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

Max q is more expensive across all brands but sager. But remember not all max q are the same some of them are on par with the normal version (the 70w 1060MQ or the 90w 1070MQ) while others are gimped trash(the 60w 1060MQ or the 80w 1070MAXQ)

 

So in other words you pay more for already underclocked GPU, but it's not alright to get it underclocked even more but still in specification, right? Is there any test between underclocked non-Max-Q and Max-Q GPUs?

 

To be honest I'm not really interested in single 3DMark scores which compare two individual laptops, that's why I spent a little time averaging scores from multiple games and benchmarks in order to give the true picture across the whole market. It's still easy to see how different models perform from different manufacturers, all the information is there for all to see in reviews. Max-Q doesn't guarantee a performance level, it provides a platform of chassis depth, sound level and so on.

"The gs63 and gs65 use the 80w version like sager so I dont understand your point." - My point is that there is a broad range of performance in the Max-Q umbrella, but that is the Max-Q specification.

My reasoning for using the 80W chip is not BS. It's not my reasoning, it's the manufacturers reasoning.

From the notebookcheck Eurocom Q5 (P950HR review):

"System Noise
Max-Q targets a fan noise of 40 dB(A) when sitting 25 cm away from the front of the notebook. Both our tested Zephyrus and Omen 15 units have been shown to meet this very specific standard. The Q5, however, can be as loud as 50 dB(A) when running Witcher 3 on the preset Automatic fan mode. In order to benchmark the Q5 more fairly against other Max-Q notebooks, we've dialed back the fan to a manual 55 percent speed setting that we've recorded to be about the same as a Zephyrus when subjected to similar gaming loads. Thus, all our benchmarks and data on this page reflect the 55 percent speed setting unless noted otherwise."

"September 6, 2017 Update: Eurocom will update the firmware for faster GPU performance whilst addressing fan noise concerns in the next few days. The 55 percent fan speed mode is expected to be the new Automatic default fan mode"



I'm not sure if you're angry at laptop manufacturers for using different specs which sit exactly in the Nvidia product envelope, or if you're angry at Nvidia for giving Max-Q a TDP range and not a single TDP value. Either way, as I mentioned before, it's 100% clear for everyone to see any performance differences in load and gaming usage cases. Just stating 80W or 90W on a laptop specification page doesn't give anywhere near the whole picture.

Call then gimped if you want, but I'll say again - Nvidia have a TDP range for Max-Q, min and max. If the product is 80W to 90W then different manufacturers will choose different options.

You don't pay more for the GPU, but you might pay more for the laptop that it's installed in. Any laptop you see that is higher priced with Max-Q than a non-Max-Q is driven by the product/manufacturer of the laptop - it's not driven by the GPU version itself.

 

1070 "normal "16.3k , runs at around 1680mhz
1070 max q 90w 16.1k , runs at around 1550mhz
1070 max q 80w 14.3k , runs at 1200mhz

Compare the gap between the 2 max q and the gap between the 90w max q and the normal card. Look at how much performance your losing and no one even knows about it. If people knew that the gtx 1070 in the aero 15x is this close to a real card they would skip the msi gs65 in a heartbeat. But gigabyte decision to only include single channel memory allowed the 80w trash max q to look good.(even when crippled with single channel memory the 90w card is still noticeably faster)

The 90W max q IS A NORMAL GTX 1070 FOR ALL INTENTS AND PURPOSES BECAUSE IT CAN RUN AT 100W and performs the same as the 1070n in lower cooled model. The other one is miles behind both cards and you cant know unless you spent weaks asking about it.

 

I believe the problem here is misinformation and/or bad review. The screenshot I've posted earlier shows that 1070 MaxQ in P950HR (hyperbook) clocks 1544 MHz in Unigine and 1200 MHz in Furmark. The numbers look kinda familiar, don't they?
Looks like GPU-Z info about clocks is not reliable source of information. Look:

 

This 3dmark performance transfers to real world performance. The witcher 3 on the 90w is 53fps vs 55fps on a 1070N laptop(asus gl502vs) compared to a measly 46fps on the 80w version. This difference will explode into tens of frames in games that love frequncy like csgo and overwatch

I am not angry at sager, especially since there device is the most reasonably priced out there.I am angry at the idea of the TDP umbrella and consumers never knowing about it . Everyone and there mothers know that max q is slower than normal cards, but no one knows that the gap between individual max q chips is ALOT LARGER.

