Clevo notebooks with 800M series coming out February 2014

Sorry, I don't know much in this area

 

1) Q3 is late. Cause Q2 is the middle. Plus Q3/Q4 are the possiblilties.

2) What is funny here? I know mobile world more or less and considering 680MX to 780M to 880M , actually, it is been more than a year of rebrands.

 

Nice article.

Thanks, hopefully someone else may care to explain?

 

Anyway. As the example has proven, lets just end this silly arguments and get back to the topic.

This way, everyone will be happy

 

There was an official chart with the price per 20nm wafer and other components compared to 28nm. If I had it with me now, it would have maybe answered some questions for u.

I think we should have some info of this 20nm being expensive as we approach the 800 gpu releases. Let them be desktop gpus, but the reviewers wont miss a chance to spoil the 20nm process info in more details.

That is my take on it. Good luck trying to find the answer tho. Someone will defiantly have smth at hands.

 

Nvidia was complaining earlier about wafer cost at 20nm vs 28nm. It is quite a complex picture because yields and such steps in, which nobody knows if will be good or bad on 20nm. Not until they started producing it in mass scale. Nvidia had horrible yields with 28nm production for their Kepler chips at TSMC, but AMD had actually pretty good results. It comes down to that, plus the process cost of producing the wafer itself.

Anyhow, Nvidia threaten to leave TSMC in 2012 because of the bad yields, but 5-6 months after that they did nothing but brag about how awesome the 28nm process have become at TSMC. Maybe they have found out a technic that works with 28nm that also can be applied to 20nm. Who knows.

Either way, wether they like it or not, they are forced to go down a node to be able to produce chips with enough transistor density for it to be able to keep up with Moore`s law. Not to mention to keep the interests of their customers at bay. A new architecture on the same node won`t nearly be enough as stepping down a node and showing in more transistors as well as changing architecture.
This is why they come up with ideas like increasing wafer size to 450mm which let them produce far more chips/wafer than current 300mm. Which will reduce production costs.

http://www.extremetech.com/computing/123529-nvidia-deeply-unhappy-with-tsmc-claims-22nm-essentially-worthless

 

Do u know why Nvidia decided to go 16nm first after 20nm?

As the cost goes up should they just try to lower a plank as low as possible for save ups?

 

You mean 20nm > 16nm > 10nm instead of 20nm > 14nm?

I have no idea since Apple will be producing 14nm A9 CPUs at TSMC. But they are filthy rich and could probably afford it. Cost like you say might be one of the reasons yes. Especially when you consider the charts above

 

Interesting so the wafer price and yields are becoming a major problem. Thanks for the info. Wonder what problems Intel is having with 14nm

 

Yea, pretty much it.

So, they are orientating on the "leading process" as well, it looks like?!

 

They have their own factories and probably way better equipment than TSMC and GlobalFoundries. They are ahead of the rest for a reason. Intel will be out with 14nm Broadwell this year while other get 20nm.
But yeah, I guess they too have their yields problems

 

While on the right forums:


What do u guys use to keep your laptop not degrade in performance? As well as other useful laptop tips a newcomer like me might not be aware of.

Thanks

 

For Nvidia, the inital price per wafer (and thus cost per die) will be higher due to limited volume and yields. But with a smaller node, they get more dies per wafer due to increased transistor density so eventually the cost per die becomes cheaper as volume and yields go up.

For consumers, the price for new tech will pretty much always be more expensive than current tech, but I wouldn't expect new 20nm GPUs to be priced much differently than when the 28nm GPUs when they initially launched.

edit: oh, cloudfire answered this in the last page... oops, didn't see

 

Nothing? A laptop's components just don't degrade. It either works or it doesn't.

The only thing that affects performance is heat. Keep your fans and heatsink clean and it will work until something breaks.

 

Yep. Just keep your vents clear, occasional sprays (like once a month is all) with compressed air will do the trick.

 

The reason I belive Cloud is right is because of the simple principle of Occam's Razor. Sin has too many assumptions to back his theory, while Cloud has articles and/or links to back up his claims.

Now, bickering aside, lets hope that nVidia will get their act together and take hold of their best chance to leave AMD in the dust as far as I've known and get 20nm GPUs out and give us something worth upgrading to.

 

Eh whatever. Doesn't matter since PS4 and XBone have both proven to be failures in terms of their expected performance. Current laptops have nothing to fear from upcoming games since AMD APU in consoles suck so much. Nvidia could release junk and it will still be fine for gaming since games will be made for console first as always and well... they suck with AMD APU.

I'll be joining the smart ones and build a Broadwell + AMD R280 system in a few months rather than build another expensive laptop that will be obselete in a year and require an entire system rebuild rather than a desktop, just upgrade GPU/CPU and hooray.

It's not really the price that got to me, but this switchable graphics. Switchable graphics ruined high end laptop gaming for me. It's made having a high end laptop pointless for me.

 

I think you are just being a jerk. The R280 and the 880M have similar performance (Check notebookcheck's hierarchy chart), and since this thread is about Clevos, laptop owners CAN upgrade GPUs (MXMs) and CPUs.

