Clevo notebooks with 800M series coming out February 2014

8GB of vRAM is most likely for marketing hype, and the +105 core clock is probably to make those serious enough to consider the 880M feel good about getting a real "upgrade".

 

True, but it may just repeat a history with desktop 780 to 780 GHZ edition with newer and more efficient chip. Anyways, no point in upgrading at all just good for those, who didn't pull the trigger on 780M laptops)

Only downside in all this could be the release of desktop Maxwell cards in a month or so after 880M comes out...

 

Maxwell isn't coming out until the end of next year, I thought that much was clear.

 

Obviously to milk the cow some more while they work on Maxwell.

June-July at latest I`d say. They have already post poned Maxwell from 2013 to 2014.

 

octiceps do you mean 28nm or 20nm maxwell GPU's (not mobile)? The latest source i found about this was
Maxwell Geforce cards shipping by March 2014

While there's no time to implement 20nm till Q1 14 i thought the first desktop Maxwell based cards are out using 28nm tech process with mobile GPU's coming around middle-end of summer. Could you share another info? Where did the end of 2014 came from?

 

First of all, that link is from Fudzilla. Enough said.

OK maybe end of 2014 was a bit of a stretch on my part, but all roadmaps indicate 20nm GPU's in second half of 2014. If it ain't 20nm, it ain't Maxwell. I don't know where all this BS about 28nm Maxwell comes from. We've had die shrinks of existing architectures in the past but there's no such thing as a brand new architecture on a previous process node. That's not how tick-tock works. Did we ever have 40nm Kepler? Exactly.

We already know that 800M is gonna be a straight-up rebadge of existing 28nm Kepler parts. If the picture still isn't clear to you at this point you need to have your eyes checked.

 

Alright so I did a +105 OC on my 780Ms in order to do a head-to-head comparison with the 880M pre-release sample, and the results are not very encouraging:

Cloud Gate
GTX 880M -- 23060 marks -- 76534 graphics score
GTX 780M +105 core -- 25080 marks -- 77104 graphics score
GTX 780M STOCK -- 23438 marks -- 64266 graphics score

Fire Strike
GTX 880M -- 9054 marks -- 11708 graphics score
GTX 780M +105 core -- 9130 marks -- 11658 graphics score
GTX 780M STOCK -- 8479 marks -- 10635 graphics score

Graphics scores are essentially the same within error, so that 8GB of vRAM is clearly doing absolutely nothing. Of course keep in mind if this is a "true" 880M it is most likely an engineering sample, and there's always the possibility this is just one big hoax.

Notes:
1. Both systems have fairly similar setups, key differences being the 880M rig has 32GB of ram (vs 16GB for mine), and is running 2 x 256 Intel SSD in RAID 0, vs me just having a single 512GB 840 Pro.
2. Even with +105 OC Futuremark still shows 780M core as being 876 MHz, whereas 880M is 954 MHz. Could just be a reporting quirk of 3DMark, especially since it's reporting the vRAM amount on the 780M wrong.

 

Seriously, could you stop writing like a little kid and become a little more humble and mature?

First of all:
Nvidia and AMD does not have a Tick-Tock plan.
That is Intel. And you seem to not know how that plan works out, explained below.

Second:
Maxwell may or may not be 20nm. Intel is putting out new architecture on previous process node. Hello Haswell? It is 22nm exactly like Ivy Bridge but a brand new architecture.
Why can`t Nvidia do the same while waiting for 20nm production to ramp up?

Third:
We didn`t see any mobile Kepler`s, until March 2012. We had rebrands just like now on CES 2012 as well. GT 640M, GT 650M and GTX 660M, which was our first Kepler cards, was announced in March. That article Red Line post which says Maxwell is releasing in March, actually makes sense considering what happend with Kepler. GTX 680 which was Nvidia`s first Kepler desktop GPU also launched in March 2012.
We didn`t see our Kepler flagship, GTX 680M until June 2012.

Could this happen again?

Forth:
Some people also thought the entire 600M series was rebadges after they saw GT630M and GTX 670M and GTX 675M as rebrands. But as sneaky as Nvidia is, they introduced the "MX" prefix, which gave birth to GTX 670MX and GTX 675MX which was pure Kepler. That could happen again. It may be a rebadge series like you say and they do 900M as Maxwell, but there are several possibilities here.

"MX" is also pretty alike "Maxwell" when you look at it so you never know

 

Haswell, Broadwell, Maxwell, what's with all the wells? Oh well, as long as all is well.

 

Looks like I struck a nerve with the Nvidia fanboi. Predictable.

Nvidia and AMD do tick-tock just like Intel, but they do it in reverse, tock-tick. New architecture and die shrink happen on the same tock, e.g. GeForce 400 and 600 with big performance gain. Followed by a tick, e.g. GeForce 500 and 700, a refinement of previous generation with smaller performance gain.

