*** Official Clevo P64xHK/Sager NP8x40 Owners' Lounge ***

I've been here two and a half years, and I finally get to start an owners' lounge for the product range of my sort.

Thanks to @Stooj for discovering that Metabox, the Australian Clevo reseller, already has listed this machine as the ' Metabox Prime P641HK'. All in all it looks like an entirely internal refresh compared to the P640RE/RF. Specs are as listed below:

I probably will not purchase this machine as I wish to move away from gaming onto more 'productive' stuff like programming and 3D designing, but it's still interesting to follow the market.

Things to note: this still doesn't have G-Sync, even though we now have a discrete GPU mode like the 15/17" models. It has a USB 3.1 Type-C port, but it isn't mentioned that it supports Thunderbolt 3. Power supply is 180 W, even though the GTX 1050 Ti supposedly draws the same power as the GTX 960M. The battery is still extremely anaemic, given that the GS43VR has a 99 Whr? battery.

All in all, it's an incremental spec-change; anyone with a current P64xRE/RF would do well to give this a pass as they'd be paying lots for a brand new machine that isn't much more powerful.

It is, however, good for someone that may come from stuff like the W110ER, or the W230ST even. It's also a good alternative to the AW13 r3 (though the graphics and battery life would definitely take a hit).

 

I think the one thing that stands out as a "reason to buy" is the P640 series have always been quite cheap (In Australia at least). They're far and away the cheapest P-series machines even going so far as to begin undercutting the N/W series.

I highly suspect the the largest reasons to buy the P640 are:
A) You want something <15" that is 2KG or lighter.
B) Your budget is only $1700 AUD (or regional equivalent, depends on relative pricing of course)

A friend of mine is looking at one for this very reason (replacing an old W350ET in fact). For that money, there's no competition.

If price isn't a factor, the new XPS15 (tiny bezels actually put it in the 14" competition), Aero14 or Aorus X3v7 are all better on paper (hell, even the Blade14 is...eek). The Aero14 and X3v7 have 94wh variants, weight less, are thinner and have the faster GTX1060. But they also cost damn near twice as much (here).

The GS43 only has a 61Wh battery in it. Clevo has always skimped on battery though. Usually to accommodate 2.5" drives which are all but abandoned by other manufacturers of "light" gaming laptops.

AFAIK the reason the P640 has only shipped with a 45Wh battery and not something 60+ is because they've retained that 2.5" HDD slot.
All of the above alternative systems are strictly 1 or 2 M.2 SSDs only and the whole palm rest is then free to be occupied with a battery.

I highly suspect this may be the last iteration of the P640 chassis as they may go to pure M.2 + large battery like the competition which would likely necessitate a new design.

 

Really interested in seeing how the 1050ti performs.. Should be around 970M at least I guess..

Sent from my LG-H850 using Tapatalk

 

That's honestly not an upgrade at all, because the P640RE already had a 970M in it.

 

Except clocks and temperatures of course..

 

Indeed. The impression I got with the original P640RE was that it was really stretching the limit on thermals in that chassis and throttled when you weren't in a cool environment.

The other thing is the 1050ti has 4GB VRAM rather than 3GB (970M) or 2GB (965M). There's a surprising amount of games which hover around the 3.5GB mark by using 2K/4K texture assets which will benefit hugely.

 

I don't know about that—according to HTWingNut's review, temperatures appeared to be plenty decent. Admittedly the 1050 Ti would supposedly give even better temps, but I don't buy it. The HK still comes with a 180 W power adapter for a reason, not a 120 W one like the W230 series if the GPU were truly less power-hungry.

This, however, is very true. 2 GB of VRAM just isn't enough in 2017; plenty of games begin to lag severely when assets begin to swap between system RAM and VRAM. Nevertheless, there have been GTX 960Ms that have released with 4 GB, such as in the Inspiron notebooks.

 

As far as straight power, the 1050Ti Notebook is marked as 70W on notebookcheck vs 81W for the 970M (according to NBC).

