Clevo notebooks with 800M series coming out February 2014

Because we want to BELIEVE!



Cookie goes to the first person that tells me what that's from...

 

8gb on a 256 bit bus rofl. What a waste of perfectly good GDDR5 chips

 

GDDR5 ain't cheap, so it's just gonna hike up the Nvidia tax even more.

 

Indeed, especially with the PS4 in full scale production. It will only make GDDR5 even more expensive

 

Definitely no point in upgrading to the 880M unless you're buying a new laptop. (that's assuming the 880M is indeed just +183 core 780M with a useless 8GB vRAM)

 

Prema, are there any news regarding the voltage settings on 880M under load? Also there're different Elpida and Samsung memory chips on different pictures, which one is true?) Seems like elpida memory could not be clocked higher than samsung according to desktop cards

 

We all knew a rebrand was possible. That have been said many times in this thread. K5100M have already been pointed out many times, with the same N15E prefix.
We had hope of a Maxwell because a GTX 780M @ 850MHz already run 85C+ on most brands. Now they upped the clock to 1000MHz.

GTX 880M is A2 btw. Which means there won`t be a B1 stepping. A2 is just the codename for the factory they are using I think?
Will be very interesting too see the heat from these GPUs...

 

The left one is GTX 880M while the right one is K5100M.
N15E-Q5 is K5100M.
N15E-GX is GTX 880M.

Here is the die shot of GTX 880M



I think atleast Clevo versions will be ELPIDA and not Samsung. I bet they use Samsung on K5100M because the GPU is bloody expensive and they can afford to use expensive Samsung VRAM. While ELPIDA is cheaper, which is why they could include 8GB GDDR5 on the GTX 880M.

 

Sorry Cloud, I thought the shots had not been posted as the speculation talk went on...

The 880M sample cards, which he is selling since a month now (LOL), may not represent Clevos final vRAM choice...

 

I hope they use Samsung. I'd be a bit disappointed if they went with the cheaper RAM just to reach the 8GB...

I mean are people who spend massive amounts of money on these laptops that stupid to think more vRAM is better, faster? Is this just a Clevo thing or are other OEM cards also coming with 8GB?

 

When can we see the true successor to 7970m and 680m?

 

What do you mean by "true successor". If by new architecture, then there's been plenty of banter about both here in this thread and forum.

nVidia Maxwell 20nm likely will be H2 2014. Radeon (Volcanic Islands?) who knows.

 

@HTWingNut

Yes I meant new architecture wise sry.
The above post was a sarcastic/desperate outcry against all the rebranding

 

I can`t watch the Nvidia stream tonight. Its 5 AM here when it starts :/

 

Its 9:30ish PM where I live, so I can watch it. I still (uselessly) hope they don't up quite as badly as we have seen from the leaks. I would like to see the justification of 8GB of GDDR5. It will either be hilarious or pissing off.

 

Progress! I like how a decent 2012 or newer gaming notebook will outpace the next gen consoles.

Hopefully there is a branch of SteamOS for notebooks someday.

The new Tegra K1 seems neat. Maxwell should be coming up soon.

 

SteamOS will work on any "standard" Windows PC hardware.

 

Yeah, IIRC, it's Linux-based and made for standard PC hardware; you could dual boot it if you wanted.

Sent from my SAMSUNG-SGH-I747 using Tapatalk

 

An hour spent on Tegra tech... Still nothing about Maxwell

Sent from my Nexus 4 using Tapatalk 4

 

Sadly, nothing about Maxwell( gotta hold till CES then!

Sent from my Nexus 4 using Tapatalk 4

 

Although Tegra K1 looks pretty beastly for mobile devices, im dissappointed they didnt show Maxwell like Semiaccurate promised.

We still have January 7-10th to hopefully see some. If not, the last trust for Semiaccurate is gone.

 

They may show something at CES, but not in a massive press release like Tegra K1. This shows how much they are focused on markettng gimmicks and integrated devices. By going through this thread and the lack of an announcement, I am more than certain nVidia will just refresh the 7-series and do marketing stuns, such as 8GBs vRAM.

 

The same site that predicted the GF100 catastrophe despite all the naysayers. People with green colored undies usually don't talk about that site anymore, I'm surprised you had faith in them at all cloudifre

 

Refresh it with a die revision may be ?
Like an 765 OC for the 860 the 8 series will have a bigger oc potential ?

 

Just heard a joke that with the new 880M, you might use the VRAM for swapping when running out of RAM.

In theory, yes. The fab should be more mature through time so overall quality "might" increase a bit.

 

I really hope they have some secret prototype running a GM107 somewhere in CES. Like hidden in a dark corner, without any specs, that is just waiting to be found be a reporter lol.
God I hope they won`t do a complete rebrand the next 6 months. :/

I should probably be offended by this post but I laughed out loud instead. Green undies is def on my must-buy list
Tried giving you rep but have given out too much the last 24 hours

AHA, so that is why we now have 8GB on our graphic cards. They serve as a system upfresher as well

 

Incredible!

 

I think at the end of the day the 7-series with a good custom BIOS will perform almost as good as the 8-series. The 8GBs vRAM is a marketing stun, as consoles got 8GBs, nVidia is trying to impress the idiots that run nowadays tech press and your average, brainless consumer. No need for anything more than 4Gbs, DirectX 11.2 and OpenGl 4.1 supposed to eliminate the need for high vRAM amounts, with texture streaming.

 

HAHAHA.