Moreover, this doesnt mean that the 1070MQ 80W isn't objectively a bad card, it is. The 1070 maximum efficiency point is at 95W . By being 15w below that 1070MQ 80W is neither efficient nor powerful.but consumers deserve to know that the 80w shortcomings when they buy it. It's barley 10% faster than a 1060N and can't be either undervolted or overclocked , limiting it's performance further.

And to add further insult to injury, the 1070MQ 90W can push to 100W with thermal headroom at stock settings, the 80W can't push a single W beyound.(that's due to the TDP target and limit)

And btw, the aero 15x drifts past this 40db "goal" by nivida and reaches 60db on max fan. Why can't sager do the same ?

 

I didnt bring this from GPU Z. I brought it because I was astonished by how much stronger the 1070MQ in the asus compared with how weak it looked in the msi GS65.

When the 80w 1070MQ is on 1544mhz the 90W is at 1730mhz and the normal card is at 1850. And the 1070MQ 80W can't sustain this BTW. its clocks acording to multiple sorces :
Jarrods tech : 1250mhz average.his aero 15 x (90w chip)average was 1520mhz btw.
Polish review : 1200mhz average
Notebookcheck : 1316~1250mhz
The quote XMG provided : 1200mhz

Peak frequencies dont matter(although the 90w still has much higher peaks). The chip at the maximum efficiency point and that eats 20w more wattage will of course destroy the one that is wattage deprived

 

PClab reviewed Hyperbook SL950VR. I've posted screens from SL950VR too. Would the put two different chips in the same model? I think it's caused by drivers and/or power management settings.

 

Nope, the cards tdp limit isnt something you can tweak. Look at Jarrod video about the aero 15x and you can see it held 1720MHZ for a good 20s despite being at 90c in GTA V . But thats "burst frequncy" its not going to stay there for long just like the 80w 1500mhz clock.

The biggest drawback is that the 80w cant hold 0.8v(nvidia undervolt minimum). Thats CRUCIAL if you undervolt. A 1070 can hold as high as 1740~1680mhz at this voltage(without passing the 100W TDP) so in a single tweak you eliminated virtually all the performance gap and your card will run at the median of the normal gtx 1070. And even if you don't undervolt the 90w is just as strong as some normal 1070N out there . So technically speaking it has almost the same performance as 1070N while the 80w is miles behind and thats HUGE

However, I PRAY that there is a clevo out there with the 90w card.I wouldn't be forced to work an extra month to extend my budget to the aero 15x(student).
But gentechpc,XMG, and xoticpc plus almost all reviews confirm the 80w card, so my hope is crushed

 

It seems that you've ignored first two sentences.

 

From pclab review



The other 1070MQ can reach a 100w while this one is limited to 80w . Of cource thats going to gut performance. Maybe hyperlink updated the card in a vbios update which is why are seeing a 1544mhz clock in your screenshot(completely possible). But I am not going to take this screenshot over the word of Sager,gentechpc,xoticpc, XMG, and eurocom

 

And this is the same model like in the review.

 

Maybe They pumped the tdp limit in a vbios update while keeping the target the same. Since the aero 15x max gpu power consumption is 96w not 89w

Its probably looks like this
Card_target _limit
1070MQ 80w_ 80w_ 80w
1070MQ 80w updated_ 80w_ 90w
1070MQ 90W_ 90w_ 100w

Or the card is simply bursting.which is the most probable explanation since you can see in the graph the card held that frequncy at no more then the last few seconds of the benchmark. And from hwinfo you can see the average clock was 800mhz so its safe to assume the benchmark ran at 1200~1316mhz most of the time so thats in line with the notebookcheck results

 

You guys are overthinking this...
Max-Q systems are artificially crippled. ODMs have to send test results to NVIDIA to get the license. If the system is too loud and/or hot under a specific load for a specific time, they have to cripple them further or won't get the certificate.
If the 15x really reaches 60db, Gigabyte played them and they will have to cripple it further or NVIDIA would have to revoke the Max-Q license, simple as that.

Any of these systems could beat the rest in performance with the proper firmware which doesn't disable CPU Turbo and thermal throttles the GPU below actual safety limits.

 

Hyperbook*. Looks like they do.

 

A comment from the legend himself, I am flattered

Its not that complicated. You can get away with 60db as long as the normal fan profile is close to 40db. While the overboost one is 60db.

