*Official* nVidia GTX 10xx Series notebook discussion thread

Does the GTX 1050 surpasses the GTX 970M?

 

No, AFAIK the 1050Ti is the one that surpasses the 970M.

Sent from my SM-G935F using Tapatalk

 

The 1050 is only a little better than a 965m and not at 970m performance.

 

just received the new kaby lake gt83VR, first firestrike run below!

http://www.3dmark.com/3dm/17463632?

 

lol better than 99% all tested hardware.
Makes my 1060 scoring 9700 and 11800 graphics score feel like 5 gens behind


Sent from my iPhone using Tapatalk

 

Graphics Score looks great, expected the Combined Score to be higher considering the amount of GPU power but it's only at 7340, which is 1000 points behind my desktop with GTX 1070 + 6700K (in sig) - I suppose that test uses more CPU power than I realised & the 7920hq is holding back your system in that test. Loads of GPU power though! *EDIT: During the Combined Test in Firestrike my CPU is only at 45% load, so that test doesn't use much CPU power - not sure why your Combined Score was so low then - hmmm - (overall power throttling to the system maybe - not enough total watts available to system?).

 

highly doubt it, it has 2 - 330 watt power supplies feeding into the system. I am limited severely in that test because its the 1080p test and CPU is bottlenecked like hell with 2 gtx 1080s at that resolution. Ill try a 4k bench and see.

 

here is the 4k fire strike bench results:

http://www.3dmark.com/3dm/17467190?

 

Yeah, but my CPU was only 45% loaded during the Combined Test, yet my system scored 13% higher than yours in that test - your CPU doesn't have something like 1/3 the power of mine, so your Combined Test score should have been higher - the score is calculated on absolute framerate achieved, so your sli GTX 1080's don't really come into the equation. I mean I don't mind, but you might want to know & make sure your system is behaving like it should.
*EDIT: tried to find some Combined Test Firestrike scores on Notebookcheck.net for similar notebooks, but they don't list the Combined Test score - so unfortunately I can't directly check.

*EDIT 2: Ah, I did find one result for a GTX 1080 notebook with 6820HK CPU, and it scored 7884 in the Combined Test of Firestrike - which is more than you got with your better CPU - I think there's something up with your system. At this link:
http://www.notebookcheck.net/Aorus-X7-DT-v6-Notebook-Review.183986.0.html

*EDIT3: Ok, my last edit! That 6820HK in that review was overclocked to 4Ghz, hadn't noticed that. I reckon that Combined Test is not very multi-thread aware then, because I did notice that only about 4 cores were being effectively loaded during that benchmark (albeit at 45% overall CPU loading), and I'd assumed it would use the extra threads if it maxed out any of the other cores - looks like that test might not do so!). @HaloGod2012 , your system might be ok afterall. I suppose the best way to be certain is to monitor clocks of GPU & CPU during testing to ensure they're running at the boost speeds that you expect with no throttling.

 

Stumbled upon the shunt resistor mod seen here... has anyone tried/have information on whether this works on mobile Pascal graphics cards?

 

I think I'm good since the 7920hq at stock only boosts to like 3.8ghz when all four cores are at high load. I'll use xtu and overclock later. Scores seem to match perfectly with what I'm running. You scared me for a second

 

Supposedly 3.7GHz

Largest possibility is a throttling CPU, much worse RAM than yours, and the GPUs being starved.

Your desktop 1070 can probably around match a single stock 1080N with your OCs, and your 4.5GHz CPU with better RAM might compensate the difference.

But honestly, it would require more testing to figure out.

 

7920HQ still has extra multiplier bins? I'd be curious to know what you find

 

max multis in XTU are:

1 core: 47X
4 cores: 43X

Haven't tried overclocking the cpu yet

 

Not sure why

I think that might be the case.

