New Clevos with Max-Q?

Those claims would be surprising if true. I'd love to see a side-by-side comparison with all their competition.

 

You mean they want to try and sell you theirs as the best? Shocking

 

Nvidia has really f***ed up this time. They released the new GPUs waaayyy to early to provide a feasible alternative to current GPUs. The current GPUs are much much better than the Max-Q GPUs in the way that matters, namely power and price. Max-Q only has slightly better thermals and power consumption for what? An extra 500$ price tag? This is really dissapointing.....

 

Max-Q is more like The Emperor's New Clothes, Same graphics but with lower performance vs. The name brother, weaker components and the same high price. Yoo shouldn't call this NEW GPUs!! Call it what it is... Milking and in the end... Scam!!

 

As disappointing as E3 2017. It would seems this years new technology sucks.

 

dat age of empires remastered edition tho! love it and want it!

classics FTW

Sent from my HUAWEI NXT-AL10 using Tapatalk

 

Two Reddit threads where users undervolted their desktop GPUs, resulting in massive power consumption and noise reductions:

https://www.reddit.com/r/nvidia/comments/6hmk55/pascal_undervolting_really_deserves_more_attention

https://www.reddit.com/r/nvidia/com..._we_have_silicon_lottery_for_graphics/diy1t63

I believe this was what Max-Q is all about.

 

No, it's not at all what Max-Q is about

It's about reducing noise, putting in a 1 size to large GPU so they can charge more money for the same performance as the 1 size down GPU, and wave their hands in special magical ways to convince people to pay more for less.

There is no technical logic to the approach, it buys nothing.

Which does seem to be the case in those threads as well, reduction in voltage caused drop in temp and less fan noise but also huge drops in performance.

So if you mean this is what happens when you try to detune a GPU "like" Max-Q is doing, then that's true.

It's just not true that "...this was what Max-Q is all about"...

Max-Q is not worth wasting brain power over trying to figure out the technical why's and wherefore's - there's no technical logic to it.

The why is to make money from nothing, providing nothing, and likely to get rid of overstock 1080 and 1070 GPU chips as most people are buying 1060's.

Nvidia saw how many suckers where paying big $$$$'s for Razer's 1080 that only performed like a 1070 due to not enough power (PSU is 250w instead of 330w as required).

Nvidia also saw how the suckers were paying even more for the 7820HK model - even though there isn't enough power budget for the 1080 somehow these guys think they are gonna be able to OC their CPU - nope, same performance as a locked 7700HQ.

Fortunately it looks like a lot of people are figuring this out finally, maybe the laptop manufacturers will start flying right, and providing true performance for the $, and try to help educate their customers in what's possible:

Magical Thin laptops that have the power of big thick weighty high power laptops aren't possible.

 

You don't need to convince me, I'm not arguing for Max-Q.

I was just giving nVidia's technical premise behind Max-Q—which, based on the two links I gave, appears to be valid. Like I said earlier, silicon always has diminishing marginal returns as one jacks up voltage and clocks. Look at Ryzen 7—it achieves nearly 900 points on CB R15, when clocked at 3 GHz and undervolted appropriately, while drawing a mere 30 W. On the other hand, the potential difference required for > 4 GHz on Ryzen is ridiculously high. Every bit of silicon has a voltage wall. It turns out that Pascal's efficiency skyrockets with even a mild undervolt and underclock, with significant differences in temperatures and noise. Apparently performance isn't that different, either. The problem, however, is that the price is going to be horrendously high, and this is the main issue.

For all intents and purposes, this is not necessarily a return to the old 'M' days, but I want to wait and see. I am fairly sceptical about the whole affair, but I would like to be pleasantly surprised.

 

Right on target. +rep M-I-L-K-I-N-G twice!!!

 

I also undervolted my regular GTX 1080 (Notebook) using afterburner's Voltage/Frequency curve editor and it dropped my GPU temperatures by 10-18°C (I remain in the lower to mid 70°Cs with max UHD load in games, professional rendering applications and stability/stresstests whereas before it would go up to 90°C, throttling like mad and the fans would become rather loud) with Clevo Control Center's "Overclocking" fan profile all while keeping the clocks steady and high (quite a bit more than it would clock without the undervolt) without any throttling and the fans relatively quiet (CPU is undervolted as well). The system has been perfectly stable as well. It takes a bit of tweaking and testing what works for one's GPU until you find a stable and cool sweetspot, but it's definitely possible even with the Stock BIOS provided the silicone lottery hasn't awarded you a turd that requires stock voltage. No need for a "Max-Q" for that, a regular "Min-Q" should do just fine.