And no our entire argument is that there are some 1070MQ with higher tdp limits that can destroy other 1070MQ cards. The difference between 14.3k and 16.2k at stock in 3dmark firestrike isnt small.a 100w 1070mq is good enough for me since I can squeeze a 1750mhz clock out of it, an 80w chip that can only hold 1250mhz belongs in a trash bag.

I am a customer trying to min max the performance I get in a small package, not settle for the 80w version that is neither efficient nor powerful.

 

Like I said, if you don't care about noise and temp limits enforced for Max-Q then the sky is the limit with all them.
Just buy the system of your choice and De-Max-Q it, but don't make a purchase decision based on how they are configured by the factory. Otherwise you would spend many hundreds just on firmware...

Also don't look at the CPU-Z clocks, that's all marketing as well. Actual Pascal clocks depend on ASIC, so it'll be a lottery either way.

 

You can't un-max Q a card. How are you supposed to change the TDP ? You're stuck at the factory TDP.

 

Can you please tell me how. The old vbios maxwell flashing no longer works at pasca and by suggesting there is an alternative you just boiled the blood in my viens.

 

Can these Max-Q boards handle TDP modding up to Max-P or higher levels? Nearly all Max-Q GPUs are BGA (Eurocom/GeCube has a 90W 1070 Max-Q in MXM 3.0b), so if they go boom it would be an expensive repair.

 

You would be a fool to do that(probably wont explode since this same chip is running at 115w in normal laptops). I want to get a CRUCIAL head room to 100w. from there undervolt and overclock and voala. You get a 1070n that is much more powerful and efficient. You would get 1750mhz at 0.8v instead of at 0.95 thats a -150mv undervolt


The 90w chip can reach 100w so i Wont need to mod it. But the 80w trash edition wont so it can't be under volted unless i mod it somehow

 

Why would it be foolish to seek desktop level performance? GTX 1070 Max-P are almost all 115W (I've heard of a few that are 125W like the newer vBIOS on the Clevo -TM models). Desktop 1070 FE is 150-170W. 115W is not enough to achieve desktop stock clocks in demanding loads, even after undervolting.

 

Just look at the Hyperbook review or XMGs own P950ER vs P955ER stock for the obvious...like I said they have been crippled further after the hardware was final just to pass certification...

And no I don't suggest to use MXM vBIOS or pull out the programmer!

 

A desktop 1070 stock with a memory oc undervolted consumes a 100w...........
95w is the sweet spot. After that you will have to feed it much more power.

And it would be foolish not because you won't be able to pull this off. But because by going 115w you would gain 2% performance for an additional 15w heat. You could try to undervolt a hold 1911mhz at 0.9v in 115w mode . But not sure if the 1070 wont overheat at 115w

 

So in the end I would have to make my purchase based on factory settings since changing tdp is too dangerous.

 

I'm just wondering as to why not just buy a notebook with a non-Max-Q 1070?

 

1 not looking to buy tanks

2 the 90w 1070mq is almost the same thing as a 1070n

 

The PA70HS-G is a tank? @ahmad hendeh

 

Its a 17.3' laptop. It would Weigh atleast 3kg

 

No undervolted 1070 is boosting to 1800+ using only 100W under a demanding 100% load e.g. running a AAA game with SSAA. It's not happening. It needs about 125W minimum.

Also no 1070 is undervolting to 1911MHz at 0.9V stable unless you won the silicon lottery.

 

No and that's whole point.
The TDP was only downgraded to get certified, not because the system can't handle it. XMG themselves received stock with old and new vBIOS.

Flash the old vBIOS, loose certification and be happy to have saved 700 bucks...

 

I said 1750 not 1800+ . And no a quarter of people with the gtx 1070 is the asus gl502vs achieved 1911 at 0.925v . So it's not that rare. Even if you can't achieve it go 1800@ 0.8625~0.875v. And remember the 1070N has a boost in cuda cores so at 1750mhz you are on par with FE editions running at 1911mhz. Which should satisfy anyone if he is achieving that with a laptop that has 0.7' thickness

 

925mV is 2 voltage points higher than 900mV, and that is still uncommon as you said.

 

I am going on the back of my head so those are estimates. But on youtube yo can find a guy with a gtx 1070 who
1 undervolted to 0.8v
2 locked the card at 1750mhz
3 OVERCLOCKED MEMORY by 350mhz
And still achieved a power consumption below 100w. Of course remember the cuda cores increase. With 1.9ghz you are on par with a desktop 2.05ghz . No need to go that far on thin laptops .

