Why does sager use the slower gtx 1070 max Q in the p950ER

sager's p950hr clocks*
1101 Mhz Base
1266 Mhz Turbe
1316 Mhz during gameplay

Compared to the max Q gtx 1070 in other laptops that run at
1215 Mhz Base
1379 Mhz Turbo
1480 Mhz during gameplay

What version does the p950er use

 

Where did you get the above information and numbers from - it will help us all to answer your question. Also are you talking about the Sager version of this chassis (the Sager NP8851), or the Clevo P950ER? There's only one GPU option in this Clevo chassis and the boost clock in games goes to 1480 MHz (as our testing confirms).

 

Eurocom Q5 (same scores as the gentech p950ER review)
While lower than other gtx 1070 max q laptops



Aero 15x gtx 1070 max q (3d mark 15500 graphics score)

quote from https://m.pclab.pl/art74974-2.html about the gtx 1070 max q in P950ER :

3. Nvidia GeForce GTX 1070 Max-Q is the latest energy-saving version of the mobile system, which TDP is limited to 80-90 W, depending on the manufacturer of the laptop. The GPU has not been truncated in any way, so we still have 2048 CUDA units at our disposal, but their timing has been significantly slowed down. The base clock frequency is 1101 MHz, while in turbo mode the graphics core can accelerate to 1266 MHz. In The Witcher 3, the actual clock speed oscillated around 1200 MHz. The limitation was the power consumption limit, which is rigidly set to 80 W.

 

Overclock it back then. It has been underclocked because Clevo has crap cooling systems.

 

Sager lowered the TDP limit as will(the same TDP limit as a laptop 1060......). Even a +50 mhz oc will be hard to achieve

 

Yeah that kinda sucks.

Don't buy clevo then.

 

What other option contains a gtx "1070" for ~1600$ ?
Msi, asus , gigabyte are all in the 2000$+ mark

 

There are a couple MSI models that I see with the full 1070 in the 1600 range. But they have the kabylake processors and not the newer Coffeelake.

 

I mean this is always the case with everything in life. There is most of the time a good reason as to why a product is more expensive than another. Yes the MSI notebooks do cost more, but have better cooling, speakers, screen, keyboard, software etc.

 

Yes but I only have 1600$ for laptop + shipping not more

Btw does what your saying apply to the gs series, since almost all other MSI laptops are tanks

 

https://www.amazon.com/MSI-GS63VR-Stealth-i7-7700HQ-Keyboard/dp/B072K85Z7N - oh look, someone has put 1070 Max-Q because they have ****ty cooling system! Or maybe because the laptop is only 0.7 inch thick? Who knows? Don't mind the price. And don't mind bad reviews either, it must've been clevo fanboiz.

 

From https://www.hardwareluxx.de/index.p...ter-allrounder-mit-max-q-im-test.html?start=1 review, english translation:

"The Max-Q variant has to deal with speeds of 1,215 and 1,379 MHz, respectively, which are clearly outbid by our SCHENKER KEY 15 at around 1,544 MHz."

Can we differentiate Sager from Clevo so that the conversation doesn't get too confused. Clevo manufacturers the chassis, Sager is a reseller/system integrator like XMG, Eurocom etc. Sager haven't limited the TDP on any laptop we're discussing here.

All the figures posted so far are from reviews of the HR version (Kaby Lake, not the ER as the OP title states), I appreciate that people might think that the same GPU performance might be reached in the ER version but that's not actually correct because the ER has a better cooling system which improves clock speeds.

I would suggest changing the title of the thread to "in the P950HR" because the "ER" performance is yet to be independantly reviewed.

Yup, on the HR this was the limiting factor - but hopefully on the ER we will see better performance (initial testing shows an improvement over the HR, but we only recently received our first sample chassis).

 

My problem is with clevo using the slower 1070 max q and limiting the tdp (which causes the card to run at a pathetic 1200mhz)

 

Idk yet, i've seen very positive reviews about the GS 65, temps are fine despite being thin, speakers are great etc. but how the durability is yet to be known, it's a completely new designed notebook, so time will tell on that, but featurewise the GS65 is a solid recommendation.

 

Still, I wonder how and when both throttles. You can have lower clocks but better overall performance. I'm also waiting for a response from Hyperbook.

 

Nope, sorry but that review is from 2017 and is the P950HR .... not the P950ER version.

 

My bad then. I can hope then

Do you know which version the ER uses . I am considering buying it in the next weeks

 

No worries.

AFAIK the ER isn't available to buy yet, we only received our sample chassis from Clevo a week or two ago I think (from memory). I don't have access to all details now as ijts a public holiday here! Will update soon.

 

Literally first bad 1 star review.
Also you're totally gonna ignore that its much thinner and has a higher clocked maxq.

 

No one can doubt that the gs65 is better machine. But you need to pay 500$+ extra for it

However,According To XMG its not certain that clevo will not use the 90w max q 1070.

 

Thanks !

Waiting for your update to pull the trigger

 

Its not even the GS65, but he literally takes people seriously who think MSI is installing bitcoin viruses on their notebooks lol.