Obsolete in a year? Are you retarded? This - NVIDIA GeForce GTX 680M - NotebookCheck.net Tech - is from 1 1/2 of a year ago. Look at the benchmarks. It plays current games FINE. CoD Ghosts, BF4, Metro : LL, AC4, F1 2013, Batman : AO, Starcraft 2, Crysis 3, etc. (Some may require dialing down the settings a bit, but it can run them. Hell, even the 580M, which is ancient by your standards can run some of them.

And, my dear sir, laptops have another advantage. They are portable.

And without switchable graphics, they would have less battery life and not be all that portable. So, the 880M Clevos would end up as more expensive versions of your desktop. Remember, the whole point of a laptop is portability (less portable for some like the P570WM), but still portable.

Maybe cut the BS and stop being so cynical and wrong, instead of telling us your desktop config for later this year.

 

Right, I need to disassemble, the laptop in order to clean them. U think it is safe enough procedure(not to break anything)?

 

If it's a Clevo then there's literally just a few screws to undo to access the vents. That's all it takes.

 

I think he had his reason to go desktop, but I do disagree with him. we laptop owners love high end stuff wouldn't care much for upgrading each year, or every 2 years. desktop is gonna be in dust sooner or later with intel pumping out close to zero upgrades on CPU end, and force you to buy X edition only.

btw perfectstorm, notebookcheck is crap, more biased on nvidia and apple. they use and test system at stock bios which is lame imo. we go for high end stuff, and push it to get the best which makes that site a joke.

 

Thanks for the heads-up, man.

 

That desktop is pointless, you're looking at incremental upgrades/improvements at best. My 4900MQ + 780M SLI can handle anything I throw at it, even Crysis 3 can average 55+ FPS on Ultra if I turn off AA, and this is just on stock clocks with the default vBIOS. I can confidently say my laptop will last at least another 2-3 years before I have to even think about upgrading, and as long as I'm not chasing the latest titles (I'm not a heavy gamer anyway), this laptop will last me for the foreseeable future.

And when I finally have the need to upgrade (and assuming this laptop lasts 6+ years), by then hex-core (or even octo-core) CPUs may have become mainstream, DDR4 will be well underway and we may even begin to see DDR5, Ramdisk may start to enter the mainstream, and the price of SSDs may actually match the price of today's mechanical drives. Now that's a desktop worth building.

As to switchable graphics, SLI capable laptops either have Optimus cut off from the hardware front (Clevo), or the system is muxed so you can choose to run only on the dGPU (Alienware). If you only want a single GPU, then you can look into the Asus G750 series that also have Optmus cut off from the hardware front.

 

for simple reference and comparison they got lots of info, anything regards to OC they got none. power limit, power draw, oc capabilities, and all those parameter tested with unlocked or modded bios/vbios. etc

@cloudfire any news on when 880m coming out? and if its 20nm ?

 

Thought you guys would be interested in these couple of posts here at NBR at this link:

http://forum.notebookreview.com/ali...r9-m290x-alienware-17-18-a-2.html#post9534021

880M is now listed on ebay to buy through UpgradeMonkey. Another forum member (Red Line) at that link above reported that they had sent a couple of 880M's to some people at Tech Inferno for testing. Apparently those 880M's throttle like crazy with their stock VBIOS and consequently are about as fast as a 780M - apparently it needs a modified VBIOS from the get-go in order for them to avoid this! (Haven't tried searching for the Tech Inferno thread that details this testing though).

EDIT: Here's a link to the initial testing of the 880M:
http://forum.techinferno.com/alienware/5734-upgrademonkey-gtx-880m-review-preview.html

It looks like the throttling that was mentioned happened during Furmark - no surprise or shame there! So, to update I think we should take the apparent throttling problem with a grain of salt. (I'll do some more searching to see if I can find anymore testing on it).

EDIT 2: There's also this thread from the suppliers of the card (Upgrade Monkey), that have outlined some of the specs, but nothing we didn't already know, but maybe a little bit more official confirmation:

http://forum.techinferno.com/alienw...review-upgrademonkey-techinferno-preview.html

 

Thanks for tips on how to clean laptop fans guys.

OT: Well well well, GTX870m is missing. That is strange. Maybe it is Maxwell as well?! Most importantly is that according to the article the launch is very soon, which also means Desktop cards are near the corner.

 

So the 880M that UpgradeMonkey has uses Elpida memory. Wonder how it'll hold up against the Samsung that 780M has.

 

It will likely clock 50 - 100 MHz less on average.

 

I guess that won't be a problem for people who don't bother with tweaking things, but let's face it, anyone who buys a 880M likely has or will start tweaking very soon.

I'm even starting to wonder whether the 880M could be a downgrade in certain ways. If temp and power consumption stays the same at stock clocks compared to 780M then it might be worth it, otherwise it's gonna be a hard sell.

 

So it really is looking likely that the 880M is the 780M with double memory and higher default clock speeds? A rebrand in all but VRAM?