 

Good point. They're just marketing and selling a fact of all GPUs as a special feature of Max-Q, and jacking up the price. They ought to just release a tweaking guide for GPUs and be done...

Anyway, the silicon lottery applies much less to GPUs than CPUs, because of the much higher core count...

 

Is Clevo brewing something new "non MAX-Q BS" or we ain't gonna see a new system anytime soon?

 

So much overclocking potential wasted... Back in the day when I got my M570TU, it was all about overclocking it. I ran my 280M at 580MHz from 500 Mhz stock for 4 years until it kicked the bucket. Temps would get around 80-85C during long gaming sessions and especially during those hot summer with no AC.

 

Back in those days the mobile chips were pre-nerfed meaning they had a lot more headroom. These days they are coming out the box near their desktop counterparts which have less headroom as the cards are clocking themselves closer to the edge.

 

I'd rather not have my system card kick the bucket.

 

They all do eventually... they all do . That's why we want MXM video cards.

 

Did you run any A/B tests for performance with / without the undervolt? To get an idea of the performance cost of the reduction in voltage?

Maybe test a render for your work before / after and that would give you an idea of the *cost*.

 

Depending on how this person did the unvervolt (esp with the newer laptops), the frequency curve can be set in such a manner where the full clock speeds of the GPU can still be achieved with lower voltage.

And even if the person is not getting full speeds, with thermal throttling out of the way and noise reduction it may be worth it. I hope you can be happy with people who are ok with making this tradeoff.

 

With regard to undervolting, Pascal behaves a bit weirdly and there is something going on behind the scenes which Afterburner doesn't pick up. I've been running A/B tests on my GTX 1080. At 0.950v @ 2012mhz, I score less than 0.975v @ 2012mhz. In both cases the card holds it's maximum clock of 2012 mhz.

The implication is that something is going on behind the scenes when voltage is set too low.

A good video of it in action is here:



That video demonstrates that undervolting your card too low, even if your Pascal card runs the same clocks, is affecting performance due to some internal management of Pascal that Afterburner doesn't report. Either the card is not actually running the set clock on the curve, or cores or shaders or whatever are being dynamically disabled due to the detection of low voltage.

So the key with undervolting Pascal is to find the sweet spot on the volt/clock line where you're not giving up performance.

For example, my GTX 1080's sweet spot is 0.975v at 2012mhz. When I raise my voltage to 1v, the score at 2012mhz doesn't improve, but when I lower my voltage to 0.950v, my score at 2012mhz worsens, even though I stay at the same clock speed.

 

It's normal. If the voltage is too low, the hardware will compensate (by either doing error correction, re-sending signals, etc. - unless it's too much to compensate, at which point there will be a crash). It's the same with CPUs, too (albeit it may not be as noticeable as they tend to crash the system quicker). That's why you have to tweak things, you need to find the voltage/clock sweetspot for your unit where it performs optimally. It takes time and patience but in the end it's worth it. Also note that this sweetspot can change slightly over time.

 

I could be wrong, but I would say this behavior by Pascal is actually a little abnormal when we consider past generations of graphics cards. Graphics cards in the past would usually either hard-locked or driver crash when voltage was insufficient for the clocks that were being attempted to run.

With Pascal, we are actually able to lower voltage and still report a full clock speed, but at worse performance, which means something else is going on, such as disabled cores or some other internal power management.

I don't remember this behavior in other generations of graphics cards, though I skipped the Fury, and never owned a Maxwell. If there are other generations that behaved this way, I would love to know them to be more informed

As for CPUs, I don't really understand what you mean. If I try to undervolt my 7700HQ by -200mv and I run a stress test, my PC will lock up and reboot. Clearly, -200mv is not enough to run the rated clock speed.