Even though it's a sample size of 1, NBC has a review of the GE72 with a 1050Ti.
http://www.notebookcheck.net/MSI-GE72-7RE-Apache-Pro-Notebook-Review.189684.0.html

Fortunately, they have a review of the same chassis with 6700HQ and 970M:
http://www.notebookcheck.net/MSI-GE72-6QF-Apache-Pro-Notebook-Review.153798.0.html

Of course, it's possible/probably some other stuff could've changed to affect temps, but (comparing the screenshot with Prim95 + Furmark stress-tests) the GPU max temp (on HWMonitor) is 9C lower on the 1050Ti. The 970M is thermal throttling as well (GPU load is down, frame-rate is much lower).

Concerning the power brick, even the P640RF shipped with the 180W PSU. The GM206 965M it shipped with used even less power than the 1050Ti. I would say this is more likely because they couldn't be bothered shipping a different charger around 150W mark. 120W would've been cutting it close and could break cross-model compatibility. MSI's 1050/Ti models use a 150W brick.

 

Yeap in the previous generation it didn't made a lot of sense that the P640RF had such AC adapter. It never went near that. I usually have my eye on the power measure at the plug, even the 970M was staying bellow 120W, i guess for the 970M i could see it as a "play it safe" move, but i think that a 965M or the new 1050Ti would be just fine with a 120W AC.

I really like the P640 models, specially the cleaner version "P641" without the chrome on the back.

 

It was likely influenced by the forum who was witch hunting any machine with a lower rated adaptor at the time.

 

The Chrome can spoil the clean professional look of the notebook.. Even I prefer the 1 models because of that!

 

Clevo's official spec pages are up, too.

Nearly a mirror image of the P640R series. I wonder. If they put the GTX 970M in the 640RE, why not the 1060 in an iteration of the 640RE?

 

Yeap it is weird but i can assure you they are not offering the 1060 in their catalog, so i can only assume there will be no P640 with 1060.
Nothing mentioned in the roadmap for 2017 either, so if anyone is waiting for a P640 with GTX 1060 i think you can forget about it.

 

1050 and 1050 TI where drop-ins for the 96xM (electrically compatible). 1060 (neither compatble with 96xM nor 970M) would have required a board re-design...so yeah...would have liked to see a 1060 in there myself.

 

That´s it then... i figured it would be related to that.
It´s actually sad, because small laptops with powerful GPUs are rare, and make a good selling motive, just like the desktop cpu and mxm gpu make with the 15" and 17" range.
We will have to wait for a new design for 14" laptops.

 

Yup, you mentioned that earlier.

Wait, so what is the 970M electrically compatible with?

Really would have liked to see a 1060... In terms of raw performance, anyone with even a P640RF isn't getting much of a boost, maybe 25-35%.

There aren't any of the nice extra stuff like G-sync or a higher refresh rate screen either. All in all, the small notebooks are shafted...

 

1060, 1070 & 1080 are not compatible with anything before them, only 1050 & 1050 TI.

 

Screw nVidia. Or maybe screw Clevo for being lazy. Didn't they redesign the W230s motherboard going from the 765M to the 860M (and netting nearly 2 hrs better battery, if I remember correctly)? Why not again?

 

The 970M used in the P640RE was a GM204 core which is pretty much a dead-end. Theoretically they could've used a Gen1 965M (the non-Ti that was presumably bottom-of-the-barrel GM204 binned chips) but they went out of production pretty quick as yields increased.

The tricky bit with Pascal is how wishy-washy the TDP is for them. Because the core clocks are so highly variable the TDP allowance of the Pascal chips is MUCH higher than you'd think it should be. For example, the GP106/1060 has a 120W TDP allowance on desktop, NBC lists the mobile as around 80W. 40W difference but the boost clock is only 38mhz lower

You're not wrong, but it's worth noting that 13" previously got things like the Sharp IGZO panel (which was light-years ahead of the 15" TN and cheap IPS panels at the time). Currently, 13" are still the only ones graced with OLED panels.

Since Nvidia don't have any easy and publicly accessible BGA spec, it's hard to know what's compatible and what's not.

