New Clevos with Max-Q?

Yes it will always be on but so is power monitoring on all the cards. Just the parameters can be changed.

 

So if a notebook chassis accepts (and adequately cools) the non-Max Q version of a GPU then the Max Q version of said GPU can be brought to non-Max Q levels (inside the same chassis)? For example, the 1070MQ in the P950HR being brought to 1070 levels (assuming the chassis can withstand it).

 

Options are left open the manufacturer I believe.

 

Would be nice to know more about how this works and the possibilities it entails. I wish things were more transparent.

Thanks for your answer

 

So for a dude who hates heavy laptops so much ( i move a lot) but also hates heating issues.. does max-q laptops worth it? i've been looking for a laptop that fits my needs for a while and found 1070 SAGER NP8952 (CLEVO P950HP6) from XoticPC. it's light + good specs. i was like i found the perfect machine

will i face huge performance difference from regular sized laptops? I was getting P650HS-G 1070 from hidevolution but honestly any thing lighter will be better, however if the performance difference is huge then it doesn't matter

my budget is 1600$ if anyone has a better option for me i'd be thankful

 

Assuming it performs somewhat like the 1070MQ version of the MSI GS73 (the only Max-Q 1070 in a normally 1060 chassis I've found) then it should be a good buy. In the case of the GS73 1070 MaxQ it's both faster and quieter than it's 1060 variant but still slower than a full 1070 by about 15%.

It always poays to wait for reviews though. One thing I haven't been able to find out is the G-Sync situation on the P950. Some resellers list Optimus only, some list dGPU/G-Sync switching (like the P600 series), some list 60hz panels only and Clevo's site lists a 120hz option. If you could get a 120hz/G-Sync/P950/1070MQ you'd have a damn nice machine.

 

I prefer optimus than G-Sync.
Switching to IGP can give me an extra 1.5 hours of battery life (non-gaming).

But if im not mistaken, Optimus + Pascal caused stuttering?

 

We both know that under real usage and load the TDP ratings are just a general guideline, not necessarily enforced, and often exceeded.

If you watch closely in the video reviews benchmarks that show noise and thermal loads live, the 39dB noise limit (fans ramp up) and the 90w-100w power limit (temps ramp up) are exceeded.

Is the 1070 (125w) actually 15w higher usage at the same performance as the Max-Q 1080 (110w) under load?

Someone will need to test this IRL with a 1070 AW 15 side by side with a Max-Q 1080 AW 15.

It would be nice to see the effort put in to Max-Q a full power / cooling 1070, and the same for a 1080 in their "native" full power full cooling habituate.

Both with new fans and chassis designs to baffle noise, new monitoring and tuning to bring the noise, temperature, and performance to the task the same as the Max-Q Design - but with full power and cooling.

That's the direction those new benefits should have been invested.

To get the same high performance out of the same GPU's, with slimmer chassis, noise limiting fans, more efficient lighter cooling, better power and cooling management software, and all at the same price, that's a goal worth investing in.

 

Asus ROG Strix GL520VS with 120Hz G-Sync IPS Panel , it's thin and light , it does not overheat like Max-Q laptop, it has GTX 1070 which is almost on par with GTX 1080 Max-Q in benchmark (Real GTX 1080 scores 31k in 3d mark 11)

 

@iunlock
@DeeX , I saw your comment in the Alienware Max-Q video, and thought you'd be interested in @Thousandmagister 's comparison where the Asus Max-Q is showing obviously waaay hotter chassis top temperatures as compared to another thin and light 1070.

We don't have the temps yet for the Alienware 1070 OC 15.6" or the Alienware Max-Q 1080, but that would be very interesting.

Andrew Deex 3 days ago
"We must OC a MaxQ. If we can get it to the same score (or more) as the 1070 it will be better as it will be cooler and will take less juice! Now to get one!"

I've asked all the owners of Max-Q 1080 laptops that have posted about tuning availability for the Max-Q Limits, fan tuning, and any other performance adjustable settings and they've all said it's locked down, no user available tuning has been found.

If you get a Max-Q 1080 please do post how far you can OC with vendor / Nvidia provided tools if any was possible, and what required Intel XTU / MSI AB + RTSS tools for additional OC, if any.