I just ran a bench with only +100 GPU , but with the CPU at 5ghz and RAM 3200mhz. It doesn't do anything for the score in 3D mark but it shows how much difference in actually FPS (or isolated scores) there is with the combined test: 34.14 fps vs 41.32

Halo's GPU score is around about right, being about double.

http://www.3dmark.com/3dm/17464153?

 

testing cpu overclocks now. 4.2ghz on all 4 cores is looking promising so far. Running prime as we speak. 2 hours of prime and I'm happy stability wise. Will shoot for 4.3ghz on all 4 cores after. Then, of course, will come the GPUs.

 

tyvm, glad to know +600MHz is still allowed for these.

Can you test for me extreme load scenario? x264 long video render etc? Can it hold its TDP properly?

 

That 7820HK should do better at OC'ing than the 6820HK's did, so try 4.0ghz to start - that might even still support a slight undervolt of -10mV to -30mV, and then try to kick it up as high as 4.5ghz with up to +100mV overvolt - the cross over point might be around 4.2ghz like it was with my 5950HQ, or maybe even higher.

The 6820HK's were pretty good at 4.0ghz, with a few reaching 4.1/4.2/4.3ghz. So the 7820HK might get another 100mhz-200mhz higher than the 6820HK

 

The 7820HK max turbo is 3.9ghz, and the 6820HK max turbo is 3.6ghz, which is an oddly large difference, an increase of 300mhz!
http://ark.intel.com/compare/97464,88969,97185,88967

 

I was checking 4-core turbo, actually. 7820HK I think is 3.5GHz 4-core turbo, up from 3.2GHz on the 6820HK.

Still a 300MHz difference, though.

 

Yup, both are big increase from 6th generation to 7th generation!

Taking that 300mhz difference into account, I am hoping to see 4.5ghz 7820HK's reporting in

 

The 7820HK should OC further depending on the binning. The problem with 6820HKs was that they needed too much voltage for 4GHz+ on average, and for the most part 4.4GHz and up was impossible to sustain, and 4.2GHz+ was unrealistic in heat/power draw for most of the notebooks they sold in.

Going by this logic, if Kaby Lake keeps that significantly lower voltage below 5GHz as a going trend, we could theoretically see 4.8GHz+ (for validations at least).

Also however, it's a known fact that most Skylake notebooks were TDP limited by the OEMs (not the chips!)... ASUS could not surpass 75W, and Alienware 100W, and I know not the limits of Svet-modded MSI notebooks. Clevo models got current-limit throttle easily with Prema mods as well.

The problem with this is that Kaby Lake is both hotter (see above noted inability to cool 6820HKs at 4GHz+) and pulls more power (see noted inability to cope with power draw; mainly due to too-small power bricks or MSI designing notebooks to pull from the battery). Also, it sucks this more power while using less voltage, which means that it requires more current (amps). This means current limit throttling is far more likely to appear.

In total, I'd say this means most Kaby Lake notebooks (which are not changed from Skylake in design) will have lower actual overclocks than Skylake was able to achieve. The ones I expect to reach the tops are Svet-modded GT73VRs which have very good CPU cooling, but not many others.

But on the better point, at least the CPUs aren't slower than a salted snail in winter. 7700HQ boosts to 3.4GHz which, while still rather slow, actually will handle a 1060 half decently.

 

I think everyone forgot that I don't have a 7820, I have a 7920, lol. My max clock in XTU is 4.7 ghz/ 4.3ghz on all 4 cores

 

Wow, awesome

The 7920HQ OC is much greater than the 6920HQ allowed - max 4.0ghz on 4 cores!

Stock 7920HQ numbers are better than 6920HQ OC!!:
http://ark.intel.com/compare/88972,97462,97464

 

Yeah I'm impressed so far with the 7920. Testing stability of 4.3ghz today.

 

@hmscott @HaloGod2012 So what's the maximum theoretical OC for the 7920HQ? 4.7 / 4.3 / 4.3 / 4.3?

 

1 core: 4.7
2 core: 4.5
3 core: 4.4
4 core: 4.3

and so far, ive been sitting at 4.3ghz stable with a 80mv under-volt.

