Clevo + Coffee Lake: Status?

there will be a clevo with a ryzen cpu at IFA2017?

 

I will never talk about machines that may or may not be released, that should be clear by now

 

I know but I am really curious about this IFA. May be some surprises

 

You can't get it more equal than this. As yoo should know... Can't put BGA in an LGA socket for equivalent, ideally identical, platforms if you want good data Your own words
Edit. Btw seen this?

 

Maybe I had not finished editing my post before you read it and responded based on the context of the unfinished post. These comparisons are under the same conditions. The BGA best isn't capable of matching the LGA best because they are inferior products. This applies to both stock and overclocked. Clock-for-clock (meaning if you overclock the BGA to match stock LGA) they are still inferior. If you think the people competing on HWBOT.org are not giving it their best shot with their BGA turdbooks, you're simply mistaken. You'll see @iunlock results on the right side on one of those links and I know he always gives it his best. He is one of those people that always gives it his best shot. Can't hate that. If the BGA CPUs were as capable the scores would be equal or greater, but they're not as capable. Anyone that believes they are is sadly misinformed.

The difference between us is you are defending an inferior product and trying to make the argument that there is an unfair comparison. All I am doing is showing numbers for other people to look at and demonstrating which product is stronger and more capable. It doesn't need to be qualified. The numbers are what the numbers are. No point in reading any more into it than that. I do not, and never have, expected an inferior product to perform the same as a superior product. That would be pretty silly.

So here we go... No handicaps, no lame excuses, no hocus-pocus formulas or smoke and mirrors... just numbers. Disposable versus non-disposable. Let those with discernment discern.

 

i'd say thats more of a software optimization issue. take CB15 for exmaple when a sw takes more advantage of the cpu performance number gets better! though imho 50% more core will benefit maybe 50% in multithreaded in software like firefox/chrome as they are well optimized, i care more about ST performance thats what make computer so snappy! that and optane ssd.

if what eurocom said is true with upcoming 17" version of tornado, im more eager to wait for that because knowing MSI has better cooling. trade off is if theres no good bios support from prema. however if eurocom offers it, there might be an unlocked bios! never konw.

for the first paragraph, couldnt agree more!! as for the 2nd part.. i'd agree stupidity is not a sin but people not striving to become better as an individual is a sin. todays society we already have mass that just lazes around, following trends/hearsay, and push their ideology unto minority. issue is since they are stupid, they donno they are stupid and usually its always others fault rather than their own.

spend the time, learn things, improve oneself is the way to go.

it seems like we are going in circle, because many times @Papusan already showed that 6700k or 7700k uses less power, voltage, produces less heat than mobile chip where are both air cooled, at the same frequency.

now you might say, mobile cpu in a laptop has different cooling capability than a desktop cpu in a clevo laptop. but its no brainer that they are same architecture regardless mobile or desktop, they draw similar power usage under load. so to truely know which is better, all you need to compare is the voltage used at different clock speed, and desktop LGA chip turns out to be quite a lot better, this is fact.

here is another fact you may not know, you think CPU can all overclock under high frequency as long as cooling allows it, which is NOT the case. plenty of time it is true to an extent, a mobile cpu can prob hit 4.7ghz instead of 4.6ghz and use less voltage had cooling is better, but that means it can only hit 4.7 max and it'll never be able to hit 4.8.

a mobile cpu like 6820HK average around 4.4ghz where as desktop 6700K average around 4.7ghz closer at similar voltage. the quality of CPU and silicon is SIMPLY DIFFERENT, face it, intel is giving mobile cpu crappier silicon.

 

from basic understanding, your thought process is correct and thats where you are wrong. say two computer with different ram spec, different storage, only similar in CPU like a mobile and desktop version running at the same frequency. CB15 benchamark doesnt use storage nor ram, its all mostly on cpu core, uncore, cache, etc, IMC is barely touched.

by doing so, the two setup will get similar score, however voltage/power draw from the CPU will be different between the two, which clearly shows which CPU is superior, in this case the LGA desktop cpu. we target problem just like you do, process of elimination by understanding the software and hardware. we have gone a further part of investigation than you have.