Given the video's showing live benchmarking of the new Max-Q 1080's (Asus), and from posts from new owners, it's clear the fan's are jumping past the 39dba noise level's at real loads and the below par "thin" cooling is heating up well beyond normal levels for 1070 level performance - see @Thousandmagister 's comparison image:


Also given the PSU's are in the 200w-230w range there isn't enough power to OC any more meaningful performance from the Max-Q 1080's, that's all you get.

Maybe you can use your contacts at Alienware to get a pair of 1070 OC + Max-Q 1080's and post the power usage, thermals, all at the same performance levels?

 

After looking at all the benchmarks, on noteb.com GTX 1080 Max-Q is 83% the performance of the GTX 1080 (laptops) . The GTX 1070 is 80%.

 

And an AW 15 1070 OC (now stock) is 85%...

And, any full 1070 with a large enough PSU (240w) should be able to OC to the same 85%+.

So far, all of the new owners of Max-Q 1080's can't find any way to tune their Max-Q limits - unlock them - so they can try to OC the Max-Q 1080.

There isn't enough power or cooling to OC any further anyway, it's dancing along the sharp edge of thermal and power limits already at it's stock performance, by design.

 

Price wise it doesn't make sense, however the GTX 1080 Max-Q should run quieter and cooler than the GTX 1070 non-OC right ?

 

Doesn't common sense tell you that a 1080 and a 1070 doing the same work, burning close to the same power, with a die and carrier for heat dissipation will produce approximately - close enough - the same amount of thermal energy?

These aren't "magical" 1080's, they are the normal complete and full real 1080, just downclocked to 1070 performance level... the same as a real 1070... is it starting to sink in?

Didn't you look at the thermal map / temps image I've posted a "few" times?

Spoiler

The 1070 is clearly running cooling in a better chassis built to handle the thermal output from such high performance GPU's, the skinny little postage stamp of metal that the Max-Q laptops use isn't near enough.

Again, the only reason the 1080 GPU was used was because Nvidia wants to sell more 1080 GPU's, giving the perception of increased performance and a reason for the high price. That's all.

To Nvidia they cost the same. Same size die, same fraction of silicon real estate, same cost.

But, if Nvidia uses a 1080 GPU they can call it a "1080", and not a "1070", as it should be called. So they can charge more money, the same money as a full performance 1080 laptop.

A 1070 performs the same given proper tuning, given the same attention to tuning as the 1080 product.

Not a 1070 Max-Q, a full 1070. Would there be physical differences in the product, maybe yes, maybe no, it's all a matter of tuning for the best performance with new fan design and creative case heat dissipation.

Perhaps a more traditional layout for keyboard and trackpad - which would be deeper and give more internal volume and space for cooling hardware - with the same thin profile.

IDK if there would be 5w-10w more than 90w-110w, or not.

Nvidia won't do it because it's too late, they've placed their bets on using the 1080

 

I see. However, I am not sure if the above image is necessarily representative. Just neat picking here maybe. But I know that for the Zephyrus, in order to keep it thin and not very loud, they had this large cooling plate that basically dumps the heat into the chassis, hence why the chassis is so much warmer. I am more interested in GPU die temperature rather the laptop temperature. How do those faire? Just trying to see what would the advantages be between the GTX 1070 and a GTX 1080 Max-Q. For now the only one would be quieter fans right ?

 

Doesn't common sense tell you that a 1080 and a 1070 doing the same work, burning close to the same power, with a die and carrier for heat dissipation will produce approximately - close enough - the same amount of thermal energy?

And, if both are introducing the same amount of thermal energy into the laptop, doing the same "work", the laptop will heat up similarly for either GPU, right?

These aren't "magical" 1080's, they are the normal complete and full real 1080, just downclocked to 1070 performance level... the same as a real 1070... is it starting to sink in?

 

Indeed, but the GTX 1070 as far as I know is rated at 140-150W while the GTX 1080 Max-Q is rated at 110W .

 

You've got your numbers wrong all over... let me help.

Max-Q 1080 = 90w, AW bumped it up to 110w. Official Max-Q 1080 range is 90w-110w.

AW made a big deal that the "other" Max-Q 1080's were 90w, while AW pushed to the maximum of 110w.

GTX 1070 = AW was 100w, they bumped it up to 125w in the new OC release.

But, those aren't necessarily the limits, as Zephyrus owners have posted seeing 100w+ already, that's with 1st day observations, likely from visual update monitoring tool refresh - if they log at sub 1 sec intervals they might see higher peaks.

The efficiency of more cores vs higher frequency, the age old conundrum, making fewer high performance core vs more lower performance cores, to match the same TDP - usually socket limited for CPU's.