 

Ok thanks for the info.

Very strong CPU, quite an improvement over the 6920.

Have you ran anything intensive with single core to test that 4.7GHz stability?

 

not yet, have been busy gaming
I will test tonight and report back with some screenshots. The power of this is really overkill so I don't even see the need to overclock too much. I occasionally hook up to my 4K OLED tv so Ill definitely go for the max when running on that. Right now I'm sitting at 4.3ghz , -80mv, prime95 stable for 45 minutes. GPU is at +150 on the core and +300 memory; will bench with these numbers in 3dmark. Temps are a little high. After gaming for about 1 hour I saw max cpu temp hit 85C because fan profile is slow to kick in and realize( average was 72 - 75C while gaming), and GPU max temps were 83C and 85C in Rise of the Tomb raider using DX12 multiGPU at max settings but averages stayed around 79 to 80C. All in all great temps once the fan profile kicks in.

 

Probably best to thoroughly test stability before using your laptop for gaming or whatever, because borderline stable CPUs can corrupt the operating system, in fact makes sense to do an image before you OC the CPU and then you can always fall back to that if you get corruption. 4.3Ghz at -80mv sounds very impressive - what is the actual voltage (Vcore in HWInfo) of the CPU when at 4.3Ghz prime on all cores? (I'm curious to see how little voltage might be required for these new mobile chips)

 

For comparison regarding the combined test, my system at 4.1Ghz does 8194, with a single GTX1080. I do remember with my 980m SLI I had the graphics score of an overclocked 980ti, but the combined score of a GTX 970. SLI most likely doesn't do anything in combined?

 

Too much voltage is almost an understatement hahaha. Going from 4.1 to 4.2 needs almost 100mv extra for me to be stable in all scenarios. 4.4 needs over 200mv extra, and draws over 100W. Compared to around 70s W for 4.1 Ghz. Hell, I can even do an undervolt at 4ghz. It is crazy how much voltage this needs for close to no gain.

 

SLI scales about 50% in combined test.

 

Can you switch between integrated and dedicated graphics?

How is the battery life?

 

And dont expect...



WITH KABY LAKE!! BGA chips will always need more Voltage than LGA chips with increased clock speed.

 

While we talk about bga jokebooks let's also touch on the subject of "disabled" computers

 

It looks like things are heating up in the Quadro World, not only was a P6000 mobile implementation spotted a while back:

Spoiler: Quadro Pascal Mobile

NVIDIA’s Pascal-Based Quadro GPU Spotted On A Laptop – Is The Company Prepping Mobile Variants Of Its Workstation Graphic Chips?
http://wccftech.com/nvidias-pascal-...ng-mobile-variants-workstation-graphic-chips/

"At the SIGGRAPH 2016 expo, NVIDIA showcased a demo of the mobile version of its Pascal-based Quadro GPU. This could mean that the company is going to be announcing the mobile variant of the Quadro family, just like the details of the GeForce GTX 1000 series of chips for notebooks were stated earlier."

...why did Nvidia wait to deliver this?... maybe they were waiting for the GP100 and it's mobile counterpart - linked with NVlink instead of SLI for multi-gpu implementations!

Now Nvidia has announced the Quadro GP100 full GPU compute / rendering solution (released March 2017):

NVIDIA Announces Quadro GP100 - Big Pascal Comes to Workstations
by Ryan Smith on February 5, 2017 6:01 PM EST
http://www.anandtech.com/show/11102/nvidia-announces-quadro-gp100

"Kicking off on this Sunday afternoon is CAD & CAE software developer Dassault Systèmes’ annual trade show, the aptly named SOLIDWORKS World. One of the major yearly gatherings for workstation hardware and software vendors, it’s often used as a backdrop for announcing new products. And this year NVIDIA is doing just that with a literal Big Pascal product launch for workstations."...

With NVlink coupling - 2 way path - daisy chainable(?) - at least 2 GPU's linked, more?