Pascal is a rather significant die shrink which allows you to reduce the entire chip size (to a degree, eventually there's a "minimum" amount of I/O solder points required). I'm guessing the tiny 135mm2 GP107 was already on the lower limit so they just made it point compatible to save development costs for ODMs.

 

True, but my question was more of: whether or not any of the current-gen GPUs were compatible with the 970M, which Prema answered with an unfortunate no.

Indeed, well said. I think that this time around, notebooks got a much better end of the deal when it came to nVidia GPUs, precisely because they were literally the exact same chips as their desktop counterparts. Given the same computational performance, the only way that one can eke out better thermal performance is to make better bins. I strongly believe the best bins of all the chips, from the 1050 all the way to the 1080, have all found their places in notebooks, and not desktops—this may be the reason why notebook Pascal GPU TDPs are so much better than those of the desktop GPUs even with only slightly lower clocks. In fact, the 1050 Ti for notebooks is clocked even higher than in desktops, and still performs thermally better.

Of course, but I'm a gamer first, and honestly speaking—I'd rather a decent IPS panel with adaptive refresh rate technology than IGZO or OLED. One man's treasure. ¯\_(ツ)_/¯

The thing is that the raw performance of the 1050 and 1050 Ti don't net much better performance than the GPUs they're meant to replace. There's a massive gulf in performance between the 1050 Ti and even the 1060 3GB which shouldn't exist.

 

This was the impression I got but it's unlikely we'll ever know. The exception (or perhaps affirmation?) is the mobile GTX1070. The mobile 1070 actually has MORE cores unlocked which means the desktop 1070 is actually the most cut down GP104 core which is unprecedented. Similarly, the boost clock is only reduced a mere 38mhz.

The other thing, is GPU Boost throws the spec'd base/boost clock out the window. Almost all 3rd party Pascal desktop cards find themselves boosting to ~2000mhz on their own which is well above spec. So there's also the possibility that the NV specified clockspeeds are extremely conservative in order to make Pascal appear to have parity with desktop when it really doesn't with real cards. That basically leaves the TDP as the sole dictator of performance.

I suggest having a go at the AW13 OLED screen if you can. Even at 60hz the pixel response time is on another planet. It doesn't have the same visual impact as going to 120hz but it's certainly worth witnessing. OLED, 120hz + G-Sync would be the holy grail of course

From what I've seen the 1050Ti operates around the 970M level of performance which is a healthy boost when you consider it replaces the 960M. The 960M, while a great chip, is old as rocks now and isn't really that fast.

 

TB3.0 means a reference card and a custom cooled card will react differently and appropriately to their cooling, the same for notebooks. It's not some kind of cloak and daggers, it's just to optimise the chip for each scenario.

 

10% in the best cases and if you don't turn on filters. Once you turn on AA/AF forget it, the 970M can perform on the same level or a bit above a 1050Ti.

The 1050Ti replaces the 965M (1st version) and the 1060 3Gb replaces the 965M (second version). I'm saying this based on the laptops price tags.
In the previous generation we didn't have the two versions of the 965M being sold at the same time (at least intentionally).

I was considering changing my P651SE for a P641HK1, if the GPU performance was worth it. I'm confident that it's best to maintain my current laptop.

 

The 1050ti has better memory compression tech and much higher base clocks, the same goes for the ROPs, it has few but running at much higher clocks.

 

Sure, but that does not make my statement false

 

Those things should allow it to cope with AA/AF at the same rate as the 970M.

 

The 1050Ti definitely replaces the 960M skew as the brackets are determined by TDP, not price tag. No variant of the 1060 can run under the same TDP limitations available to the 965M chassis. The 965M in both variants was all but ignored by Gigabyte and Clevo for that reason as many of the chassis targeted at the GM107 TDP weren't well enough equipped to deal with the 965M.

As for pricing you have to consider that the first 4-6 months of each new generation brings a $100-200 premium. This functions as the "new tech tax" while also allowing retailers to run out older models at the lower price without taking too much of a hit. Eventually they'll settle into the same brackets.


