Clevo notebooks with 800M series coming out February 2014

Oh I know their history pretty well.
I guess I like giving them the benefit of the doubt too many times

 

Nvidia Maxwell coming in Q1 is GM117

 

Oh, I see. The 800M is the more powerful GPUs, right?

 

Maxwell should/will be the most powerful, just when it's available is still in question. 800m will be the next version and more or less is same as 700m series with some slight clock increases.

 

GM117 ? means GT 840M is gonna be a bad low end 64bit maxwell ? same perf of the GT750M OC?

 

Out of the semiaccurate, into the Fudzilla

 

Thats what we get. Fishy sites like those and chinese forums. Its up to us to separate whats BS and what could be plausible.

You don`t expect Anandtech to write about this do you? They will, but not until Nvidia giving them the clear and the launch is like on the same day

GM117 is meant to replace GK208 yes.

GK107:
384 cores, up to 128bit, 16ROPS
GK208:
384 cores, 64bit, 8ROPS

I have a feeling GM117 will be 64bit too but who knows, could be 128bit as well

 

I don't know if nVidia are dumb enough to think the customers are dumb enough to fall for their crap, or if they're plain stupid.

They will probably be forgiven (maybe even by me, and even then only at the lower end) if the GT 840M is a 128-bit (hopefully GDDR5) card with the performance of a OC 750M. Then, it can be used to give ultrabooks a nice boost in the graphics department.

Before the actually new cards come out in June/July, I'm looking forward to the 865M (hopefully - 960 or 1152 shaders, 880 core clock with 940 boost, and [for the love of god please] a 192-bit bus and 2GB GDDR5). Put that in a 14"-15.6" (with sorted out thermals) with a quad-core Haswell i7 and put it on Xoticpc, GenTech or any of the other boutique guys, and you have a recipe for success.

 

hey cloudfire, lets hope thats 20nm. can't wait some serious OC and extra cores in that card, imagine that or the next gen goes with AW or clevo's notebook with 16core CPU? yeahh

 

It actually could be 20nm. If they start with only GM117 and only use that chip on 2-3 GPUs, they can easily meet demand with 20nm production. But who knows..

Anyhow, I think GM117 will be this one I guessed about earlier.
They can do GM117, GM116, GM114, GM110 as names this round. Impossible to guess.

 

The ASUS G750JZ has also been confirmed to come with a GTX 880M 4GB. They were already showing it off at CES.

 

Asus have really been firing on all cylinders lately. First G750 with GTX 780M, now GTX 880M. Looks like they found a design they are happy with and been meeting their sales expectations.
I have always thought they could do a hell lot better than the midrange GPUs they previously put in their machines.

Good for them.

 

asus has nice looks that aside its hella hard to upgrade in asus =(

 

Yeah but still fixed hardware which is sad.

 

They got rid of the 15" version too.

 

Fixed hardware but nice one, it's not like 4700hq will be outdated in 1 year.. the GPU is proprietary but removable right ?

 

Its removable but good luck finding a MXM card shaped like it...

 

BTW, I think we will have to wait until CeBit 10-14th March before we get to see GTX 860M and GT 840M.
Since they didn`t show anything at CES

 

I dont understand why no one opened the system information tab in nvidia control panel ! in it we could have seen how much CC there is the that gtx 860m !

 

I don`t think Nvidia have added support for Maxwell yet. Neither have GPU-Z so who knows what they would have showed. But yeah, they should have atleast tried running them at CES. But then again, they could have put in an old GPU, renamed it to GTX 860M so it shows that on Device Manager, to be able to show something on the show.

We will absolutely see leaks before CeBit though. I just hope we don`t have to wait months :/

 

Can anyone who knows chinese translate this? I`m just curious to what it says.

 

It's a joke about some hospital named Nvidia. Has nothing to do with any GPU news.

 

Ok, thanks

 

Cloudfire, Meeker since the 800 series is going to be rebrands with minimal performance increases, I would like to buy an msi gt 60 2oc that was at 1700 and now 1550.. do I take advantage of the deal and get the 770 on discount or wait for 870 and hope its priced in the 1500-1600 range?

