New Clevos with Max-Q?

They understood what they were selling and what the product needed to be, and consumers understood it too.

These days, however, people expect to be able to game on a Dell XPS 13.

Few OEMs have ever produced anything that comes close to that build quality. I have yet to lay my hands on a machine that was built as well as my AW 17 R1, and even that wasn't quite up to the standards of previous AW machines.

Pretty much what I did. Bought a used (but excellent) Latitude E6520 (from back when they were still good and made to last) and built another desktop.

Come on, you know you want to.

 

I see people getting payback for wanting thin and flimsy. Battery boost will be reintroduced as the big saver, Oh'well

 

So I take it future 1950X owners here?...

Sent from my SM-G900P using Tapatalk

 

It depends on how that turns out. Too early to tell if 1950X is actually going to overclock well enough to want it. Being better and matching Intel's equivalent at stock clocks won't be a good enough reason to buy one. It will need to still be able to win when pushed to it's overclocked limits. Also depends on whether they release a GPU worthy of being paired with it. Never have liked the idea of mixing red and green parts on the same machine.

So, ultimately still waiting for the fat lady to sing. No more leaps of faith on anything sold under any label... way too risky, but especially where AMD is concerned. Measure first, buy later. I will measure using Futuremark search and HWBOT scores, not based on "professional" reviews. Ryzen 7 1800X still isn't faring all that well on the leaderboards from what I can see. Hopefully, Threadripper will change that sad paradigm.

 

Have you seen Intel's chips lately. They are an f-ing joke! Hell, base clock on the 7920X is 2.9, calling into question the 100-200mhz drop with 2 extra cores. I'm not expecting over 4GHz all cores on AMD, but with the 400mhz drop from the 10-core to the 12-core with Intel, the forced delid, the MB VRM cooling and 10-core pulling 400W, I'm less inclined to give Intel any pass this round. With 7nm vs 10nm coming, I think Intel is headed for darker days.

But, as you said, we need more info and good reviews.

Sent from my SM-G900P using Tapatalk

 

Yes, I have seen and I am not impressed. They've never been one to allow a competitor to beat them at their own game, but seems like tech everywhere is circling the drain and trending downward in terms of performance. That's why I am watching to see what happens. 4940MX was a joke as well. It sucked severely compared to 4930MX. SKL-X and KBL-X are both a joke. I won't accept crap from either one. Definitely need more time to see real results from passionate overclockers that know what they are doing (guys like us) and not marketing hype or hokey-pokey baloney from "professional" reviewers.

 

Definitely agreed! The only thing I'm waiting for is the motherboard reviews and the confirmation of TR using B2 stepping like Epyc, which cut down the inter-ccx latency by 40ns from Ryzen 7 (which also cuts down the cross-die latency to the same amount, although the memory latency is around 10ns greater than the inter-ccx latency). Here is some articles addressing the latency numbers (including Intel Mesh):

https://www.pcper.com/reviews/Proce...Core-i5/CCX-Latency-Testing-Pinging-between-t
http://www.anandtech.com/show/11544/intel-skylake-ep-vs-amd-epyc-7000-cpu-battle-of-the-decade/13
https://www.pcper.com/reviews/Proce...X-Processor-Review/Thread-Thread-Latency-and-

Both scale with faster ram, although you get slightly more benefit on AMD. But, I also need to see ram overclocking with the new chips, including 3733+ speeds without using bclk and whether they fixed the defaulting of PCIe to 2.0 if bclk is over 105. Lots left to find out!

 

Worth noting, people were doing this long before the AMD/ATI merge. I even had an AMD board with nForce chipset back in the day.

With Epyc on the way, I wouldn't be surprised if Nvidia is silently hoping it takes off as the greatest thing ever and especially in the HPC/Datacentre markets. Intel has been hamstringing GPUs with a lack of PCIE lanes for quite a while now and the new AMD chips are pushing that envelope hard. AMD chips have a larger number of PCIE lanes per cpu, lower power consumption and lower cost per board which ticks all the big boxes for anyone setting up GPU clusters.