I know the details of some upcoming chips. There will be atleast one Maxwell in the 800M bunch we have heard about so far.

1. GT 820M
GF119. Forget about it. GF means Geforce Fermi. Garbage!

2. GT 840M
GM108.
576 CUDA Cores running @ 1029-1124MHz.
64bit, DDR3 running @ 900MHz

Source say he tested GT 840M ES with those specs :thumbsup:
http://www.chiphell.com/thread-942040-1-1.html




Close enough to the HP Envy 15 specs posted a while ago

 

You can already have 12GB RAM with Quadro K6000 (not mobile). Given how Hawaii's memory controller is designed, 16GB is possible in the next FirePro line up. Both can still run into VRAM size problems, especially in compute.

In gaming, 8GB might be too much given the throughput of the GPU core, but only if we continue to do things the old way. There are already a number of game engines that use memory heavy techniques like vexol model and multi-layer procedural generation, and they are all fighting with VRAM size limitations. If available, any engine of this kind will eat 16GB for breakfast, even if core throughput and VRAM bandwidth do not increase.

Having 8GB or more VRAM may sound ridiculous at first, but it can be useful if the software we use catch up.

 

Not in a mobile consumer card, 256-bit or less.

 

Maybe if it pushes 4k with AA and lots of big textures?

 

4k on a laptop thats a joke, even with a 20" laptop I think its far too much for average joes. probably cant tell difference anything beyond 2kp

nvidia won't release the press that easily, I'd only imagine their stock goes way down if somethings not done properly lol. SA did mention not seeing maxwell till end of this year or possibly early next year but that was a speculation from a long time ago and things might have changed. maxwell for desktop only maybe.

i can feel it, rebranded 780m with more vram LETS GO NVIDIA

 

Because everybody is fine with good old static polygon models forever...

 

Do "everybody" have a choice?

 

The hardware need to exist first, then devs can code for it. Before this happens, nobody has a choice.

 

Anything besides polygons would require a shift in current hardware design because they're very good with polygons and shading them. Scalability with something like Ray tracing would be nice but require all new api and hardware and ways to make games.

Beamed from my G2 Tricorder

 

Current gen cards are doing fine with polygon-voxel hybrid engines using rasterization <=> cube-marching if there's enough VRAM to hold the data. I agree that the API needs some work. Once ploygon generation is complete the rest of the pipeline stays (largely) the same.

Ray tracing (against polygon or voxel) has to wait. We simply don't have the throughput needed to doing good-looking (recursive) ray tracing.

 

The W110ER needs to see a refresh. I have such an example Rebranded by Aftershock as the X11 I won from a Raffle. There has been no other laptop save the overpriced Alienware M11 which even so was already long in the tomb with the even older GT 540M so it's about time that this gets a model to fulfill the role of the smallest gaming laptop in the world. It currently doubles as my LAN party machine.

 

You jammy dodger!

 

Throw out the 2.5 HDD on W230ST and squeeze the rest of it into a 11".

 

The w230st is the replacement for the w110er, About same depth, w230st is thinner but a bit wider. If you put one on top of another, or side by side they are remarkably similar.

Beamed from my G2 Tricorder

 

What about the 870m will the re-brand be as bad? or maybe there is more potential for improvement? and Is the information given in this post 100% accurate or do we wait for CES to give us the 100%? and what about an improvement over GPU boost, is that to be expected?

 

I was expecting new ROG laptops ! nothing, no refresh or s.th.. no announcement !

 

imo asus sucks now cause they like to do their own style of graphics card, very difficult to upgrade. the disassembling procedure is also a pain too. after owning g74sx and g75vw, dont wanna continue with ASUS anymore. go for clevo or alienware man, its the way to go!

 

but clevo sound is just medium (high end laptops) or bad (mid range laptops)
Alienware are so, so, sooooooooooo expensive too !
MSi are single fan
(driving crasy)

 

Just use headphones. Problem solved.

 

EDIT: Nevermind this post :/

I`m gonna be brave here. This is a shot in the dark.

GT 840M
Maxwell, GM108
576 CUDA cores (2/3 SMX)
64 or 128bit
1-2GB DDR3

GTX 860M
GM108
1152 CUDA cores (4/6 SMX, full GM108)
128bit
2GB GDDR5

GTX 870M
GK104
1344 CUDA cores running at around 1GHz
256bit

GTX 880M
GK104
1536 CUDA cores running at around 1GHz
256bit


Looking at this, GTX 860M will perform lower than GTX 870M and will not interfere with the numbering (GTX 860M > GTX 870M > GTX 880M)
One might think 128bit is too little for over thousand CUDA cores, but they are slightly weaker in FLOPS/core than Kepler and the memory bus have improved. Fermi Geforce cards had 384bit bus. Kepler had 256bit. Maxwell....who knows.
We haven`t heard any news about any 800M cards, but SemiAccurate promised us some Maxwell news during CES. The OEMs that is showcasing the notebooks are awefully quiet about the cards. Nobody have dared run some GPU-Z for us (strange) and since the notebooks that are there that features 800M series only feature the two cards, GT 840M (Lenovo Z40) and GTX 860M (Lenovo Y50, 4K display), it makes it a little more fishy.


Now please don`t shoot me if this is very wrong. They are all based on rumors and some logic. This post is mostly for myself to what I think will happen

 

That would be pretty cool. But going by the specs, GM108 seems a little big. I would expect a 1152 core chip to be called GM106 or something.

 

Still gk106 for the 780m most likely.