With Pascal, if I undervolt to 0.9v, I still report "2012mhz" but my performance is nothing like 2012mhz @ 0.962v. So that way, assuming Afterburner is reporting accurate clocks, there is other management going on internally.

The analog to this with CPU would be if I set a -200mv undervolt, and my CPU still boosted to it's max 4-core 3.4ghz turbo, and yet my CPU benchmark scores were lower. This, however, is not what happens. What happens is I hard-lock and reboot.

I do agree that you need to find a sweet spot for sure, especially because ultimately you don't want to compromise performance.

I initially made my post on Pascal undervolting because I discovered that it's actually quite easy to compromise your performance on Pascal by undervolting without realising because your card will still boost up to the clock rate you set and be "stable", but you'll receive less performance for it. Many people may undervolt their Pascal's to 900mv at 2ghz and think sweet it passes a FS 1 hour loop, and not realise they're performing lower because they're not paying attention to performance as much as stability, because they assume performance remains the same at the same clock speed.

 

It's just another strategy for renaming GPU (680M, 780M, 880M) or holding one gen for 2 years (980M). This time they just downlocked them and say they are pro awesome for pancakes class laptops. They forgot to mention that they will be pathetically weak after.

I don't see any rational reason for Max-Q except trying to push more GP cores and hold longer til' volta. More profit per year with one technology investment.

 

A downclocked 1080 for 1080p is still more than capable.

Nvidia have taken it one way with advertising and the people on here have gone the opposite way *sigh*

 

Is that a sarcasm ? Most GTX 1080 owners don't game at 1080p . For 1080p , GTX 1060 is more than enough . You could buy a GTX 1070 gaming laptop with G-Sync for like $1400~$1600 and still have $1000 to throw at a big party or a new PC/ultrabook
Or if you really want a thin gaming notebook that bad , you could buy Gigabyte Aero 15 , this thing can withstand Furmark + Aida64 extreme test and it's a whole lot cheaper than MaxQ laptop
Seriously , who would pay 2 grand for this crap ? Who this is for ? Knowing GTX 1080MQ is not faster than a normal GTX 1070 (according to that Dell guy) , the CPU is most likely to throttle to da max

If MaxQ laptops sell well in the future, full power laptop will be hard to find . Manufacturers only care about profit anyway . Nowadays , it's hard to find a gaming laptop that has removable batterry , most laptops have built-in battery because of "laptop must be thin" kind of trend

 

You noticed that too, huh?

That's not exactly a step backward. Big desktop replacement laptops are a niche market.

 

I bought my first business notebook in 2013 , just like any other "light-weight" laptops at that time, it has removable battery. One year later , ultrabook became popular in Japan and that's the end of tradional notebook....

 

What a surprise, Nvidia has marketed Max-Q so as to sell it to people that are easily fooled into believing that they can have a full power GTX 1080 in a slim laptop. Just like the Razer 1080.

We worked out the truth about Max-Q, confirmed by Alienware - as part of their marketing to "one up" the "weaker Max-Q" laptops, it's a GTX 1080 in name only, performance wise it's no better than a GTX 1070.

We have taken Max-Q at face value, for what it is worth, and it's not worth the asking price.

And, you still suggest Max-Q has value because "a down-clocked 1080 for 1080p is still more than capable"...

You are too close to the "drink the kool-aid" dispensary.

Paying big bucks for a "down-clocked" GTX 1080 only to get GTX 1070 performance, has no value.

If all the vendors selling GTX 1080 Max-Q laptops price them at their true value as GTX 1070 laptops, then Max-Q will be an interesting curiosity at best => "Why would laptop makers put a perfectly good GTX 1080 GPU into a laptop only to deliver GTX 1070 like performance?"

At least at it's proper cost for performance value Max-Q would then be fairly priced.

Better yet, correctly name the GPU to a GTX1070 - call it a GTX 1070 special edition if needs be but don't call it a GTX 1080 as that will lead people to believe they are buying GTX 1080 level performance.

Purchase would then be a matter of preference in the final package vs trying to buy GTX 1080 performance in a thin package, which Max-Q does not deliver.

 

You're own needs and requirements are blinding you from looking at the needs of others.

Let's say I want the best experience possible but I had a hard limit on weight of 2kg.