Go to what ever frequncy you can hold at 0.825v and you would get desktop 1070 levels of performance

Interestingly this entire discussion is irrelevant for the 80w 1070mq since it runs at below 0.8v almost all the time

 

Plus part of me is slightly wary to trust exact output readings from a notebook vrm.

 

It worked wonders for gl502vs owners trying to fix the catastrophe that was their laptop's cooling system. And worked with 1060 max q. Why not the 1070mq ?

 

By trust I mean to be making precise comparisons between needed voltages for different models.

 

Yes it won't be precise as I said but you're differently going to get a higher frequncy compared to stock since nividia overvolted their cards to prevent boost 3from bricking cards

 

Where can i find the old vbios.I am buying from gentechpc not xmg so that might complicate things.

 

No worries it'll pop up, just give it some time...

 

Hopefully you're right.

 

Yes it does. You must be a whelp of a person. I find it quite fine for carrying about. Very slim for what it does.

 

Its just that the 90w MQ is the right compromise for me. 1070n performance at recoed thickness and weight. I trained my self to carry my 14lb school bag for as much as possible in the last 3 years so I dont think its a question of if I can . its just that I want a light machine with a small form factor. Maybe if it had a 15.6 version i would have bought it

 

You should ask those people with really low undervolts to check their benchmark scores and make sure they're not actually getting worse performance than stock. When you go too low on the voltage, sometimes Pascal doesn't crash, it just lowers your performance by a lot.

Look up the YouTube video titled "The Pascal Problem" by Actually Hardcore Overclocking/Buildzoid if you want see an example.

 

I am not looking to go greedy with voltages. As long as you dont do that your fine. Tens of people have been undervolting the gl502vs and I am certain if a performance lose was the norm people would have noticed. Its just fools trying to run 1900mhz which isnt going to work properly on anything lower than 0.925v if your lucky. Moreover, the card runs at around 1520mhz stock and is close to a 1070N.if you want to go conservative set 0.8v at 1600mhz, I am certain no Pascal card will fail here no matter how unlucky you are with silicon lottery. And for boost 3 to work nividia had to use overvolted values for every frequncy step or the badly binned cards will carsh.

 

It's all in jest but I am tired of this kind of stuff. Everyone has different use cases for their laptops. I personally choose slim and light because I already carry one other laptop with me for business trips. I tried doing the 5.5lb+ thing and yeah the performance is of course, better - but it's worth it for me to get the lightest, best gaming laptop for more $$$.

Another fun thing: Sager/Clevo/etc. "off brand" laptops are great for performance but horrible for resale. The general public does not know them nearly as well as ASUS, MSI, Razer, Alienware, etc.

Just looked up the PA70HS-G - 7.5lbs - there's just no comparison when the thin and lights are all around 4.5lbs. Huge weight diff when portability is an important factor...

 

So... I am currently tempting in purchase a P955er, in my city I can get a P955er with 8750H and 1070 max-q around 800~1000cad cheaper than other models (gs65, etc., after tax). My question is how much impact does this slower 1070max-q actually have during gaming? Is the extra price justifiable for faster version of 1070max-q?

Thanks

 

(delete)

 

To know what you're losing I made an excel for the average of 3d mark fire strike when overclocked

16 aero 15x had an average graphics score of 16112

38 razer 2018 had an average graphics score of 16396 with 4 of them reaching 17k

15 Sager P95XER had an average of 14883 with highest score being 15.8k

It's 11% slower on average. The 90w is 5% behind the full fat 1070 while the 80w is 14%(this is assuming you overclocked the 1070MQ)

So assume if your bf1 fps with 80w is 100 the 90w is ~130

This should give you a picture to what expect when they are overclocked

Try looking into the razer blade 2018 from microsoft store. With student discount you can get it at 2070$ and that thing blows the sager out of the water

Microsoft support told me they will add the MS65(1070MQ) to their store soon. The 1060 version is 1600$ with student discount. So expect the GS65 for 1800~1890$ at most

Beware the MSI uses the 80w card like the sager.

I would reccomend the aero 15x, you can find it in physical microsoft stores for 2070$ with student discount


1 better cooling
2 better battery life
3 beter gaming performance
4 better screen
5 better software
6 two years of international warrenty
7 180w PSU vs 150W PSU