 

Sager NP8952 (Clevo P950HR) Detailed Review & Benchmarks
nVidia GTX 1070 Max-Q 8GB


Sager NP8952 (Clevo P950HR) nVidia GTX 1070 Max-Q 8GB, 7th Gen Intel Core i7
https://www.gentechpc.com/product-p/sager-np8952.htm

 

Skip to 11:33 to see GTX 1070MQ's 3dmark score which is 12097
A normal GTX 1060 6GB scores 12960 :/
https://www.futuremark.com/hardware/gpu/NVIDIA+GeForce+GTX+1060-6GB/review
GTX 1060 3GB : 12220
https://www.futuremark.com/hardware/gpu/NVIDIA+GeForce+GTX+1060-3GB/review
Sound like you will pay full price to game on a cripple GTX 1060 3GB , that's pretty smart

 

It would be better to scroll down in your links to compare the nVIDIA GeForce GTX 1060 (Notebook) score of 11,100. Click on See full list. Or, link to https://www.futuremark.com/hardware/gpu/NVIDIA+GeForce+GTX+1060+(Notebook)/review. You are quoting nVIDIA GeForce GTX 1060 desktop scores.

 

I don't like how it's so thin.

 

Funny how people are never pleased, companies make a beefy desktop replacement, then people complain it's nice but it's too thick. Then they give you an option of a thinner laptop, and you complain that it's too thin

well, if it's too thin, get a Clevo P870KM1 and enjoy the thickness

That's why manufacturers have a variety of laptops to cater to all people's needs, if you don't like it, then it's not for you, look for another model. what a useless comment

 

Nice one

 

you are comparing a desktop GPU to a laptop GPU, nuff said

 

They are *ALL* supposed to perform like "DESKTOP" GPU's

How soon we forget...

Nvidia stuffs desktop GTX 1080, 1070, 1060 into laptops, drops the “M”
https://arstechnica.com/gadgets/2016/08/nvidia-pascal-laptop-specs-gtx-1080/
"Just under a year since Nvidia brought the full desktop version of the GTX 980 to laptops, it is beginning to put an end to cut-down laptop chips altogether.

Starting today, the desktop versions of the GTX 1080, GTX 1070, and GTX 1060—with some very slight tweaks—are inside laptops from the likes of MSI, Asus, Alienware, Lenovo, and Razer, to name but a few. They're even overclockable. Yes, if you want the very best graphics card outside of a Titan X inside something you can carry around with you to LAN parties, Nvidia has you covered."

 

supposed to, but there is a difference in the RAM size as I can see above.

PS: you've liked everyone's posts, now I don't know what you agree on and what you don't

 

Like doesn't mean I Agree with what you said necessarily... Like means I like your post.

I can Like your post, and you - not necessarily the same - and still not agree with what you are saying.

It's also an effort thing, well stated arguments and all, points for form. I can agree with what you say, but if it's a crap post I probably won't Like it.

Put in links and provide insight and actionably useful information, and you probably get a Like and a +rep! I still may not agree with your post.

The text in *my* posts explains *my* position.

 

So do the "DESKTOP" GPU's have different VRAM sizes...

Besides, the Clevo laptop Max-Q 1070 has 8GB for a supposed advantage, and it's performing worse than a 3GB desktop 1060... very sad.

 

Just said that I don't like it. Which I agree it's not for me.

 

That's right . Check DigitalFoundry's review
GTX 1070 notebook is actually on par with desktop GTX 1070 in most games

 

With all due respect, @Thousandmagister, the GTX 1070MQ is a fair bit better than the 1060. Typical 1060s achieve ~12500 in Fire Strike graphics. In that very video, at that very timestamp, the 1070 Max-Q has achieved 14711 points. That's about 2000 points better. It may not seem like much, but the P950HR only $185 more than the 1060 version, the P950HP6. Slightly less than $0.10 for each point increase. Compare to the price increase from the P650HP to the P650HS: $170, from a GTX 1060 to a GTX 1070. Prices from HIDEvolution. However, the P650HS has dimensions 385 × 271 × 28.7 mm and is 2.65 kg with the battery, while the P950HR has dimensions 380 × 249 × 18.5 mm (nearly a centimetre thinner), and is 1.9 kg massive, i.e. 700g lighter.