As a consumer you are saying I should be stripped of the choice of getting a 1080 chip lower clocked to fit the form factor and instead should settle for a downclocked 1070 and get overall lower performance, that I am not allowed to buy the best chip possible because its a waste not to fully clock the chip.

That's the same logic desktop gamers use to dismiss laptop gamers that I have had to listen to time and again and now to see this community trotting it out is quite disappointing.

Nvidia are giving it a moniker, they are not just calling it the same thing. People will be able to compare benchmarks and make a value comparison. Will it be the right choice for everyone? No. But then again what is?

I don't see why we should strip choice from people when you believe some may be led astray. That's belittling the consumer.

 

citation needed

I keep seeing people regurgitate this and yet with zero research. The "Big Bucks" going on here is almost exclusively based on the super high end Max-Q pricing for one-of-a-kind models with no point of reference. Previously it was things like the Razer Blade Pro which had such high prices and the fact is, people who bought those cared little about perf/$$ anyway.

For reference, the only price I can see online for the ASUS Zephyrus is AU$3600 for the 1080MQ and yet a P870KM with a GTX1070 costs $3999. So...

All the models I can see which actually have multiple variants with lower GPUs (ie MSI GS63/73) are only about 15-20% more expensive for the next "Max-Q" model up. Most of these are also the 1060-1070MQ change, which as I've said before, is what will impact the most buyers. Lets be real here, very few people even have the budget for high end models 1080 models, MaxQ or not.

We still don't have good bench tests either. We've got 1 3Dmark score (and it's 3DMark 11..I mean come on....) which is a combined score with CPU taken into account and nothing else. No power consumption or thermals.

I think it's because too many people here seem to believe they're better qualified to spend other consumers' money and are quick to lose sight of the fact that pure performance is not the sole factor in a laptop purchase

 

Yeah, dammit... don't forget about NVIDIA's need to make money, right? It's OK (with them) if customers get screwed in the process... just... collateral damage. Sell a castrated 1080 that runs like an overclocked 1060 and charge as much for it as a regular 1080, give or take a few bucks. Yeah, that's the ticket. That's customer love, Green Goblin style. And, never mind any of the adverse aspects of this. It's all for the great cause of advancing turdbooks that are more emasculated than the last batch. Optics are the only thing that matter here. And, don't forget... Max-Q is also the answer for global warming, third-world genocide, Islamic terrorism, acne, warts, and the suffering of innocent animals around the globe. Now, let's all lock arms and sway to a round of Kumbaya.

 

Helping people avoid being taken advantage of is often met with that kind of defense from the deceivers.

"How can you be so sure they are going to be fooled?, you are belittling them (the consumer) by saying these things (unveiled truths).

You should let them decide for themselves, if they want to be fooled (buy it), let them be fooled (buy it)."

Which of course really means, "Hey!, we got a good thing going on here, why are you messing it up for us??"

I'm not stopping people from throwing their money away, I'm helping educate them in the full light of day that what is being sold isn't as it appears.

A 1080 Max-Q is not a full performance GTX 1080 - it's not going to give them GTX 1080 performance as they expect from a desktop or a full GTX 1080 in a laptop. Using the name 1080 is a deception.

The name GTX 1080 call's to mind top performance and power, normally only available in large, heavy, thick, "desktop replacement" laptops.

To sell a "GTX 1080" and then mumble some other made up word (Max-Q) to absolve yourself from culpability in the deception of the consumer, doesn't absolve you in any way, it's merely another mechanism of the deception.

If you were really trying to sell "another choice" or "another option" in the form of a highest performance slim profile laptop, you wouldn't use the name GTX 1080, as that would be misleading, you would use the description "GTX 1070 level performance", the highest available in a thin laptop.

People still have the choice, they can throw away $1000 more on a GTX 1080 Max-Q laptop performing at the level of a GTX 1070, or purchase a full performance GTX 1070 laptop and save $1000.

 

When you call out the name GTX 1080 "first time in a thin laptop" as a selling point, you are selling performance.

Unbelievable performance in an unbelievable package. A text book example of "if it sounds too good to be true... it's too good to be true".

A classic marketing method, presentation of a fallacy to deceive the listener into paying for more than they are getting.