The 1080 vs 1070 with proper tuning would be less than the 10w-15w range between full 1070 OC and high Max-Q 1080 TDP

That's a very small amount of thermal energy as a percentage, and as an absolute amount.

I think Asus is up to the challenge of squeezing out 10w-15w more thermal dissipation out of a thin laptop, if they were given the go ahead for a 1070 based super thin gaming laptop.

I think that would have dropped the end user price back down to the $1700 range, plus or minus whatever for goodies

With such an easy straightforward story to tell the customer, "Hey, it's a super thin 1070 laptop!", Asus could have delivered a range of models, from lightly adorned to top shelf, and made far more sales across a wide range of happy customers.

Instead of gouging each customer another $1000 for the same performance as their previous thinnest 1070 laptop.

Hopefully this Max-Q experiment will fail forward with a 1 level down priced delivery of new thin laptops using full performance GPU's.

I'd like to see how thin a full performance 1080 can be made using the same advanced cooling, fans, etc would be.

 

Max-Q 1080 minimum is 90W, Alienware are still in the official TDP envelope at 110W.

 

For the GTX 1070 (laptop) is 120W right ? Or is it lower?

 

115w up to mentioned 125W

 

I'm pretty sure the 15" 1070 came out at 100w, maybe you are thinking of the 17"?

 

recommended TDP is set by Nvidia, manufacturers can play around with clock speeds and voltages to change the TDP to whatever they want, see the GTX 1080 Razer version.

From notebookcheck:
The power consumption of the GTX 1070 is rumored to be slightly above the old GeForce GTX 980M and therefore around 120 Watt (including memory). Therefore, the 1070 can only be found in big and clunky gaming laptops. A low power (90 W) variant with less power consumption was released at Computex 2017 with reduced performance for thin and light laptops.

Official GTX 1070 Max-Q is rated at 80-90W: https://www.nvidia.com/en-us/geforce/products/10series/laptops/ , I am fairly certain a non-Max-Q version is around 120W. Anyway, You were about right, the GTX 1080 Max-Q and the GTX 1070 are roughly same power consumption. It would seem that you get 10% better performance from the GTX 1080 Max-Q at the same TDP as the GTX 1070. No way worth the price difference.

 

Yup, and the TDP is definitely close enough that a nice thin full performance 1070 laptop could have been made similar to the Zephyrus.

The TDP range for all the 10xx series GPU's was initially advertised, and then manufacturers went their own way, tuning the power for their chassis and cooling, that's why we get variations in performance results for the same GPU, dependent on the power and cooling spec'd for a laptop.

For example, the 1080 was supposed to be 180w (up to 200w) on mobile and desktop, but most chassis couldn't handle that, and all except the top Clevo air-cooled and Asus water-cooled laptops had less TDP.

Again, those were stock tuned, with OC the power draw can grow much higher.

I don't have the interest or energy right how to dredge up all the TDP stock values we've discovered and pooled together over time, but there are several people active in this thread that can fill you in too.

I was trying to focus on the Max-Q 1080 and AW 15" 1070/1070 OC TDP numbers, as they are the only relevant ones for this discussion, it's easy to get wildly off topic quickly with very interesting seemingly related details.

Although to see and contrast the wide range of power limits used in 14,15,17,18" laptops for all the ranges of Pascal GPU's, is definitely worth seeing, especially with sample images
@Papusan ?

 

The 1070 needs higher clocks to match the 1080 and uses GDDR5. The savings come from having more units at lower clocks and voltages and using GDDR5x.

 

Strangely the DDR5x memory suffers from performance issues with applications sensitive to latency.

Even though the throughput is higher with DDR5x, surprisingly weak latency has made the 1070 w/DDR5 outperform for Crytocurrency applications, by a wide margin.

As shown by AW, it's pretty simple to tune up the 1070 vBIOS to match the advantage's brought by a wasted downclocked 1080 shoehorned into a solution better suited for a real 1070.

That is, a 1070 makes a better 1070 than a 1080, and it does this for a much lower price to the customer.

 

Origin PC Shows Off the Evo 15-S With Max-Q Design

 

@1:01

"Its got a 1070 and it's like all the other cards in the 10 range this year. They basically as good inside a notebook as they would be in your desktop right."

 

The marketing is real with them. They are pushing it for the uninformed.

 

I'm not sure for BGA, but 1070 MMX 115W
Expect lower for BGA as MMX will use slightly higher power.