Now, let's cross our fingers for mobile NVlink multi-GPU solution(s).

 

MSI WT73VR and New MSI Mobile Workstations Offer Something for Everyone
January 5, 2017 by Jerry Jackson
http://www.notebookreview.com/news/msi-wt73vr-new-msi-mobile-workstations/

" MSI WT73VR: Specialized VR Mobile Workstation
As more professional programs like Autodesk Stingray and an increasing number of CAD/CAM applications take advantage of the latest VR headset technology the MSI WT73VR makes for an ideal companion. The WT73VR is available with either 7th generation Intel Core i7 processors or Intel Xeon processors.

The VR-ready graphics are handled by the Nvidia Quadro P5000 GPU with massive cooling vents to keep the high-power CPU and GPU running at their best."

Professional Graphics with Pascal
https://www.nvidia.com/object/quadro.html
https://www.nvidia.com/object/quadro-graphics-with-pascal.html

QUADRO FOR MOBILE WORKSTATIONS
https://www.nvidia.com/object/quadro-for-mobile-workstations.html

 

http://www.notebookcheck.com/Test-Alienware-15-R3-Laptop.194724.0.html

On the subject of crappy and uneven heatsinks - on the stress test, there is a 20C difference between two cores.

 

And more Info out to the people
Specializing in Razer laptops, enthusiast repairman aims to fix what Razer can't

http://www.notebookcheck.net/Specia...an-aims-to-fix-what-Razer-can-t.195612.0.html

 

Sorry double post. Problem with editing post in the forum site.
WRONG There is a 23C difference between two cores. In my math is 97 ÷ 74 = 23

 

Haha. One of the biggest core differentials I've seen recently.

Sent from my SM-G935F using Tapatalk

 

Ha! I knew that component level repair would come back in fashion as these turd companies make thin stuff that breaks more easy more often and more expensive, and lock out tweakers from making tools and spreading the information to help folks out so they don't have to chuck a $3000 laptop in the bin

Any electrical engineer not fixing these dinner tray firebombs on the side is missing out on easy money.

 

" Henry believes most of these failures are caused by "incredible heat and inadequate cooling", as the components used are operating near or at their maximum heat specifications. The clock chip, for example, is rated for a maximum temperature of 85C, yet is located directly under the heatsink."

Brilliant job Razer. Confirms what we already knew. Surprised Linus is such a staunch supporter of them.

 

Agreed. Just going through the Blade threads makes me cringe about some of the things users have had happen to them.

Sent from my SM-G935F using Tapatalk

 

When can we expect HDMI 2.1 to be in our laptop builds?

HDMI 2.1 as Fast As Possible

 

At this rate? like in 10 years. Even today we have issues finding if a laptop has HDMI 2.0!

My hope is in display port tech. I can live with adapters if necessary, but HDMI support has been weird haha.

 

Yeah, totally agree. I mean even a premium laptop like the GT62VR only has HDMI 1.4 - for HDMI 2.0, you need to go to the Tornado F5.

Sent from my SM-G935F using Tapatalk

 

For whatever reason, MSI isn't listing the HDMI as 2.0, instead for all their new models they say "1x (4K @ 60Hz) HDMI", which essentially is what HDMI 2.0 offers that we are looking for - this has been the same since Skylake models released, and continues with the Kabylake models.

The GT62VR 7RE (Kabylake) Specifications lists this specification as well
1x (4K @ 60Hz) HDMI
https://www.msi.com/Laptop/GT62VR-7RE-Dominator-Pro.html#hero-specification

So you don't *need* to go to the MSI 16L13, from any boutique vendor to get HDMI 4k @ 60hz, it's available from the retail boxed MSI models that list that specification in it's product page

 

Thanks for the clarification. Jesus MSI, please get your specifications right on the spec sheet (like the way they list the weight for every damn model wrong), otherwise you are just confusing customers and giving them less reason to buy your product.

Sent from my SM-G935F using Tapatalk