 

Well, as much as I love the Tornado F5/16L-G-1080... and, what's not to love about the most powerful 15-inch notebook known to man? ...I have to say that I am not impressed with it's stock cooling, or the amount of cancer MSI embedded in it's firmware. @Prema bent over backwards trying to exorcise the throttling demons and it still has some nasty crud lurking somewhere in the shadows. I believe the firmware is fixed and all that is left for now is hardware limitations. So, I'm not all that eager to give props to MSI before the fat lady has a chance to sing. But, you are correct about it being a great product. It's definitely greater than any other 15-inch option. Unless HIDevolution offers the 17-inch version and he does it as a special favor for a wonderful Prema Parther Shop, it is unlikely that @Prema will support it with a firmware mod because it is not a Clevo. The cooling system in the P870 is truly outstanding if you don't get one with poorly fitting heat sinks. The unified heat sinks in the lesser Clevos are inferior to the one in the Tornado F5... I would agree with that.

I'm only quoting this part because it's an awesome statement of fact and I totally agree with it.

 

Still haven't seen BGA scores equal LGA with an fixed exactly same overclock!

Yeah, my bet. Because 50% more performance shouldn't give a lot lower scores.

Mind me about Kool-Aid drinkers.

FYI Never in Hell the 6820hk's average OC is 4.4GHz. This silicon is one of the worst from Intel ever. 4.2GHz and voltage easy passing 1.4v

 

- i think @Mr. Fox meant 4ghz something mobile cpu can reach. 4ghz bga/4ghz lga, and compare the voltage used for the two and LGA easily come out on top. we can't go too high because voltage increase exponentially and judgement becomes a lot harder to decide which is better. 4ghz might be too harsh on the BGA chips though, already using more than 1.2v..

- i see what a convenient term thats what @Mr. Fox meant about kool-aid drinkers LOL i read it b4 didn't know what it means, but now i know.

- for 6820hk i took johnksss as an example but as you know that guy always do the extreme. his does 4.6ghz and its clearly too high of a standard for even overclockers like me so i put 4.4 LOL. 4.2 is just sad

 

I often forget that many may not be aware of what that means. I'm not young and many in this forum were either children or not even born yet. For others that may not know, this is the tragic story behind the "Kool-Aid Drinkers" expression (very sad).

Drinking the Kool-Aid: A Survivor Remembers Jim Jones

Spoiler: The BGA is Good Fairy Tale

Intel, and now NVIDIA, and the BGA turdbook peddlers are the "Jim Jones" of PC Gaming. They lie and deceive innocent and trusting people into buying their nasty garbage.

 

Desktop and mobile chips are in two different classes, therefore will produce two types of results.

Purely from a numbers standpoint the desktop chips are expected to perform more efficiently in terms of frequency to power ratio.

The oxymoron factor here is how the mobile chips cost just as much, if not more in some cases than desktop chips. Crazy I know.

There are different laptops that suit a situation better than others for portability and practicality. Having options is good.

When it comes to actual desktop replacements, it is expected to have desktop parts so in that regard having a lga/mxm is good.

What it comes to portability, having a laptop that fits the occasion and task is good.

Since the release of Kaby Lake and Pascal, the gap has narrowed significantly between lga and bga chipsets, performance wise. Long gone are the days of the maxwell era.

Not all data should be based on the extremes, because most people don't need an extreme OC to successfully complete their work tasks.

The same can be said for gaming. A game doesn't need an OC of Mach 10 to have an enjoyable gaming experience.

The important thing here is to realize that bga and lga are in two different classes. Sure they share the same skin, but what's under the hood is different, in terms of the system that it is in.

Knowing this, to compare a desktop chip and to be surprised by the fact that it performs better a mobile chip is not very crafty. There's is no point, because it is obvious. It's like putting a varsity team up against a jv team. Common sense.

It's not all about numbers. The 7820HK can easily do 44x all day and most can comfortably cruise at 46x. Where the air gets thin is at 47x and 48x. Yes it's bga (which I don't condone in a high end gaming laptop ), but to call it a turd is where it starts to cross the line of being silly vs being practical.