Also yes, that's what I said, you always pay 500USD more for a reason. If you can accept inferior keyboard, screen, drivers, BIOS etc. and are only interested in performance, then clevo is your best bet.

also sidenote:
My overclocked (non modified TDP) GTX 1060 is slightly faster than the nerfed maxQ on the clevo and slighly slower than the "good" maxq, make of that what you will.

 

I doubt a gtx 1060 OC can beat a max Q 1070

The "nerfed" version gets 14500 graphics score at 3dmark firestrike and can run witcher 3 at 46 FPS at ultra

The "good" version gets 15800 at 3dmark fire strike and can run witcher 3 ultra 54 fps

The normal 1070 gets 17500 in 3dmark fire strike and can run witcher 3 ultra at 60 fps



An oc 1060 is in the 12K range in 3d mark fire strike(the 1060 will beat the bad 1070 in some games that prefer frequncy like csgo, overwatch. But in genral it will be still slower)

Note :Temp effect those results

 

https://www.3dmark.com/fs/15384583

https://www.notebookcheck.net/NVIDIA-GeForce-GTX-1070-Max-Q-GPU-Benchmarks-and-Specs.224732.0.html
avg: 14926.8

"good" MaxQs, not the nerfed Clevo one.

 

The nerfed p950hr one scores ~14600
The real 1070 max q is ~16200*

*those scores are when there is no throttling. however as most 1070 max q laptops throttle the avrage is 14900

However, how did you achieve that with a LAPTOP 1060 ????
Whats your laptop ?

Asus GX501vs*

 

It's not that unusual for 1060s to reach good scores. You just need to do proper MSI Afterburner OC, I have a GTX 1060N on a P375SM-A, the scores I had previously on my P157SM with the GTX 1060N were:
https://www.3dmark.com/3dm11/12520447

I should probably do an updated firestrike score on my P375SM-A.
The notebook you saw on the firestrike score was an Alienware 18 R1 with a GTX 1060N.

 

I've just received email from Hyperbook about their SL950 VR w/ 1070 Max-Q. Looks like FurMark does not push the GPU all the way in. In Unigine Valley it clocked at 1544 MHz

 

Amazing, thats the 90W version

Is the hyperkink sl950vr the p950er or the p950hr ?

* note. Furmark pushes the gpu very hard. Thats why almost all modern gpu downclock once they detect it .

 

HR - http://sklep.hyperbook.pl/hyperbook-sl950-vrp950hr-gtx-1070-maxq-p-26317.html

But still, the max clock should be around 1500 but it was not.

Mine 1060
Furmark:

Unigine:

 

All gpus will downclock once they detect furmark since furmark is known for destroying gpus

Thats wierd since the HR reviews and q5 all pointed to the 80w chip. Maybe it was fixed through a bios update ?

 

Which would be correct ;-)

 

Doesn't it make it a bad benchmark?

 

Its not a benchmark. Its a gpu torture test (alongside msi kombuster)

Its only good if you want to see how hot does your system get at the absolute limit

 

Can you confirm if the er uses the 90 w chip?

 

@XMG
UPDATE : it seems the er uses the nerfed 80w version

 

FurMark causes the GPU to downclock more because it's more power hungry than most loads, so the card hits its power limit at lower clocks than usual.

 

FurMark is also an unrealistic amount of stress on the GPU. It's the epitome of a torture test.

 

Now that GPUs are power monitored its the easiest way of controlling such apps rather than the cludge fixes put in drivers originally.

 

Holy crap 700mhz throttle on furrmark.

Do they have any firestrike or demanding games on the clockspeeds? Since furmark is quite the dramatic torture test.

 

the X4C from Eurocomm is ~1600 at lowest config (i5 8400, 16GB RAM) for a 1070 last time I checked

 

Get a Powerspec 1510 from Microcenter. You can get it with a gtx 1070 and a i7 7700HQ for $1300.

 

Right now the ones in that range are the GE Raider ones. There is a GS one I see that is $1699 with the kabylake and 1070 GPU

 

Max-Q specification is 80" - 90W 1070, so though it's at the lower end of the scale it is driven by 40 dBA noise level ultimately. Some companies' (I meanother companies which don't use Clevo chassis) laptops exceed this depending on the profile setting, but I'm not here to defend MAX-Q.

I completely forgot to say that the P955ER has been out for a while, basically cosmetic differences but that uses the 80W version - same as the previous HR models.

 

What I truly hate is the lack of transparency . You are losing 12% performance + your OC headroom and nobody even knows about it. For everybody its all the same card("gtx 1070 max q")

Hopefully it's fixable with a hardware programmer

 

The official Nvidia guideline for Max-Q 1070 is 80W - 90W. It's completely within specification to be at 80W.

To be fair, the performance of a Max-Q 1070is noticeably below that of a standard 1070 - weather this is transparent or not in the first place is one question. If the GPU and thus laptop is fully within Nvidia guidelines for their nomenclature, you have independant reviews and the manufacturer's specifications online to give you more of a performance indication. Also you could argue the same about non Max-Q laptops - they perform differently due to different cooling systems, thinner/thicker chassis, TDP limits and so on.