Whether or not the price/weight/performance trade-off is worth it, is up to the purchaser to decide. It's not as lousy a deal as notebooks with the GTX 1080 Max-Q, but still worse off than purchasing a notebook with the GTX 1070.

For reference: the GTX 860M (GM107) achieves ~4000 Fire Strike Graphics points. 2000 points more, and that's a whole other GPU: the 965M.

 

Really ?


Edit : I got what you meant but the overall score : 12097 is still below the overall score of a normal GTX 1060 3GB which is 12220
A real GTX 1070"m" is on par with the desktop version which is way faster than a GTX 1060 6GB. GTX 1070MQ is bullsh#t , let's call it "GTX 1050 Ti Boost"
Edit 2 : Nevermind then . I would opt for Acer Helio 300 (i7 7700HQ + GTX 1060 6GB) instead , it's $600 cheaper
https://www.amazon.com/Acer-Predator-Helios-GeForce-G3-571-77QK/dp/B06Y4GZS9C

 

2000 points x $1 = $2,000.

2000 points x $0.10 = $200

 

Yup...really. Since we are talking about GPU’s, look at the GRAPHICS score, to which @Ionising_Radiation referred...not the overall score.

 

Yes, really. I said Fire Strike Graphics. If you're doubting me, your image just proves my point twice over.

Look at all the tiny numbers below the big-ass 12097. See the first line, 'Graphics score'? What's the number that's next to it? 14711? That's what matters in Fire Strike, and it's why I ignore any reviewer that quotes a FS bench number without the Graphics score (NotebookCheck is a notable exception, which is why they're so damn good).

The large number that Fire Strike spits out is a composite benchmark of the two graphics tests, the single physics test (which is a CPU test) and the combined test at the end. It is pointless and statistically incorrect to compare Fire Strike benches without using the Graphics score.

Sent from my HTC 10 using Tapatalk

 

Ah, yeah, sorry, I derped out there. Will make the appropriate edit

Sent from my HTC 10 using Tapatalk

 

Hey, don't sweat it, we've all slipped a column from time to time.

Max-Q sucks for value and performance, and that's enough said

Get a *real* 1060, 1070, 1080, then there's no doubt for what you are getting!!

Unless you get a Razer Pro 1080, then its' a 1070's sickly running mate

 

Let's just call it a 1074. In a sweatsuit.


Sent from my iPhone using Tapatalk

 

Certainly way inferior than a 1070N and certainly way better than a gtx1060N but for the full 200 bucks more.

 

It's as if people have a choice of form factors and prices depending on what best suits them

 

Yes, that is correct.

The consumer is KING...if you don't like it, don't buy it.

All this wind bashing it is a terrible waste of time and energy.

 

Logical fallacy right there. Just because everyone 'slips a column' doesn't mean this mistake is any less wrong. In fact Fire Strike is notorious for being questionably wild, and illogical in the way it presents benchmarks. I benched my GTX 860M, see what we get:


Where does the software get that 4339 score from? It's sure as heck not an average of 4683, 9497 and 1835, which is 5338.3. That's about 1000 points more than 4339, or 23% more, which is not insignificant.

If one wants to scientifically and thoroughly benchmark graphics cards, let alone complete systems, Fire Strike is the worst benchmark to choose, primarily because it's a CPU+GPU composite bench and highly normalised, with who-knows-what weird statistical manipulation going on behind the scenes. The least we could do is to at least use just the Graphics score and discard the large number, which is, I say again, erroneous. It is bad statistics, full stop.

I will concede that Fire Strike is good for a ballpark estimation, a rough guideline, but if one wants to do any sort of serious comparative benching, the big score should be taken with a sackful of salt, and each score with at least a 10% error margin (meaning a score of 4000 can have an error of ±400, and a score of 20000 can have an error range of ±2000). As a scientist through and through, this error margin is way too high, and put more bluntly, quite rubbish.

A much better GPU benchmark would not spit out any sort of 'score' at all, but give us the raw data, i.e. frame-times (preferably in a CSV file for us Excel nuts) and it'd even do a bit of graphical statistical analysis by itself: mean, median and mode frame-time, standard deviation, and even draw a nice normal distribution graph. Any benchmark that spits out a single score, hence obfuscating the backstage work, has already introduced some form of inaccuracy.