It's not delivering the performance of a GTX 1080, so we are calling it out as being deceptive advertising.

That's it for a start, but there's more.

The deceptive mis-direction continues with "Quiet Operation, limited to 40db".

You can't have maximum performance, that of a GTX 1070 (using a GTX 1080) at 40db noise output.

You can have *either* Quiet operation at 40db *or* full GTX 1070 performance (using a GTX1080), those are mutually exclusive functional modes.

It's the pattern of sustained deception that is the most annoying, and actually helps point out the danger of believing anything said in the marketing of these Max-Q laptops.

MSI knows this, they haven't even used the term "Max-Q" in their product pages.

Nvidia is driving this, vendors are stuck defending it, per usual

 

Apparently (according to KitGuruTech's recent video on the pre-production model), the 1080MaxQ will be exclusive to the Zephyrus for about 6 months. Other manufacturers will follow afterwards:


Unfortunately, most of the useful info is still under NDA for 2 weeks according to them which is also nice to know. So...we'll be back in 2 weeks I guess

 

The thing about vendors defending it is the part I don't understand. They don't need to be cheerleaders for the Green Goblin, and when they're going through the motions just to make money it's just as bad as NVIDIA selling junk to keep the cash flowing.

If some of the big names would have big cajones and tell NVIDA " Hey, where not selling that crippled panty-waist [insert expletive for feces] to our customers" there would be no market for garbage to speak of. Nobody would be deceived because the effort would be DOA... an abject failure. It would dry up and blow away, like dust in the wind. Average consumers typically tip the Kool-Aid glass and go with whatever their pet brand tells them they should go with. Offering to sell it is a carte blanche endorsement. If they were telling customers " we're not putting our name on this filth, if you want garbage like that then go buy it elsewhere" most of them would not go buy it elsewhere. They would decide that if MSI or Alienware, or ASUS... whatever brand they venerate... says it's junk, then they will have no part of it. That's exactly what should have happened with BGA filth, but it didn't. The big name ODMs and OEMs where Intel butt-kissing yes men that went with the flow and spinning lies from every angle to mask the stench. And, here we are now. The cancer has metastasized and the fatality of high performance mobile computing is imminent. But, shame on the harbingers for raining on the parade, LOL.

 

I agree with you, everyone have their own needs.
I just hope the performance differences justifies the price differences.
In my country, MSI GS63VR-7RF (GTX 1060) and MSI GS63VR-7RG (GTX 1070, no MaxQ label but I assume it's MaxQ) have $825 price gap.
So, MSRP $2399 vs $3225.

 

That's a good observation, it's going to be tough to tell the Max-Q 1070 / 1060 from the full performance models if MSI doesn't tell us.

The unlabeled "Max-Q" model will likely be the thinner of the two, and touted as such by the ad copy - as the thinnest 1070 / 1060 ever!!

Worst comes to worst, pay careful attention to the model number details and match them exactly to performance reviews.

At some point these Max-Q laptops are gonna stack up in the warehouse and they will need to be unloaded a lower prices, which will make them look like bargain 1060 / 1070 laptop's, except they will run slower.

 

Don't count on that last part happening. It didn't happen when it should have before. There is no shortage on the popularity of stupid. Just because a cow patty is covered with hundreds of flies doesn't mean it's something good to eat. The flies don't know any better. They are just hungry and they have acquired a taste for manure. In fact, don't have your surprised look on your face if this trash jumps off the shelves due to deceptive marketing. We see it everywhere we look. If things were as they should be and enough people were smart enough to know better we would not have even a third of the veritable smorgasbord of castrated gamer-boy rubbish that is available today. It would be reviled and shunned as unacceptable wuss fodder.

BTW - fixed that typo for you.

 

We can only hope that the reviewers will test the limits of the Max-Q laptops in the ways the owners will expect them to perform.

Running with the 40db limit, and measuring performance.

Then running with all the limits removed and measuring performance - with temperature and noise levels noted for that full powered maximum performance.

It's likely that the 1080 Max-Q laptop at full performance will be louder than the GTX 1070 full sized laptops, and run hotter.

It remains to be seen how many people will be taken in by Max-Q. I hope most people will learn about and fully realize the limitations on performance before they purchase, so they won't be disappointed when finding out for themselves after purchase.