 

I think that's pretty much 100% confirmed now, especially given the Upgrade Monkey sample, and all the other information & leaks we have seen. Although, it looks like some versions of the 880M will come with 4GB VRAM, saw that somewhere linked earlier in this thread I think.

 

GTX 880M is pushing it. Like really pushing it to milk the cow some more.
GTX 780M @ 850MHz is already running high 80s in Clevo`s and can sometime touch 90s with MSI machines. And now they have increased the clock to 1000MHz...
Current Alienware`s can perhaps take it but I don`t think that card will have any success due to the fact it will run hot as hell (which that throttle might be an indication of, idk) plus its a rebrand.

I hope to see some GTX 860M numbers soon. I have great hope that its one of the first Maxwell`s we get

Here is some GTX 750 Ti numbers I posted today (the card thats suppose to be Maxwell). But the result was kinda depressing. :/
But we don`t know the complete picture yet about power consumption, features and such. So fingers crossed for something better about that card...

http://videocardz.com/49409/nvidia-geforce-gtx-750-ti-15-slower-gtx-660

 

Yep, we don't know the efficiency/performance of the Maxwell architecture until we know some more details about that 750ti.

 

This sums up my feelings here.

 

better ask tsmc

am i the only one here that doesn't believe that new gpus will come out in Q1 and Q2?

 

The apparently got up to mass production of 20nm in Early Jan (The actual time is somewhere in this thread).

 

The current 880M if released as it is would be a disgrace to the 800 series. At best this is a 785M, and if this is still a 28nm Kepler, then calling it a 780M OC edition isn't exactly off the mark either.

 

if that is true I'm right, Q3 is a very possible time, specially that production is divided between 2 manufacturers, couple that with held issues, Q3 and Q4 makes sense

 

Q3 definitely makes more sense to see the full line, we might get one or two by Q2..

 

28nm Maxwell has its place. Don't displace it too quickly...

It is quiet a long way till Q3 Q4 and what will be your toy till that time if it is not 28nm Maxwell

And 860 cant go left out as it seems promising.

 

I still think GT 840M, GT 850M and GTX 860M will all be based on Maxwell like previously speculated.

99.8% certain

I sincerely hope they will be built on 20nm too. That Lenovo Y50 with GTX 860M release in May is good news regarding that.
However Lenovo Z40 with GT 840M will release in March.

Hmmm, maybe they are two different chips?

 

Will 860M be faster than 660M?

Why? Lenovo Z is low end and Y is high end. Why should they be different chips?

 

Who knows. I would certainly expect so.

GTX 660M was like 30% faster than GTX 560M when it came out I think. I`m curious what Unified memory could do, and that oboard Denver ARM (if that happens).
I wonder what the Denver ARM will do. Make the idle consumption as low as possible? Work as an IGP instead of using the Intel CPU, and therefor allow Optimus users to finally get 120Hz support? Create better PhysX? Better GPGPU performance? Help the CPU in various tasks?

Exciting

 

Unified Memory should be bread and butter for Maxwell chips if there are no ARMs there.

Funny how X60 -30%
and X80-12%

 

Also in my previous posts I said that I expect GTX 870M to be a rebrand.

Looks like its true as well since GTX 870M is on sale at ebay like GTX 880M.
Business as usual from Nvidia and history seems to repeats itself once again

NVIDIA GeForce GTX 870M 6GB DDR5 for MSI Clevo Alienware Upgrademonkey | eBay

http://www.ebay.com/itm/Nvidia-GeForce-GTX-880M-8GB-DDR5-for-MSI-Clevo-Alienware-upgrademonkey-/261374015476?pt=PCC_Video_TV_Cards&hash=item3cdb1af7f4

 

All the high ends are rebrands, as expected. 870M= 770M?

 

Hard to say. I think its either GTX 770M or GTX 675MX.

 

They should fall in line with 790 release. Nvidia is hiding Maxwell for real.

 

Well it seems they like following routine.

8800M > 9800M > 280M > 285M

480M > 485M > 580M

680M > 780M > 880M

So I guess we shouldn't be surprised. We already saw high-end Fermi rebrands in the 600M series which were later superseded by the true Kepler offerings. The 880M is a stop-gap GPU with a MX edition coming later in the year. I guess Summer/Autumn time for a 20nm mobile Maxwell - all depends on timing of the desktop cards. They will prioritise desktops, that is definite. So I expect to see the high-end laptop GPUs available a few months after the desktop launch of high-end Maxwell.

The real interesting thing is whether the MXM versions will be compatible with existing Clevos. Sure hope so!

I'm definitely skipping a 880M that is essentially an overclocked 780M. I would be a fool to buy into that. A rich fool.

 

The ARM chip is what I'm looking forward to as well. It would be awesome to have the IGP on the dedicated card, if we can call it an IGP then, and bypass Intel altogether. That way nVidia has control over the display output and we're not tied to Intel's crappy drivers and support. Not only that but the ARM should have access to the fast vRAM allowing for reasonable performance on battery.