I'm not advocating Max-Q (especially not the marketing associated with it), but if one wants to prove a point, there has to be no holes in their argument. If I find that the GTX 1070 Max-Q is objectively decent (which it is—the P950HR appears to be a better compromise for a lighter machine than the Asus Zephyrus, and is also much more reasonably priced), then I will say so, and I will not bluntly round up every GPU with the word 'Max-Q' and figuratively do it the Nazi way, and gas them all in the chambers.

 

You viscously eviscerated that poor guy for no reason, instead of simply correcting him, notice how he's gone now and probably won't come back.

I was trying to encourage him to come back. Nice job drilling him back into the ground.

Continuing the prosecution of your unreasonably vicious "righting" of a "wrong" was unpleasant to read.

I don't like encouraging those kind of responses, so I ignore them.

 

Viciously eviscerated? If you put it that way, fine. That's not how I began my response, here. Instead of checking his scores, he replied with a screenshot of the video, with a rather snarky 'really'. I admit I got somewhat annoyed, as evident in the tone of my next response.

I then noted his edits, which is why I chose not to respond—point made and understood, no need to press it across again.

But what I absolutely could not stand was your comment, paraphrased: 'it's alright, others make the mistake too,' and then 'whatever the case, Max-Q is rubbish,' clearly deriding a detailed and polite breakdown of the facts which I first presented.

I could say the same of you, accusing Max-Q of being an all-around worthless concept despite new facts being presented. I was willing to change my opinion, you aren't.

This slighting of fact-checking, detailed, raw, transparent statistics, and a tendency towards excessive simplification and obfuscation (and frequently, downright changing the facts to suit trends) translates to the real world, where serious decisions are made. It has translated to poor choices in electing leaders, or in making geopolitical decisions. I regret to drag politics in here, but it is what it is. Statistical analysis is seriously lacking in many of the lay person. I don't claim to be a professional statistician, but I try to follow their methods, and be as transparent and thorough as possible.

If, at any point, @Thousandmagister was offended by my posts, I apologise. I type directly, straightforwardly, and it is often mistaken for rudeness. I ought to change this. Thank you @hmscott for pointing it out. However, the facts will not change.

 

You are welcome.

@Thousandmagister , please come on back

 

Exactly bruh. They could offer f.eks the Clevo barbone series laptops with socket hardware + the Msi Barbones with exactly same lovely socket hardware. This will give people a variety of options

 

Nope
I stated my opinion , you stated your opinion . There is nothing wrong
GTX 1070MQ was not bad as I thought it would be , that's why I said never mind . But still... $1600 is too high , as you can buy Acer GTX 1060 notebook for only $1000 or the super slim form factor Asus GTX 1070 (non Max-Q) notebook with 120Hz Gsync display for $1600 . This GTX 1070MQ notebook is kinda out of place , don't you think ?

 

Here's the technical guide for 3dmark, unfortunately it is not updated and only up to firestrike extreme. See page 35 to see how the scores are calculated

 

Thin is secret code for " this is a wimpy piece of trash with a crappy cooling system and a fancy looking wrapper" where notebooks are concerned.

I'm not seeing much variety for enthusiasts at all. There is hardly anything worth looking at because of the over-utilization of anemic BGA filth. You get a whopping 2 or 3 options if you don't like the soldered wuss crap.

Vanilla ice cream comes in round containers, square containers, paper and plastic, some brands are extra expensive, and there is a wide variety of colors with fancy images and fonts to make the container attractive. Some brands of vanilla ice cream tastes better than others, but at the end of the day it's all just vanilla ice cream. The package that it ships in ends up in the garbage and has no effect on the flavor of the contents. I like the huge plastic one gallon buckets with the carry handle the best. The plastic is great for landfills since it never wears out or deteriorates over time. You can even use it for something else after the ice cream is gone. Like diamonds, plastic is forever... unless it catches on fire. And, we know it's all gonna burn in the end.

Bon Appétit.

Beauty is only skin deep, but ugly starts on the inside and goes to the bone.

 

That P950HR is actually a good bit faster than my P650RE and the temps aren't too bad. My cpu is 3-4C cooler on a FS run, probably because it's a 6700hq clocked lower and the GPU is about 5-6C cooler. Without knowing the power draw that could just be the performance differential throwing out more heat or superior cooler in the P650.