A lot of desktop owners cruise at 46x with their 7700K's and for some it's even a stretch for them to oc any higher without running into issues such as heat etc... A 50x-51x oc is obtained by a very little group of people. It is not the norm.

So again, it's important to keep practicality in mind. One doesn't need such extremes to enjoy a given system, bga or lga, in order to game, work etc...

OK now... How about that F7? I'm very curious and it's in my scope. Please let there be some design and quality improvements. However, most importantly let there be a bios with @Prema magic. Without it, one is just as doomed as a soldered chip in many ways...ie "limitations."

Oh my I'm typing this from my phone...









::iunlock::

 

I can't see this as yoo bro iunlock. No changes!! I'm old but not blind. Even the gap between old low entry mobile socket i7 and newer BGA aint CHANGED!!

 

i'd have to agree with @Papusan there. though i understand you are doing a general statement. laptop mxm GPU always followed a traditional path of shrinked down of a desktop version. 580m = 570, 680m = 670 etc and thats very much due to heat and power constraint in a small form factor. by doing a full 980 and 1080 in laptop power and heat increased, hence laptop's cooling design have to reflect that accordingly to not overheat. GT83VR GPU heatsink is an excellent example, very well made GPU heatsink.

on CPU side of things, it goes far back to first gen i7 which is the 920xm or 940xm. take a laptop cpu and desktop cpu from any of the 4c/8t i7 gen 1-7 (sandy, ivy.... to kaby), clock them at same frequency 3.8ghz, undervolt them as much as you possibly can and bench without crashing. desktop chip will have same performance as laptop cpu, with exception uses less voltage/power, anyday, in ANY generation.

it is clear that intel use to use these lower quality silicon as i3 is no longer the absolute truth, they also throw massive amount of low end silicon to mobile BGA i7 and limit its clock so people dont find out.

we are NOT blind and stupid.



as for mobile BGA and portability, yes laptop can be made thinner if BGA, but thats just it. paying possibly more for less performance, unable to repair machine, all for .5" to 1" thinner which regardless all can be put into a backpack.. no can do

 

I think what @iunlock is saying, in less of a blunt manner, is that BGA performance is not as pathetic as before. I think that is a reasonable statement to make. It doesn't mean it's a good product or that welding it to the motherboard is OK. The gap between BGA and LGA is a little smaller than in was in the beginning when Intel stopped supporting socketed mobile CPUs.

The fact that they cost as much or more rises to the level of immorality as far as I am concerned. The OEM/ODM is charging more for less and making more money offering less. From the customer's perspective, that's just messed up.

Even if the performance were identical, BGA would be inferior by design. It would not make it OK if 7820HK and 7700K performance were identical in every respect because the product using the 7820HK is a disposable, dead-end, inferior product that is more expensive to repair and offers diminished flexibility. The only people it benefits are the people selling it. It would run too hot in thin and light notebooks to be useful and it would be a really stupid choice in the notebooks that are not thin and light. In other words, it would be OK only to the people that are already OK with it.

When mobile Extreme CPUs were still a thing, those were socketed and their performance was not remarkably different than the same generation desktop i7 quad core. Sandy, Ivy and Haswell mobile Extreme CPUs were right on par with the K-series desktop quad cores by almost every measurement. If they were nicely binned samples, some even performed better than the average desktop K equivalent quad core. So, in that respect it is one step forward, two steps backward and BGA still has some catching up to do on performance.

The problem of the performance disparity, in and of itself, is not as much of a problem as the misrepresentation and deceptive marketing crap. We see examples almost daily of people that have regrets over buying their first BGA turdbook. It's not the people that wanted something extra thin and light, and Lord knows they are not more affordable. Those people want what they want and performance isn't the most important thing on their list. It is the people that asked for a beast and were handed a bunny that are not happy. Who in their right mind would intentionally spend the same or more money for a gigantic notebook that is crippled? I know that we cannot say nobody would, but clearly, a lot of folks would not and only did because they were duped. If you look at the amount of outright nonsense that is being used to market inferior products as being something they are incapable of being, it's atrocious. We have seen a number of press releases just recently that exemplify the magnitude of this deception, and it's startling how many people that are goo-goo-gah-gah over this stuff don't have a clue they are being suckered.