EDIT: On a side-note, Clevo is now listed it as the P641HK1. The old P641HK (no number 1) page isn't available anymore so I can't be sure of exactly what changed, but what seems to stick out to me is:

• HM175 Chipset now listed
• i5-7300HQ option listed
• DDR4-2400mhz listed as default (up from 2133mhz)
• 1x Mini-DP 1.3 AND 1x Mini-DP 1.2. As opposed to 1x Mini-DP-1.2 (Mini-DP1.3 should allow 4K@120hz)
• All USB 3.0 ports now listed as USB3.1 Gen1 (likely due to the chipset change or full spec being known)
• Battery life went down? Now listed as 290 Minutes. Could've sworn it was 300 or 330 previously.
• Key Features lists XTU overclocking, but no 7820HK processor listed. I'm guessing this is a mistake as there's no indication the standard HQ's will get any kind of OC.

 

Fair enough, but notebooks with the GTX 1050 Ti (and the i7-7700HQ) upon release are already sub-$1000. Upon release, notebooks with the Maxwell GTX 860M and i7-47xxMQ/HQ were $1300-1500. I understand your point about TDP, but the general consumer (which is what I was more than two years ago when I first purchased my W230SS) looks at prices first, and not TDP. Going by that logic, the notebook GTX 1060 sits squarely where the 860M was. Now, if you're talking about the 840M/850M, then they were positioned as the budget GPUs. Now, nVidia has pulled a very fast one on consumers by effectively shifting prices (and brackets) up where it matters.

Furthermore, if we're really talking about TDPs and bins here, the 1060 actually replaces the 965M, going from GM206 to GP106. The corresponding performance increase is more than 200%. On the other hand, the 1050 is really a joke (hardly 30% faster than the various GM107 GPUs), and the 1050 Ti, honestly speaking, could have been more powerful than it currently is. It doesn't even hit GTX 970 levels of performance, which is a little sad.

There's still a massive performance gap between the 1050 Ti and the 1060, and this is easily seen in Fire Strike scores: the 1050 Ti scores roughly between 7000-8000, while the 1060 scores typically 11000-12000. That's a solid 4000 point difference, and it would have been better if the (unannounced) 1040 performed at 960M levels, the 1050 at 1050 Ti levels, and the 1050 Ti at 9000-10000 levels. In some cases, the GTX 1060 is nearly twice as fast as the 1050 Ti, and that is an unfunny joke. Even the 970M wasn't that much faster than the 960M (around 50% faster), and the 860M was such a good overclocker that it easily reached 870M stock performance.

I hardly need mention that there's been an overall shift to using smaller chips for consumer applications. Big Pascal, from what used to be enthusiast-level chips (GTX 580 contained a GF110), have been relegated to the ultra-expensive Titan category, or even to the supercomputer market, while the top-notch GPUs like the GTX 1080 already begin with the pared-down GP104. The 560 Ti had a GF114 chip; the GTX 1060 now has a GP106 chip.

Interesting. I was hoping for G-sync, but no dice.

 

I'm mostly going off the 900M line-up as far as relative performance/TDP differences. The 860M/850M was an aberration at the time since it was newer Maxwell chips in the Kepler line-up. It literally created a whole new gaming laptop bracket due to it's 50W TDP which could actually play games. A similar GK107 50W chip was so slow in comparison.

As for pricing, absolute figures are kind of useless as it doesn't take into consideration things like inflation and more importantly, currency strength. For example, the US dollar (I'm assuming this is most useful for people here?) was literally at it's weakest against the CNY (Chinese Yuan) in 2014 and is currently at it's highest in nearly 10 years. The USD simply had less buying power at the time. That alone could be that pricing difference. Given most computing gear comes out of China that exchange is particularly important. I imagine it's easy to forget in the USA, but for the rest of the world the status of the USD has significant visible impact on technology prices.

e.g. Here in AUS, prices were actually the opposite in 2014 as our dollar was strong at the time. My W230SS was $1359 AUD in mid-late 2014 and a current N850HK (1050Ti) is $1679 AUD.