The resulting sales, or lack of sales should bring down the price of the Max-Q laptops to match their performance against full performance laptops.

The $1000 difference in price between a $2700 Max-Q 1080 vs. a full performance GTX 1070 @ $1700 should erode quickly.

The $1000 is a convenient round number, that will undoubtedly change as the Max-Q laptops need to be sold off, reducing cost to the point where they may sell at a fair price in the end, or even a bargain.

A 1080 Max-Q should go for the price of a 1070, a 1070 Max-Q should go for the price of a 1060, and a 1060 Max-Q should go for the price of a 1050 / 1050ti / 1060 depending on if the TDP of the 1060 Max-Q matches TDP of a standard 1060 - which has been rumored.

After reviews come out we will have a better idea of just how the Max-Q models perform as compared to their numeric namesake's, there will likely be a lot of variance in performance depending on the implementation.

If Nvidia rolls out the Max-Q technology to all Pascal laptops, then the Pascal laptops with excellent cooling and already performing at the level of their numbered name, might actually benefit from the tunings available. That will be interesting to see as well

 

Correction - If nVidia roll out Max-Q to all laptops, then everyone gets screwed and has to buy painfully expensive stuff for cards that underperform for the name they are given, even those who can see through it like you, me, @Mr. Fox and a few others.

Sent from my SM-G935F using Tapatalk

 

We are considering two different sceneries

My more positive scenario is where the Max-Q control over power and cooling is shipped in device drivers (vbios firmware too?) for all existing and future laptops for fine tuning and control of GPU's.

I'm imagining Max-Q control over GPU tuning to be rolled out to laptops already equipped with adequate power and cooling to get the best performance, quietest operation, or some user specified gradient in between.

I'm not suggesting all high-end GPU's be neutered into lower power low cooling laptop configurations

The current Max-Q laptop roll out will certainly dissuade most of the people that experience it from ever thinking of using a high performance GPU in a thin laptop incapable of powering or cooling it, ever again.

 

This mobile generation is turning out to be quite the complicated mess and screw job from Nvidia.
Nothing will change unless AMD get in the game. Even then, I fear it's too late. What an absolute s#$& show the mobile scene is in right now.
I have a lot of respect for MSI at least attempting to make upgrades an official thing - even if they handled the fallout like a bunch of amateurs. I see Clevo ( more so Prema and co and a few distributors) also fighting the good fight. But boy has the portable laptop scene been consolized.

 

Why should Nvidia charge less for a chip that was no easier to produce?

I know my views are not going to make me popular on this one but why do people feel entitled to get it cheaper? When has a smaller and lighter machine cost significantly less when it has similar capabilities?

 

I've already answered this question:

"Paying big bucks for a "down-clocked" GTX 1080 only to get GTX 1070 performance, has no value."

 

Who says a full power 1070 would fit into the same envelope the 1080 would fit into? It's likely it would need to be downclocked itself.

 

Because something underclocked 300mhz is presumably a lower binned part and the fair value price of lower binned parts is lower, not higher?

I have a feeling a lot of sloppy reviewers are going to get caught out and not notice the on-the-fly quality downgrades. I also wonder if Nvidia will do a VW and ensure they run better on benchmarks than in RL

#maxqgate

 

THIS↑↑↑↑↑ See @hmscott's post!! I wonder why they defending it And yeah, per usual!!

 

That hasn't happened, we are discussing what has happened

 

That's exactly his point, just the other way around. Why should consumers pay more for less.
The part about it being "similar" in capabilities seems to be what many people are taking issue with. Being 1070 performance I'd argue they're not anywhere similar capabilities.
All it is in my eyes is a 1080m which I don't have a problem with, other than it being more disposable junk and only recently Nvidia and others* fed everyone BS that MXM was no longer viable. Not hard to see why many people here think this stinks.

 

Now that Nvidia has begun to tamper with the design (destroying) Aka cripple the performance of its best chips. This is just the beginning... Everything they learn from the Max-Q tech, will be passed on to their next released graphics. Volta will be the next man on the list. With even more locked and crippled firmware + more project designed drivers like they now do for Max- Qrippled.