The more the OEMs are able to identify ignorance in their customer base, the more comprised the trash will be that they sell. Just look at the ridiculous excuse for thermal management we see anywhere we look, even among the most expensive turdbooks. They can rationalize the acceptableness of selling crap on the basis that most of their customers have demonstrated they don't know anything and will believe whatever they are told.

 

its so sad but so true.

 

Ah my head hurts. Too much fighting, not enough togetherness. Let me dive in and clear up some things.

@Mr. Fox Sandy, Ivy, Haswell, Broadwell, Skylake, Kaby Lake mobile and desktop chips, with the same RAM and same clockspeeds, will score the same results in any test assuming the systems are configured the same and the same procedures are taken. Their IPC is equal. If you have a 4.7GHz 7820HK and a 4.7GHz 7700K they're the same, in effect. Not trying to change your hatred of BGA here, just don't want you to bash what isn't bashable. Same @Papusan on this particular point.

@Ionising_Radiation Notebook CPUs do indeed draw less power at idle and whatnot, they're designed to pretty much use less voltage and there's other benefits like deeper sleep states that save battery power when asleep and all sorts of things, but when overclocking these pretty much go out the window. The unlocked mobile chips are, 100%, without a shadow of a doubt, binned far worse in terms of extracting performance. They need significantly more voltage to achieve the same speeds their desktop counterparts can achieve, and draw a lot more power as a result, and hit silicon limits much faster. I haven't seen anyone with a 7700K here unable to hit 5GHz stable on a Prema mod once they knew what they were doing, despite a plethora of reviewers claiming the chips often don't hit 4.9GHz+ stable... I guess it means they don't know how to overclock. I didn't, either. When I tried 5GHz I kept getting errors or current limit throttle in some tests (not Cinebench or TSbench) and I had to adjust other settings that had nothing to directly do with the CPU in my BIOS. But here we're tossing experienced overclockers at things and 7820HKs aren't hitting 5GHz stable. Or 4.9. I haven't even seen 4.8 anywhere. I know @iunlock can get 4.7, I'm not sure if he has 4.8 stable, but runs 4.6 for his daily system. Either way, the mobile chips WILL DRAW MORE POWER WHEN OVERCLOCKED, and their efficiency only applies to lower clockspeeds/usage states. You were making a big statement against this earlier, I want to clear that up too.

@Wormwood has a point about equal testing, I'd rather relegate 7700K and 6700K testing to mobile systems only, because even if you LN2'd a notebook you can't go above a certain voltage point which I hear many complaining about for getting that beloved 5.5GHz, so even if it's ice cold the super high clockspeeds should need more voltage than we could give, right? However on this existing point too, Mr. Fox and others have often benched notebooks without any kind of desktop cooling solutions or mods in the past when the mobile extremes were a thing; there has to be some give and take. If 7700Ks are available in laptops with a BIOS to take advantage of it out the box (and they are) then fair game, laptop v laptop benching. But I will concede rank 1 water/LN2 cooling tests on a desktop probably don't compare to what one can expect in a laptop; otherwise might as well check desktop vs laptop entirely.

I have my own feelings on BGA vs LGA and whatnot, but this doesn't really affect most peoples' choices. But on the performance side of things, most people are happy if they can get what they want out of a unit, and overclocking isn't usually what they want... but if a person wants a productivity focused unit I'm still going to pick a $3000 7700K delidded over a $2900 7820HK to recommend.

Now, no more fighting. Everybody together. We can discuss everything awesomely. If someone isn't understanding what you're saying let's ask what they don't get and make stuff more clear. The more people understand ALL sides of the story, the better.

 

Should probably point out here that the 7820HK may very well be limited by power delivery and not voltage. Cranking up voltage probably fails because the VRM can't keep up.
There's probably fewer power phases since the ODM technically only has to allow for 65W or whatever the boost TDP is. Any system built for the 7700K would need 100W+.