 

Tough spot man. Nobody knows what GTX 870 will be. The only cards in the 800M series we know for sure what is, is GT 820M (Fermi, really old) and GTX 880M (Kepler, GTX 780M rebrand).

The rest could be anything. Me personally think GT 840M, GT 850M and GTX 860M is Maxwell since the notebooks with those isn`t coming out until March earliest. When we had Fermi and Kepler came, we also saw Kepler launched in March. If you could wait until March, do it.

Again, I dont know. Yesterday someone broke the news that GM117 (Maxwell) is coming in Q1. March is in Q1, so yeah, I think that will be the month.

I think buying any notebook with Kepler now is stupid now that Maxwell is so close. We saw a 80-90% jump from Fermi to Kepler. You will be very depressed if you buy a notebook with Kepler now and somewhere in March-June, Maxwell comes along and beats the living crap out of it. That is not money well spent in my opinion. If you were gonna buy a rebrand, the time was when 700M series launched. Then you could atleast use those GPUs a year until Maxwell came along. Now its only a matter of months.
It also looks like Maxwell will have ARM (Denver CPU) on the GPUs, which will offload tasks from the CPU to do PhysX and other CPU oriented tasks, so that will also be a pretty big bonus. But nobody knows which cards they are going on (GT or GTX? Only for desktops?).

 

The high end chips are probably still ~6 months out, maybe even more if there are any additional delays. And the new notebooks certainly aren't going to be priced cheaper than the current ones.

 

I dunno I bought my Fermi 675M in June 2012 and it's still rocking games at 1080p.. Nothing wrong with being behind 1 gen.

 

Go for the 770m machine now.

 

Whyyyy?

Its like a hot sun inside your notebook :/
GTX 680M, cooler and so much more powerful.

Yeah I know, money right? :/

 

Used 680m get my vote ~~

 

So, will Maxwell be the 900m series? I'm confused.

 

We don't know yet. *SOME* 800m series may be Maxwell, but perhaps not. 880m is Kepler (current tech), but there may be an 880MX or 890M or even 980M later this year with maxwell. Who knows?

 

I'm just hoping that the GTX860M is a Maxwell chip, available by August, and in machines that are ~$1,200 price point.

 

Found this. Its from GTC 2013.

Live: Jen-Hsun Huang at GTC 2013

 

hmmmm...My guess is first maxwell parts will be low-mid range chips. They will allow nVIDIA to test the next gen chips before they launch high end stuff.

 

痔疮 means Hemorrhoid and 肛肠 means Anorectal.
It's asking you if you have Hemorrhoid come to Nvidia Anorectal hospital.

I guess they think the Nvidia logo looks like the anus.

 

I think it will be Maxwell, but if it isn`t, you will absolutely have other Maxwell GPUs ready to buy in August. 99.999998% certain.

LOL thanks for the translation. haha

 

that 99.9999% as in 20nm? im hearing maxwell this and that so much but theres maxwell 28 and 20nm, so lets just drop maxwell and use 28/20nm instead. thanks

 

All Maxwell GPUs will be 20nm now that we didnt see any launch at CES. TSMC, which is producing the 20nm chips for nvidia, have already stepped up to mass production as we speak, or should do so any day now. We got to buy 28nm Kepler for the first time in late March. This will be the case with 20nm Maxwell as well according to my own forecast.
It will give TSMC and Nvidia 2-3 months to accumulate enough chips for their first 20nm GPUs. Which will only be for a few select Maxwell GPUs, GM117/GM107 whatever it will be called. And then as 20nm production get better, yields get better, and Nvidia get enough chips from TSMC, GM2014/GM104, is launched too.

TSMC poised to be leading 20nm chip supplier: analysts - Taipei Times


Xilinx which also produce their chips at TSMC have already begun to ship out 20nm chips to consumers

https://www.semiwiki.com/forum/content/2927-xilinx-begins-shipping-tsmc-20nm-fpgas.html


The signs are all there that we will have Maxwell 20nm in late March. Whoever told you 20nm wasnt coming until late this year is full of it. Maxwell was suppose to come out in 2013 btw. They are more than ready for launch now.