 

Games don't care about latency, are we seriously discussing laptops as mining machines now?

Show me the numbers where the 1070 is performing the same at the 1080 for the same power. If not then it's an expensive option sure, but for those with the cash they can get the best option.

 

Correct on Max-Q - as good as correct on normal GTX 1070; as discussed in below quotes it's 115W.

 

You're the one that brought up the idea that DDR5x was an advantage for the 1080 over the 1070 DDR5, I'm only pointing out that in some applications DDR5x can be a detriment instead of an advantage.

It doesn't matter whether you do cryptocurrency mining on a laptop, it's a matter of showing there are disadvantages of DDR5x.

It would be interesting to see a DDR5 version of the 1080 and 1080ti to be able to see in direct comparison where DDR5x helps performance, and where DDR5x makes performance worse.

Spoiler: I did the googling for you...

I don't do cryptocurrency mining, but due to the GPU shortage caused by the mining I've heard the issues surrounding the use of various cards, and it stuck in my mind when I heard about the 1080 hashing slower than a 1070, and the reason given was the latency of the DDR5x.

1080 Specific Ethereum Mining Issues
https://forum.ethereum.org/discussion/9277/1080-specific-ethereum-mining-issues

"4) How fast will the 1080 mine when "fixed" (opinion zone)

On Ethereum mining, the GTX 1070 is roughly 20% faster than the 1080 right out of the box (24 Mh/s verse 21 Mh/s). Yep, sounds broken.

5) Known Issue: GDDR5X (1080) isn't performing as well as GDDR5 (1070) on mining

GDDR5x is better on spec, but GDDR5X is new tech and doesn't appear to out shine GDDR5 on mining... yet."

These Are The Best GPUs For Ethereum Crypto-Mining
https://www.forbes.com/sites/daveal...gpus-for-ethereum-crypto-mining/#5bd144762675

In this 2017 article it looks like the additional performance of the GTX1080ti powers through the DDR5x problem enough to squeak by a 1070 in hashing, by 0.696, with 31.775 vs 31.079, well below what one would think a 1080ti would do in comparison to a 1070. The article doesn't even waste an entry for the GTX 1080.

If you check availabilty and pricing for the 1080 and 1080ti, you'll find plenty available and at stock or discounted pricing. Mining doesn't have any demand for those GPU's, due to DDR5x and the low hash rate / dollar / watt.

After I made that post about DDR5x being a detriment for the 1080 in some applications, it dawned on me this may be another the reason Nvidia selected using a 1080 instead of a 1070 in the super thin laptop project.

If Nvidia can sell 1070 GPU's in Crypto Mining cards, as well as gaming GPU's, it's possible Nvidia doesn't want to use up the 1070's in the thin laptop project.

With lower sales of the 1080 than the 1070 for both gaming and mining, due to expense and performance respectively, Nvidia needs places to sell the 1080's and their stock of DDR5X, so why not detune the 1080 and make them pseudo 1070's?

Of course I really believe it's their ability to use the "1080" name in the advertising of the laptop as the main reason. But, maybe that's really just a nice marketing side effect of the real reason - Nvidia's got too many 1080's sitting around not getting sold?

Either way, using a 1080 in a thin laptop only to detune it to 1070 performance is stupid

 

Yeah, this is typical of what I've seen in all the reviews. They either say it's got a "1080" or a "1070", completely dropping the "Max-Q", and then draw comparisons based on the "real" GPU's, not the performance reduced "Max-Q" results a potential owner will see.

It's exactly what we talked about early on, the "Max-Q 1080" performance will be assumed to be that of a real "1080" because of the name.

I really do hope disappointed owners return their "Max-Q" laptops, and not slink into silence. They shouldn't get stuck with these things.

The problem is most people don't test for performance, they just use them, so they might never know, unless someone tells them.

 

That's highly unlikely. The 1080 and 1070 are both the same GP104 core anyway.

More to the point, the mobile 1070 is actually different to the desktop 1070 used in mining so I don't think it has to do with saving up any chips.

NV are only making dedicated mining cards out of the GP106 which probably has a higher yield level. The GP106 chips are using Samsung's 14nm process, whereas the GP104 chips are the TSMC 16nm process. They might just be making more on the 14nm process.