 

Yes that is true, to say that a 6820HK's avg OC is 44x is a comedy skit. I've seen literally about over a hundred of these to know that the true avg for a 6820HK is between 38x to 40x. Anything beyond that involves luck, a proper repaste and proper tuning.

I did have a couple winners in the 6820HK lottery and have achieved 44x and up to 46x. 44x was stable, but 46x ran a little thin. This was long ago...With the 17R3's...

Generally speaking, it is also true that the 6820HK chips were among the worst batches of the silicon....

The highest I've gotten a 6820HK is also 46x. Very rare as I've seen only two systems be able to achieve this. 44x wasn't a huge stretch with a proper repaste and tuning, but it wasn't very confident in terms of being able to rely on it.

I find that the 6820HK chips were very weak in integrity from my experience.

As @Mr. Fox as stated, I was speaking in general terms in that the gap between Pascal GPU's is a lot smaller than the Maxwell's.

The trickery that has evolved in the industry as evolved along side the evolution of the silicons themselves lol...sad I know...

Paying more for practicality is the same as paying more for better performance. With the form factor of ball grid array aside, a full blown lga system is not for everyone and it does not suit every situation. If on a road trip or plane, bringing out the tablet or more portable device is more practical than a DTR.

Now when it comes to high end gaming laptops / work stations is where the whole ball grid array is not acceptable.

There are different classes for different situations.

You've hit the nail on the head brother with clarifying what I was trying to say. There is no arguing that the bga is far inferior by design. When it comes to those who have been let down by expecting more from their purchases, a large percentage of them took things for face value, straight out of the box without any attention to addressing the thermal issues with a proper repaste / repad.

I know several people who went from an AW/MSI/ASUS to Clevo, only to come back to the bga systems. - The biggest reason was due to the limitations that even lga systems had with the power draw limitations within the bios. Therefore, at this point it becomes choosing the lesser of two evils, because in a way a lga system without @Prema magic is just as bad in many ways (limitation wise) to a bga system with the same issue.

However, for most users I believe they can still get by just fine with a Clevo 775 or 870 on stock bios, as not everyone is about OC'ing things to the moon. My 775 ran flawlessly at 46x comfortably all day.

On the other hand, there are bga systems like the AW17R4/15R3, ASUS G701VI, MSI GT73VR etc... that can all push 200W+ on the GPU without issues. I believe I've hit 230W+ on the ASUS at one point and it stood on its two feet just fine. As for the HK CPU's they can handle their own for what they are. Sure it's no K series chip, but for what they are as the (jv league) of chips compared to the K series (varsity), being able to push 90W+ on the HK's is something that is often overlooked. (Purely talking about the ability of the chip to handle such watts.)

I've recommended Clevo's to several of the people that I've dealt with over the past few years without hesitation over the bga offerings. It really boiled down to their usage habits.

If the F7 pans out to check the boxes on my list, I'll be one happy camper.

Take Razer for example. What a crap of a company. They completey rip off of Apple's unibody design, skimping out on innovation and only to spend all that money (the chunk for innovation) on brainwashing-marketing. The brainwashing is so strong that people will buy it regardless. Very sad indeed...gosh what a crap of a company. I've never seen a company with ethics, integrity and morals as equivalent as poop. It just baffles me...

 

No idea. Nobody knows. Alienware is the breadwinner of the mobile system line, but probably lack the settings I said I needed to change to get higher speeds. They also have a wattage limit.

But to say that selling an unlocked CPU and then limiting your TDP to 56W (the default long power TDP limit) is like selling snake oil to a burn victim. Snake oil in itself isn't useless, but in that context it's pointless. Same for that CPU with that TDP. Just because I can change the multipliers doesn't mean it's going to do anything beyond X amount. And if I have any say about it, things like this in reviews over at notebookcheck.net are going to point out when an unlocked CPU cannot reasonably overclock as well. I'd love to see the look on the faces of Gigabyte's Aorus team when they say 4.4GHz works through auto-OC but it barely holds 3.8 in a benchmark or sustained system stress without thermal throttling without a way to raise the power limit.