Meanwhile Intel does the die shrink of an existing architecture on the tick (smaller performance gain) and the new architecture on the current process on the tock (bigger performance gain). Penryn-Nehalem, Westmere-Sandy Bridge, Ivy Bridge-Haswell, tick-tock, tick-tock...

And Intel does not release new architectures on a previous process node. 22nm is their current process, you look at their entire lineup and everything is on it. Now if Haswell were 32nm, then you'd be correct.

 

We'll see about that. I kinda think that it's way too early for rebadging a 780m, would not make any sense from my point of view. So I still hope for a 10SMX GK110 based chip, B2 stepping... now that could be really awesome if the notebook mobos can keep up

 

I said you should stop writing like a kid. Grow up

No. They do not follow tick-tock. They usually do new architecture on a new node. I said "usually" because it isn`t set in stone. They could do Maxwell on 28nm because they do not have a protocol like Intel.
We could argue what is new architecture here and not, since they do share a lot of similarities but thankfully there are articles that cover it.

Wrong again.
Broadwell is Haswell. Its the exact same architecture. But Broadwell is 14nm while Haswell is 22nm. So they do release new architectures on previous node. The next architecture is called Skylake. It will also be on 14nm like Broadwell.

Read about Tick-Tock and educate yourself.
Hint: Tick is new node. Tock is new architecture.

 

C'mon octiceps, you bring some of this stuff on yourself quite a lot of the time, you're sometimes rude & condescending without being provoked. It does make some entertaining reading at first, but it can become predictable.

 

Based on those performance results, it's hard to tell what the hell is going on. I was thinking it would be a 28nm Maxwell, but now I'm not too sure considering the resulting performance. Only thing that we should be able to see is if there's any power improvements that will be gained from a new architecture. Because 20nm got delayed significantly that there's no reason for nVidia NOT to produce silicon using their new architecture when they can get performance gains later with the die shrink.

 

My exact thoughts as well HT. I just wonder if its plausible or if they need to use smaller node to fit all those transistors they built on the particular Maxwell GPU. Say they managed to double the core count with 20nm. Would that even be possible with 28nm or does it require building a Maxwell GPU precisely for 28nm? Could they downclock the Maxwell cores which was meant for 20nm to make it thermally plausible for 28nm?

I think its very confusing as well.

Just found this:
Hardware.info

Looks like history will repeat itself. Maxwell like Kepler will launch in March like they wrote about here
NVIDIA Maxwell Based GeForce Graphic Cards Reportedly Shipping In March 2014


This bring more confusion in though:
Does they assume those are Maxwell out of the rumors floating around lately?
Have Remco Van Dorrn specified that it is Maxwell?


Man I give up.


Cloudfire out

 

1. Nvidia doesn't follow a tick-tock release cycle.
2. Nvidia initially released their Tesla uarch on a 'previous process node'.

So your reasoning doesn't make much sense. Nobody here knows what the 800m series is going to be at this point, but I highly doubt Nvidia would go 3 generations without introducing a new uarch.

 

Wouldn't they just need a larger die to fit all the extra transistors, just fewer chips per wafer? I dunno, I give up too. We should know soon though if February is true for the 800m series.

 

Any ES sample reviews out? for 880m?

 

nothing but the 3dm 13 scores up there.. (seems like an OC GTX 780M)

 

Sigh. 3rd Gen Kepler? lol.

 

Guys, I thought clevo's leak stated that gtx 800 series will be released in Feb not March . I am so frustrated and do not know whether to buy a gtx 770 laptop now or wait for 870? I have been waiting since 580's release...

 

Def wait and see what happens. Nobody knows yet. If its not Maxwell, we might get hints when they could arrive and how long you must wait for them

 

Anything in feb would be a rebrand anyway lol.

 

Larger dies are hotter, more expensive, lower yielding, lower clocking and harder to fit onto the package.

 

Since 580? You must've saved up enough to get yourself quad-Titan or sth by now

But then again what's a couple more months to.. 2.5 years you've already waited.

 

People doubt Nvidia would stretch Kepler another generation, when we saw 8800M GTX -> 9800M GTX -> GTX 280M?

Don't pretend they've never done it before. They are going to release something every cycle, whether they have new technology or not.

 

Well we don`t have to wonder much longer. Benchmarks and full reveal coming tomorrow.

Stay tuned

 

The GTX 280m uses a 55nm G92b core (die shrink of the 65nm G92 core in the 9800m GTX) and chips like the GTS 260m used a GT215 core (i.e. Tesla, which was the newest uarch at the time). So it wasn't a straight rehash.

The only time Nvidia went 3 generations without a new uarch was 8xxx->9xxx->1xx, and that was only because the 1xx was a short lived stop-gap gen. And those gens (i.e. 1xx, 3xx) have never had high end cards whereas the GTX 880m has already been leaked.