GM206 was a blip in the ocean as far as mobile skews so I really don't see how the GP106 is a replacement for it at all. GM206 cores didn't appear until nearly a year after GM204 was released.
The reality is, almost nothing had the GM206 core in it. A few Gigabyte machines and some Clevos. For the most part, when people were buying laptops the brackets were 960M or 970M.

Gotta keep in mind, while 4000 points sounds like a lot, relatively speaking it's nothing like the gap that was present between the 960M and the 970M. While the 965M technically filled that gap, very few machines actually carried it and it only existed (in GM206 form) for less than half the Maxwell life cycle.

Going by the NBC median figures (Firestrike):
1050 Ti: 7913
1060: 11506
About 45% increase.

960M: 4313
970M: 7460
About 72% increase.

For curiosity, the 950M scored 3174.5
1050: 5784
About 82% faster.
960M -> 1050 is 34%.

It certainly puzzles me as to why the Pascal chips seem to be less efficient the lower they go (I'm still a firm believer that all the listed TDPs for mobile Pascal are hugely inflated for some reason. The GP107 cores are even on the slightly smaller 14nm process (GP106/GP104/GP100 are all on 16nm).

That's assuming Nvidia don't just release a 1080Ti in the same vain as the 980Ti (which I think is highly probably). It you look at the release dates, it's highly likely it'll just arrive in June in the same way the 980Ti did. Only this time they're actually putting a longer gap in there so they can actually sell their Titan X cards instead of undercut them a mere 3 months later.

Similarly, that gives them more time to ride the "notebook GPUs are at parity with desktop GPUs" train before a potential mainstream 1080ti rains all over that parade.

 

Different processes and wafers can make a huge difference to how the chips react to voltages and clocks.

Lower end chips are usually made on more bulk processes to help bring costs down and 14nm vs 16nm is literally meaningless when trying to compare to processes as it's the real details of the process that dictate the actual transistor size.

 

And here I was thinking it was 1050Ti max due to cooling limits since it's hotter than 970M and they wanted a small 14" footprint. Shows what I know I guess haha.

 

AFAIK The GP107 chips are made by Samsung while the 16nm GP106+ are made at TSMC. Both are FinFET and the core layouts are fairly similar.

I'm really not sure what's up with the TDP numbers for the 1050Ti (or Pascal for that matter). Notebookcheck and Wiki both list 75W, except desktop card reviews show even stock overclocked cards drawing 70W total and including big coolers (which are about 5W on their own). That's in stress tests as well, gaming floats around 55-60W.

The only possibility I can think of is that the shrink to 14/16nm has actually created a situation where the heat generated is just harder to pull away (because it's more dense?).

 

Pascal's chips are indeed extremely hard to cool due to the thermal density. They're worse than Fermi, by a long shot. The 250W TDP vapor chamber cooler on the 580 (which regularly pulled 250W+) was used, to worse effect, on the 180W-rated 1080 Scammer's Edition, which, without adjusting the power limits to max, was only allowed 165W. The same 250W cooler was also used on the Titan X Maxwell, also to better effect than on the 1080.

Let that sink in for the heat.

As for the P640HK1, I believed that 1050TiN was used instead of 1060N because the machine was indeed smaller (it can't even hold one 9mm 2.5" HDD) but this was proven false by Prema, as it's basically a cop-out due to electrical compatibility with the 970M. Basically, Clevo cannot sell 900M anymore, because it isn't sold to them anymore (probably, not sure) and they didn't feel like spending R&D into redesigning the P640RE into a P640HP6.

I know us gamers aren't their primary audience... but I wish they'd be slightly less lazy. The P6xxRx maxwell models were overcompensated on the GPU cooling significantly (so much so that the CPU cooling was considered bad and the GPU cooling considered overpowered for the cool maxwell cards) but when pascal launched, this proved to be a blessing for them since they did not need to modify much in cooling to achieve the difficult task of cooling a 1070N in a thin model. So that worked out fine. Their P7 and P8 models though, were barely changed and honestly need significant after-market work to handle the pascal heat. They should have just run a redesign, they knew it was coming for ages before we did.