According to the NBC test , the typical 3DMark workload operates ~10W below a 1070 (comparing the Zephyrus to the P651-HS-G which is very similarly specced). In fact, the 3Dmark loop power usage is more like some 1060 machines. That translates directly into battery time (assuming you actually put a decent battery in there, which the Zephyrus didn't...).

So for the purposes of gaming, they've pretty much accomplished their goal there.

 

Do you really think AW would have gone through the trouble to increase the 1070 OC TDP to 125w from 115w? Maybe someone can verify from AW what the 1070 TDP was before the recent OC; I thought it was 100w for the AW 15 1070.

Spoiler: Alienware Configurator showing 125w 1070

 

My 15R3's 1070 Power limiter is set to 115W

 

Excellent, an actual reading off the laptop, thank you!!

So that's only a 10w bump up on the TDP from stock 1070, and 15w bump up from the AW 15 Max-Q 1080 110w to match performance.

 

You're welcome. Those are very interesting datas you're collecting. If people wan't to check their GPU's Power Limit, they can dump their Vbios with Nvflash and read this value with Mobile Pascal TDP Tweaker

 

You're comparing ^^^^ with the 125W 1070 OC TDP if I understand correctly?

 

Yeah, I guess it must be hard to follow along as we gather the info's

New AW 15 stock (max Max-Q) Max-Q 1080 => 110w TDP.
Previous AW 15 stock 1070 => 115w TDP +5w
New AW 15 stock (now OC) 1070 => 125w TDP +15w (+10 from previous 1070)

So that's a 10w increase from 1070 to 1070 OC, which is now 15w higher than the Max-Q 1080.

Not a very big range of TDP change, making the full power 1070 OC easily cooled in a slim chassis much like the Zephyrus with the same performance as the Max-Q 1080.

Or even easier to cool, use the lower TDP full 1070 for the slim chassis Zephyrus like laptop.

Not a Max-Q 1070, but a full power full performance 1070, and charge $1000 less than the Zephyrus Max-Q 1080, which also provides 1070 level performance.

 

A) GDDR5x is better in every way for gamers.

B) Mining has only increased sales, the 1080 is shifting plenty of units and due to price pressure on the 1070 the 1080 becomes a no brainer for gamers so they have no difficulty in shifting cores at the moment.

C) Nvidia can just laser cut 1080 to 1070 if they really want to.

D) 1080 max-q designs are the niche of this launch and is not significant in volume levels.

 

@hmscott - great detective work will be following this closely. I totally agree when you say that the 1080 Max-Q is a rip-off.

By the way guys, what is the TDP of the 1080N in the Aorus X7 DT v7 ? I'm searching and I can't find a mention of it. Is it 150W? 180W? 200W?

@D2 Ultima (great article by you and @Eason on Notebookcheck, BTW) @Donald@HIDevolution @Kevin@GenTechPC

Sent from my SM-G935F using Tapatalk

 

Manufacturers tend not give out official specs for the TDP of the cards/chips.

 

Let's hope Max-Q dies a quiet death, and never returns to haunt us again.

Full power GPU's for laptops, FTW!!

 

Your cheering for less choice as a consumer.....

I am not saying you have to go out buying these machines but you want to deny this particular choice of form factor/performance/cost to someone just because it does not suit you.

 

I'm not saying to not make a slim laptop at all, I'm saying don't do it in a deceptive way.

I am saying, don't market a 1070 level performance laptop as a 1080 - that *mumble* Max-Q's and whizz's and bang's in mysterious and magical ways to deliver 1070 level performance.

To add injury to insult they then grab another $1000 out of the consumers wallet for no good reason.

I am asking for them to be honest, straightforward, and truthful.

Market a 1070 performance level laptop as a 1070, presented as an evolution of the previous slim 1070 laptops, and charge a fair price.

 

It is a 1080s power efficiency and it's the same chip just tuned for a different form factor.

You don't pay less for a core i7 mobile chip compared to the desktop for instance.

I disagree the price should be less of the lower power version, I think the price of both versions should be lower.

 

Are you on board with advertising the Max-Q 1080's as such?

Already people in review video's are calling them 1080's, dropping the Max-Q, and some even saying that they are performing only 10% slower than the desktop 1080, which of course is impossible.

This naming problem, resulting in deceptive marketing, selling these 1070 performance level laptops as "1080" is the issue.

Are you on board as a vendor with these intentional deceptions of the consumer?

 

It's not impossible for these chips to come within 10% of a 1080 in certain circumstances if you think about it without even seeing the numbers but where are you seeing that?