Nobody is saying that there won't be Kepler in the next gen. Heck, the 710m and 720m are still Fermi. But it is very unlikely that there will only be Kepler in the next gen.

 

Cloudfire, "Well we don`t have to wonder much longer. Benchmarks and full reveal coming tomorrow." I thought the big reveal was in CES in about a week, why tomorrow in particular? and by the way as far as notebookcheck is concerned gtx 680>870 could that be true ? if it is how can after 2 long years the second high end card (gtx 870) cant beat the 2012 high end card?

 

Because a cut down version of a chip can't beat a less cut down one no matter what you call it.

 

I know, but that doesn't mean they won't do it. But it's sounding more like another Kepler rebrand. Otherwise they'd HAVE to do that if they offered Maxwell in 28nm.

 

I hope the w230ss has nice sleek brushed metal aesthetics. better mouse buttons and a nice silenter fan noise to match the sleekness to go with it. (you don't a sleek machine and it be very noisy) the 860m will be a nice bonus boost too for gta5.

 

I have a feeling 860m == 765m.

And you can't have a small fan with those powerful components and have it be silent. The fan profile needs to be fixed at idle for sure, but they should be able to do that with an EC update on the current model. I'd rather see a larger CFM fan and dual heatpipe off the CPU to be honest.

 

I know it won't happen soon but anyone wants to throw out the 2.5" HDD slot and use the extra space for a MXM GPU plus needed cooling?

 

Absolutely. Use 2 or 3 mSATA slots instead. And I'm OK with embedded GPU, just use that space as a dedicated fan for the CPU. Damn, I wish I won the $600M lottery, I'd design and build my own laptop.

 

Hey cloudfire, where did you find out the reveal was happening today? Notebookcheck says the Date of Announcement is 01/02/2014, meaning the 1st of February. Did you read that, and think it was supposed to mean the 2nd of January? (I'm not sure where Notebookcheck gets their announcement dates from anyway).

Here's the link to the 880M at Notebookcheck saying the 1st of February:
http://www.notebookcheck.net/NVIDIA-GeForce-GTX-880M.107622.0.html

 

I always keep my word. Here is all you want to know about GTX 880M

Spoiler

All credits goes to: 慕容蛛











All credits goes to: 慕容蛛

 

God damn it nVidia! Another bloody rebadging. They really need to get their act together or they are SCREWED when AMD come out with Crystal Series.

 

You don't need $600M to build your own laptop.

You need way more than that for mass production.

 

just see the GTX 880M = OC GTX 780M ... they will add the rest as GTX 8__MX

 

No, but it would help. Point being is that you can't design and produce a PCB and case with the components you want on a hobbyist budget.

860m will likely have a slightly higher base clock than the 765m, sure, but that doesn't mean it isn't the same chip with similar OC abilities. Would be nice if they gave it 2500MHz vRAM though.

 

So know we know the 880m is a rebadge should i wait till it comes out and get a clevo with 780m for cheaper or get the 880m?

 

I should hope so. Otherwise, they will look like massive idiots.

 

I feel you man. Makes me sad to see them doing this. Milking the cow until 20nm is ready.

Sadly I`m here with more dissappointing news. :/

All aboard the rebrand train


7970M: 850MHz/1200MHz
8970M: 900MHz/1200MHz
R9 M290X: 950MHz/1250MHz

Source:
http://forum.notebookreview.com/gaming-software-graphics-cards/725226-incoming-amd-9970m-28.html#post9510494
http://forum.notebookreview.com/gaming-software-graphics-cards/725226-incoming-amd-9970m-29.html#post9510990

 

Well given that my 780M's don't really run stable at 954MHz, I'd say having those clocks at stock isn't half bad Wish I had a nice pair of 780M's so I could spit on the rebrand too

 

Jesus F**king Christ. So, both the big players trolling us with rebadging for H1 2014. Screw them. So, at the end of the day, nVidia will probably have the performance crown. Our only hope, it seems, is for MX versions of the nVidia cards.

 

I'm confused. Who cares if they are labeled MX or M? It's the same damn thing. 880m is already a rebadge, why shouldn't the others be 8xxM as well?

 

An MX version may actually be new architecture.

 

In the same way 680MX was a new architecture over 680M? Oh wait...

 

Yep, both AMD and Nvidia will rebrand the top cards. 7970M rebranded the third time, 780M the second time.
A comedy indeed :/

I think GTX 880M is based on B1-stepping just like GTX 780 GHz card. Both running at 1GHz because B1 stepping is more stable.

It was a new card different than 680M (1536 cores). GTX 670MX and 675MX was Kepler while GTX 670M and GTX 675M was Fermi.

This could be a way to introduce Maxwell without going to the 900M series