Well, if MSI starts killing them with LGA models through Eurocom maybe they might start doing things yeah

 

can someone recomend me ram for this notebook? what's the difference in performance in higher mhz ram? tkx

 

The performance impact is pretty small so some tighter timing 2400mhz ram does most of the high performance work.

 

deleted

 

so I'm having a problem with this notebook. I've installed a NVMe samsung 860 EVO drive and it gets so hot during load +100C that the notebook shuts down. is it bc of the drive? faulty cooling system? I've noticed there isn't many holes on the bottom part over the fan like on the other side. is this normal? how to solve?

 

When you say +100ºC are you talking about the SSD?
If it´s the SSD you can add thermal pads for it.

Now about the shutdowns usually the SSD would not cause this, what are your CPU and GPU temps?
Also when is the notebook shutting down? During games for example?

 

it's only during load during games etc and after an hour or so. cpu and gpu temps are just fine. then the notebook reboots and tries to boot again and it reboots again and again. I've to turn off and wait a couple minutes or so for it to cool. then it boots normally without any issue whatsoever.
doesn't happen anything without being in load. seems like a thermal issue caused by the nvme SSD. I'm monitoring temps with hwmonitor64. I also have a 2.5 SSD but it remains cool.
normal temps for the nvme ssd are @70-80C without being in load. doesn't seem right. it's a brand new one. maybe it's faulty? googling around it seems it runs really hot. I'm not 100% sure if it's the SSD. maybe it's software but I've a clean install. it doesn't seem the problem. there's no bdso. there's no issues.
if it was the gpu or cpu when it rebooted it wouldn't keep rebooting. I've experienced that issues with other notebooks. all things point to the nvme ssd.
It's a brand new notebook and I've noticed that the back part of it is not the same on both sides and a fan is obstructed without many holes to ventilate. not sure if that's normal.
can someone show me a pic of the back part?

 

Did you apply the thermal pad?

 

what thermal pad? didn't come with any :X is it needed for nvme ssd's? never saw something like it.

anyway, is the back of the notebook like that? or does it have more holes on the left side? see picture.

 

OK now this is going to sound weird, but please trust me.
Please email your re-seller, explain your problem, send them your serial number, and tell them to take actions.
We had a similar problem which does have a "complicated" solution. If your re-seller got a rep in this forum please ask them to contact me and i will help them solve that.

 

what complicated solution?
and the back part is supposed to be like that or not?
it was from a private seller. it's used.

 

but what about the back part?

 

Yeap that´s how it is.

 

yesterday I heard an issue like this by an italian user

 

Hi,

I've got a P640HK1 with exactly the same problem as __-_-_-__ (laptop rebooting, often several times, after spending a certain time on more demanding games), and the problem persisted even after sending it to RMA, even though the reseller told me the problem was fixed. CPU and GPU temps were below throttling values, so no reason to shutdown so suddenly, and even though I never thought it could be the M.2 SSD, I play games from an external hard drive and that did not fix the issue. Suspecting RAM problems, I bought another 8 Gb piece of HyperX DDR4 RAM, which did not fix it either, but at least now I have 16 Gb. Also tried with a 180 W AC adapter instead of the given 120 W one, and the problem persisted.
Therefore, it might well be a software issue, however I cannot pinpoint which program could be causing this. If anyone who has had this same issue is still active on this thread or forum, what has fixed it for you?
Also, not sure if related, but since I cannot find the latest vanilla Clevo bios (1.05.06 I believe) for this laptop, has anyone got a working link to it, or do you know whether installing a bios from a reseller selling a renamed model of the same laptop would be safe? (I found a reseller called dreammachines that had the latest bios but their laptop is a rebranded P640HK, so while this is the only option I found, I'm hesitating to flash it).

Thank you very much for your help, and have a nice day.

 

Have you checked your secondary temps like PCH?

 

I cannot seem to find a software that would display PCH diode temps for this laptop (I've tried HWiNFO, AIDA64 and HWMonitor), is there any way I could check this temp in particular?

 

the problem came back so it's not software it's hardware. also, I've upgrade to the latest bios 1.05.07 and the problem still persists. so it's not bios related. all temps are good. it's not temperature related.