Also as a sidenote: Maxwell will have an ARM processor (Denver) built on the same PCB as the GPU die. That would be pretty difficult without the reduction of die size you get on 20nm as well as the reduction of power consumption.

 

Brave move, Cloud.

Although, valuable sources say the concrete opposite.

CES is no indicator of Maxwell launch. It might have been introduced to selected press journalists. U never know.

Regarding TSMC, yes, u would be absolutely right with your forecast if Apple did not join in for 20nm.
Game is changed a bit, as of right now. Plus, working Maxwell samples(desktop) were fully functional on 28 nm, long before 20 nm process begun to take a flow.

Making the desktop on 28 nm draws the conclusion of mobile SKUs as well.

And what about 880m, 780`s twin brother?!

Final Edit: I truly eager for 20nm 800 series (both D&M).

 

Apple will be no threat to Nvidia`s production. From the same link I posted above

6.5% of 100%. Thats not very much is it? I posted something earlier where I tried calculating that if Apple got around 20-30% of all 450nm wafers, it would cover their need for 20nm chips for iPad and iPhones, while leaving a lot for the other players. Which have been in business with TSMC for quite some time and are trusted partners with TSMC. Apple Inc., Qualcomm Inc., Xilinx, Altera, Supermicro, NVIDIA, MediaTek and Broadcom Corp are all producing 20nm with TSMC. I don`t think Apple can come in and expect a royal treatment over the other established players.
Xilinx and Altera are already shipping 20nm. It would be safe to assume that the rest (Nvidia) are atleast producing 20nm too since they all produce 20nm at the same factory. Right?

Where have you seen 28nm Maxwell samples btw? You can bet that Nvidia recieved their engineering samples in 20nm before TSMC stepped up to mass production.

GTX 880M makes very little difference to 20nm plans. When Kepler launched in March 2012, we also had GTX 670M and GTX 675M launching at the same time. They were Fermi rebrands in 40nm, while the Kepler cards that launched at that time (GT 640M/650M/GTX660M) were in 20nm. Its was a way to offer some high end GPUs that they can`t offer in new node due to low production at TSMC. Luckily we here at NBR are pretty good at detecting what is rebrands and what is not.

I could be wrong but I cant help but to see a ton of similarities between Kepler launch and this one

 

Hopefully, it all goes well and we receive the desired product. I would like to take a part in testing this Denver cpu and 20nm process as 28nm shades away.

What u say has a strong hold of its own but almost all the info leakers set on 28nm launch. Of course they could be wrong.

Supposedly, if 20nm will indeed see its light in late March, why is it heavily been ignored and all attention focused on 28nm rebranding?!

 

Kepler:
Mass production of 28nm @ TSMC began early November 2011.

First Kepler from Nvidia launched March 2012. They were: GTX 680 (GK104), GTX 660M (GK107), GTX 650M/640M (GK107). One month later, GT 640 launched (GK107) as well as GTX 645 (GK106).
GeForce 600 Series - Wikipedia, the free encyclopedia

Maxwell:
Mass production of 20nm @ TSMC began in January 2014. First Maxwell from Nvidia launched in late March 2014 (my forecast).

Since Maxwell mass production began 2 months after Kepler (November vs January) , Nvidia can`t do a GTX 880 (desktop, GM104) at the same time since 20nm availability is smaller than 28nm was. So we will see only GM117/107 first, then a month or two after, GM104 launch with GTX 880 (desktop) and perhaps GTX 880MX (mobile).

I think its the way humans act and the way internet works. If they see one rumor, the websites copy each other article and it is posted on so many sites it becomes the truth.
So among all those articles, there is one original leaker who may have posted an article based on nothing.

Truth is that we have never seen Fermi in 28nm or Kepler in 40nm. Nvidia do both new architecture and a new node at the same time.
Either we get Maxwell in 20nm in late March or we don`t get Maxwell at all until Nvidia is ready. Thats my opinion of it.