My P650RE (for reference, NBC's median 1060 Mobile GPU score is 11426 so I'm definitely not fudging this to make it look better):
Overall Score: 9775
Graphics: 11656
Physics: 9717

P950HR:
Overall: 12097
Graphics: 14711
Physics: 10745

Overall score (the big one at the top) is rather meaningless as it integrates the CPU and GPU scores in some weird way. If I could be bothered, I could probably figure it out and plot it....
The important one is the Graphics score. Since Firestrike is not CPU bottlenecked in any way (with modern cpu), the graphics score will scale pretty much 100% with GPU power available.

So the jump from 11656->14711 is about +26% which is certainly noteworthy.

As for pricing, looks OK to me. For Australian pricing:
P950HP (1060 not MQ) = $2249
P950HR (1070MQ) = $2649

So $400 more expensive (~$317 USD) for +26%.

For reference, the standard P650HP is $1899 AUD. So you already have to pay +$350 for the thinner chassis with the same GPU+CPU combo.

 

How to interpret FS overall score is in the PDF dspboys linked a few posts above

Its a weighted mean where unbalanced setups are penalised but basically starts as 75% influenced by graphics score

As in my P370EM/1070SLI's case, bad physics and combined test scores bring a >30000 graphics score down to under 18000 overall score)

As for 1070 maxq the peak bench score is next to meaningless, pascal uses available thermal headroom for boost. What you want to know before claiming its +26% across the board is how it behaves under consistent, constant load, especially since its anticipated the thinness and fan cap places an extra expectation of thermal constraint.

 

Which... still makes it meaningless as a GPU benchmark. In most general usage, a Pentium G4560 or Ryzen 3 1300 would probably not bottleneck a GTX 1060, but such CPUs would score very poorly indeed in the Fire Strike Physics test, compared to, say, a Ryzen 7 or Core i7, or even a Ryzen 5 1600 or i5-6600K.

Using the Fire Strike overall score like I've mentioned above still defeats the purpose of comparing graphics cards, even with identical CPUs, and worse still, with different CPUs. It may be a "decent" system benchmark, but certainly completely useless if you want to compare raw graphics card performance, which is what is of contention here.

I think CineBench R15 OpenGL is a more relevant benchmark, which is ray-traced. Or perhaps a Blender render.

 

You sound like you're disagreeing with points I didn't make? I agree in a graphics-only discussion, the graphics-only score is the most appropriate stat to consider. But if for whatever reason I wanted one single number as a summary of a whole-of-system gaming performance, overall is the one I'd look at. Getting the right answer depends on first asking the right question.

I do however disagree with using OGL and CUDA benchmarks as surrogates for gaming performance which is (and FWICT will remain) almost always a D3D workload.

Anyway my point is, any FS score doesn't satisfy me that in the long term that the 1070M offers a true +26% over the 1060 due to its 5-minute nature.

 

I'm that's the case you're far better off using 3D Mark 11 or Vantage. They give far a better gauge of overall system potency in the overall score. FS is fluff and doesn't tell the whole story even if you want to look at the overall score, but for strict GPU comparisons its satisfactory. That'a assuming they've sorted out the buffed or nerfed scoring on the updates @Papusan ?

 

It's sad that no one use Heaven Benchmark anymore . I love it, even now I'm still using it

 

I would think Thin and flimsy will experience more thermal throttling if you run more demanding tests.Aka more load on Cpu than the ladies bench 3DM Firestrike. Streaming will be more and more important.
The i7 Jokes who is the normal in all those JokeBook's will be pushed harder and the shared pipes/heatsink design will make Pascal run hotter with the results of lower clocks in the end. Thin and flimsy is equal Max-Crippled!! The more you tax the processor, the more will Pascal suffer.

 

That can often depend on the generation of game you are looking at and games still are mostly GPU bottlenecked.

 

Certainly a constant load test would be nice, but my FS test was also a one-off at stock levels so given the temps are fairly similar I'd expect results to trail of at a similar rate.

 

There's always going to be exceptions to the rule for gaming. I'm referring to overall system performance being measured from comparing a benchmarks overall scores.