If you want to overclock then you shouldn't be looking at Max-Q anyway - performance isn't the key issue at stake with this product line and the chassis have must less headroom to clock because of the limits introduced for thickness and fan noise in order to meet the Max-Q specification.

 

A 90w max q performs at around 16.2k in 3dmark firestrike graphics stock . Compared with the normal 1070 16.5k on gl502vs and 17k from the zephryus. With a few tweaks to the voltage curve the 90w max q is the same thing as a 1070N since it has a target of 90 and a limit of 100. It reaches 1711mhz on the aero 15x STOCK in some instances FFS

The 80w version can never reach anything close to a normal gtx 1070. It gets 14.8k tops at 3dmark fire strike. you would be lucky to get it to slightly above its specified boost frequency and might end up BELOW depending on silicon lottery since both the target and limit are rigidly set at 80w

That's like selling two gtx 1070 desktop versions. One that can boost 2ghz and another that can only boost at best cases to 1.6ghz and sell them under one product without any brand ever specifying in the spec sheet. What about customers who bought a laptop 1070 max q seeing its performance in the gx501vs and then findout that the card under the same name in their gs65 performs 12% LOWER NO MATTER HOW COOL IT RUNS ?????. The only company with the decency to specify is alianware.its a scam and people need to know that they aren't even going to get the max q 1070 performance on the NP8954 and the GS65. This version is within 10% performance of a 1060N . Don't you think customers deserve to know and not get charts about the 90w version faring much better ?

And when I say overclocking I am talking about msi afterburner undervolting. The 1070 can hold 1640~1750mhz at 0.8v constant which allows you to "cheese" the powerlimit and get a performance very close to the gtx1070N. The 80w runs at below 0.8v all the time so you cant edit it.

 

If you look at a large group of Max-Q laptops in independant reviews, you can see comments like this:

"The model used in our review device has a base clock of 1101 MHz and a boost clock of up to 1544 MHz. However, the boost clock cannot be maintained under continuous load. During our benchmarks, we observed an average core clock of 1200 MHz."

Firestrike:

Aorus X5 v7 16349 (normal 1070)

Max-Q:
Zephyrus 16165(this is the exception, does it have a better cooling system or is it the only 90W in thr group?)
Aero 15X 14607
GS63VR 14394

If you look at all the game benchmarks that notebookcheck has run on 1070, 1070 Max-Q and 1060, you can roughly see that:

- 1070 runs on average 26% higher FPS than 1070 Max-Q
- 1070 Max-Q runs on average 14% higher FPS than 1060

My point is that if you take a large data set, a Max-Q at stock is a noticable level beloe the non-Max-Q. This is dependant on a number of factors, such as chassis, cooling implimentation, fan performance and so on. It can also be a result of a TDP limit, however if an average consumer looks at these figures, they can see that the very high performing models are the exception.

You can also make a very strong case that just because a GPU is limited to the higher end of the specification envelope i.e. 90W, it doesn't actually mean that it will perform better overall. Perhaps in quick benchmarks yes, as there's less time to hit the thermal limit and chassis heat build up. But you can easilly have a 90W TDP limited chip running at the same or very similar performance level to an 80W limited chip because it's limited by other factors (cooling).

 

The aero 15x sets at around 15.3k(from notebookcheck) . It would probably reach the 16k if it had daul channel ram since according to hardware unboxed that improved the aero 15x fps by 10%.

I want you to look how close the 90w version scores to a real 1070n (hell they might even delete the max q labelling on it) to know why I am pissed off. The difference between 16.2k and 16.3k is non existent. So the 1070 max q 90w IS a normal 1070N for all intents and purposes. Alianware as an example has its normal 1070N set to a 100w hard power limit just like the 90w version, do we really go around calling the 1070's in alianware laptops 1070 max-q ????

The gs63 and gs65 use the 80w version like sager so I dont understand your point.

And unlike hard tdp limits. Thermals can be improved with thermal grizzly conductonaut and Fujipoly Extreme Thermal Pads.

Edit the voltage curve and you get 17k+ at no increase in temps since your undervolting as well.

The 1070 sweet spot is 95W. The 90w version has a hard limit of a 100w so it can reach it if it has the thermal headroom. The 80w version will stay at a pathetic 1200mhz all the damn time under 99% load because its hard limit is 80w.



This is A DESKTOP CARD holding [email protected](at stock this frequncy requires around 0.95v) at 95w with an undervolt. With the 90w version you can probably hold it or something close to it without breaking your 100w hard limit.this makes your max q card a real 1070 card in a good chassis

And your reasoning for using the 80w chip is BS . At load the np8851 has a 55db reading acording to gentechpc review of the p950hr. It would probably be loauder because the i7 8750h pushes more heat than an i7 7700hq

And your max q 1070 being faster than a gtx 1060n is inflated. The 80w(the majority of 1070 max q out there) is only 10% faster than a 1060n

 

Which GPU costs more Max-Q or non-Max-Q? I mean the price difference between laptops with same specs but different GPUs.