 

When 680 Desktop was first launched, Nvidia had a big shortage problem. Nevertheless, it did not prevent 680, 670.. being released first, followed by 660 and lower end.

Looks like gpu releases differ on Mobile/Desktop platform. I can be wrong tho. Mobile have lower end first, followed by high end. Desktops have the opposite, at least for the last 2 years.

As for the sources, I am not talking about sources like Fud, VC etc... As they are known for being 50% correct when it comes to information.

 

Well yeah, GM104 might launch at the same time as low end too. I`m talking worst case scenario where they do a soft launch with GM107/117.

I`d like to see the sources that say 28nm Maxwell btw. Ive seen many sites listing that rumor. I think they all originate from 1 or two articles

 

Regarding the source. I cant reveal the name of it as it is a pic PMed to me. Not too long ago, I was lurking on various forums looking for Maxwell info, until I stumbled on a little thread. After a short while, the messages begun to pop up faster, I had to translate asking why is it happening. Afterwards I was PMed a picture of an unknown gpu which looks nothing like what is on the market now(completely new gpu). A handful of people over at that thread, including the poster were talking of 28nm process and great power efficiency. It was all deleted within hours(including Pms to me, I dunno why but it looked like the owner of the site/thread had a free access to PMs, something that would explain the deletion of the Pm).

Anyhow. This might all sound like a fairy tail but still. Maybe I was in the right place at the right time

What is interesting is that later on, sources started to publish Maxwell to be based on 28nm chip with various evidences. Someone saying it is cause of fabrication shortage, others are: 20nm is not matured, etc.

Everything can happen though. If the 20nm chip quantity matches the expectations, Nvidia will release them by revamping the current so called "28nm Maxwell" into 20nm Maxwell. It is not carved in stone.

Also, birds have spoken that AMD`s 200x Desktop series are no more but an ace up their sleeve to counter 700 series. Which could mean, all this time both of them have been working on 20nm.

We have the following so far:

Maxwell info has been strongly tightened up because of

1) 790 and 780 Ti 6GB releases-those cards will surely come out.

Edit: 2) Maxwell is 28nm, (which is not what a lot have expected- incorrect) too early for leaks.

3) Maxwell is 20nm, but it is too early for leaks.

If something, 20nm will have the softest launch to date. Where only the lucky ones will be quick enough to grab one or two cards. While others gotta wait ~6 weeks till resellers have fulfilled their stock. Similar to what happened with 680 but with unavailability being the issue this time around.

 

Shrinking a core down a process node is no simple task, to do that as a stop gap would be very expensive, we wont see kepler shrunk unless maxwell does not scale down very well, which considering they want to keep their phone line in sync with desktops I doubt we will see that.

 

Nvidia historically launches their big cores first when they have a new uarch (though the smaller GK107 was also part of the initial launch with Kepler). And the lead times with with new nodes has been anywhere from 5 months 28nm to 7 months with 40nm. So if the initial Maxwell are indeed 20nm, I'd expect them sometime in May/June or later with a big core (GM104?) and maybe a smaller one (GM107?).

Or maybe Nvidia will do what they did with Tesla and launch on an old node first and shrink it down later.

Also, Tesla was shrunk twice (65nm->55nm->40nm) iirc and Fermi was shrunk once (40nm->28nm). I'm not sure that Nvidia will stretch Fermi another gen, so we might end up seeing 20nm Kepler as the 820m or 920m, like how the 620m/720m (GF117) is 28nm Fermi.

 

OMG I hope all the recent speculation will end up true, that would mean less waiting for me, I really want my upgrade, been using this laptop for 4 years now ugggh, microphone jack just died on me to, luckily I have got a Xonar U7, my guess is it wont be to long before the rest craps out .
Here is hoping by april I can get my hands on a spanking new 20nm Maxwell laptop . ^^

 

Tesla is a brand like geforce not a chip. GK110 is the same core as that in the titan super computer, it just ate them all up.