GF104, which was a smaller core, and accidentally broke the flagship GF100 model's performance. The GF114 etc models are just better optimized again for power consumption/heat, which makes them able to run even higher clocks.

If they could bring a die shrink out on that base architecture, that's less heat/power per cuda core. Therefore more cores/clocks can very well be a possibility, and the GTX 6xx high end cards might even reach the level of a GTX 580 in desktop world. I don't think it's too far fetched.

 

WoW, imagine the performance of a Desktop 580/590 in a 15.6" Notebook!!!!

But, The prices... (X_x)

 

not too sure about performance and release dates, but i do have the list of models available from clevo from here: Clevo | Channel Authorized Manufacturer Showcase

 

Seems legit to me.

 

I dont get it! do all you guys dont understand anything at all? when you go from 40nm to 28nm as part of the less heat generated and a million other reasons why they do it is you can have more transistors in the given space,less power and all that. so you will have a performance increase! no matter what!!! they are not gonna shrink!!! even it is rebranded it usually comes with a 15% increase 460m to 560m!!!! and million other examples!

 

so am i correct to say that moving from 40nm to 28nm will theoretically result in both lower power consumption and better overall performance due to an increased in transistor count over the same area size?

 

The most important question is still on whether Nvidia will come off of their insane pricing structure. I can see them charging a hell of a lot for the GTX 680M.

 

The 680M will probably cost as much as the 580M's launch price. If they go much higher, they'll simply lose sales. And as they find easier ways to do the same 28nm process, it becomes more cost effective, and eventually they throw in price drops. =D.

But their pricing, as you must know, is due to the CUDA and PhysX technologies, rather than raw graphics performance. CUDA saves me HOURS when I get to use it (I can't use it often because I render in 1920 x 1200, and few editing programs output at that size, but with a new 1920 x 1080p native monitor machine, I would use it 24/7) and PhysX is a nice addition to certain games that use it, like Batman, Metro 2033 and Alice: Madness Returns. That's the real reason for the high premium on nVidia cards. You could probably include their pristine driver updates/support as of late as well, but that was a recent development and not part of the original price premium.

Honestly, if someone told me they would never do video editing and only play FPS/MMO/RPG games, all of which are very unlikely to use PhysX, I would tell em I don't blame them for taking an AMD card. And it's the truth, if you only want graphics power, why not get a cheaper card with rival performance? I personally do a lot of videos and I enjoy SP and games where PhysX is present, so nVidia is the correct choice for me.

 

They wont be increasing the price, should just be the same as now. ivy with nvidia/amd 28nm, momentus xt 2 , clevo refresh with a killer new design! I cant wait for the next gen stuff and I havent even received my new laptop yet! lol

 

I'm not keeping my hopes up in terms of that big of a performance boost...as long as the power consumption is lower and the new cards run significantly cooler..

 

Correct. If you think about it, the transfer of information basically determines on how fast bits of information can travel from one place to the next. Shrinking the dye therefore means the information has less distance to travel, giving it a performance boost.

I took a class about astrophysics and this was brought up as part of a discussion about light speed and whatnot, really interesting stuff.

 

yeah.. thanks for clearing things up for me!

 

"Like the AMD 7000M launch, GeForce 600M looks to be more about marketing and product positioning than anything."
AnandTech - Introducing AMD?s Radeon 7000M and NVIDIA?s GeForce 600M Mobile GPUs

 

So.. I'm confused... Will my new AMD 6990 be basically obsolete in a few months when this refresh occurs?

Will there really be a 3x performance boost for basically equally cost to todays prices?

 

No, you probably won't see a 3x increase in just a few months. While it may be technically outdated in 3 months, you probably won't be that far behind. I have a 485m in my P150HM which is ~8 months old. It's been superseded by the 580m in that time, but it only lags behind about 10%. For a 3x difference, you'd have to have a card that is several years old, or is being compared to a whole different class (mid-range card compared to high end).

 

Thanks... wheww.. I can live w/ that.

 

I think, if you really want to see big increase in perfomances and you already got high end card and processor from 2011, you'll have to wait until 2013.

 

I expect the 7970M and 680M to be hmm.... a 50% leap over current high-end offerings.

6990M is 1120sp, 715/900? I'll be surprised if the 7970M doesn't pack 1400sp, with clocks around 850/1000.

 

I can put my money down at 30% increase minimum, 28nm sounds exciting! having worked with nano graphene particles for my Li-ion battery paper which got published, I did smaller and smaller tech, it packs on so many improvements!

 

All I have to say is with Ivy Bridge's slight boost and smaller process, Optimus, and Backlit keyboard, if it is true that the GTX 650M > GTX 580M in performance, then I'm set and definitely wanting that. May do 660M just for the slightly better future proofing. Can't wait for actual news to hit though.

 

If this is the case, what do you expect from the 7990M?

 

It's hard to say, with the desktop chips being kept so under wraps. Once they release details on the high-end desktop chips, I'll be able to point out exactly which chips can be what on the mobile side.

There's no guarantee we'll see a 7990M, depending on how the performance tiers of the desktop chips are spaced out. There may not be a 6850 (6970M) -> 6870 (6990M) pairing available.

 

If we would assume that the performance tier is the same as now, slightly more balanced maybe and probably a lot stronger ofc, but really, if the 7990M would to be released, it should be a monster, presuming that AMD continues to develop as they do.