Either way, selling a locked or limited-OC chip is vastly different from an unlocked one. The 6820HKs in some units simply ran into the limit of silicon, or voltage (when near 1.4 started becoming a requirement to hit 4.3GHz+ on some systems)... it was a crappy PRODUCT, not that the OEMs underhandedly sold it. Intel is to blame. The 7820HKs, however, are pretty much what 6700Ks are in terms of how far/easily they overclock, and at what voltage they overclock.

 

Very well said. In all fairness, LGA notebooks should not be comparing anything to bga laptops (that's going backwards), rather LGA notebooks should be comparing themselves to desktop-desktops.

This is why I don't understand why two completely different class of chips are being compared to each other. @Mr. Fox gets it as he has clearly specified his views toward the bga form factor itself, rather than comparing numbers against a mobile chip which is obviously inferior.

The gap being not as wide that I've stated in my earlier post with Pascal and even Kaby Lake was nowhere close to implying that bga is equal to lga chips. I had to go dig through my files to pull this screen capture out, but for a full bga system to be pulling 228W+ is something we'd never hope to imagine in previous gen chipsets. That was my point brother @Papusan.



Mind you this was upon the first boot up, straight out of the box before I had realized that the cpu fan wasn't even plugged in lol... I won't mention what boutique I bought this from to save embarrassment, but someone dropped the ball big time as the power cable to the fan was tucked away under the mobo. SMH.

Anyhow the point here is to demonstrate that even bga laptops can hold their own for what they are. I pull 210W on my 17R4 as well.

I have a screen shot somewhere showing my 7820HK hitting upper 90W+ (before I had tuned it). This was when I was tuning 48x. If I find it I'll gladly share...

 

Oh, don't worry. It's not possible to change my hatred of BGA. As I have said, and will forever continue to say, they are inferior by design... the fact they are BGA makes it so. It's simply a fact and I won't even entertain the notion of debate on the matter. This has nothing to do with performance, and achieving equal performance (should it someday occur) will never excuse the BGA filth. It's a half-assed way to build computers that are sold at a premium and referred to as "gaming" or "high performance" notebook. For the cheap stuff it's fine... who cares, right? It's cheap.

What you're saying is 100% accurate about the IPC. The limitations on performance have to do with their generally poor bin quality, their inability to match the desktop K CPUs in overclocking, and the fact that they are generally attached to inferior motherboards that are under-engineered and less capable. At the end of the day, they end up getting left in the dust. Not because they are BGA per se, just because they are an inferior product and generally soldered to another inferior product. And, the firmware crippling is generally worse with the turdbooks due to the stupid preconceived notions of the retards that build them castrating them on purpose. Even so, OEMs/ODMs are somewhat agnostic when it comes to their digital genocide practices. Cancer firmware takes no prisoners and often ruins things on BGA and LGA notebooks alike.

 

Point understood.

So basically notebook HK CPUs behave like AMD's Zen architecture: binned for very high efficiency at lower clock speeds, and hit voltage walls as the clock speed is ramped up. Isn't this typical of every single mobile process?

At 4.0 GHz the 6820HK apparently draws 54 W (according to that NBC link I mentioned earlier), and it'd probably go up to 60 at 4.5. Doesn't that roughly match desktop CPUs, then? The 6700K has a base clock of 4 GHz, going up to 4.2 GHz.

The latter are rated to draw 95 W at full tilt. Compare that to the HK processors. Half the power draw for 95% of the performance at 100% the clock rate seems like a very good deal to me, even if I can't get it beyond 4.5 GHz (which, incidentally, is the stock clock for the 7700K).

And @Mr. Fox: your 'it's cheap, who cares' attitude contributes to the problem. Most people buy cheap computers precisely because they're cheap. But they use them for five years, far longer than you've had your P870 (which I bet you will want to change when Coffee Lake is released and turns out that the Z170 chipset will not be supported).

However, just because something is cheap doesn't mean it ought to be disposable. Three or four years ago literally every single laptop with an i5-5410M, i7-4700MQ and so on could be upgraded. Not so with what we have today.

 

Not exactly. AMD is silicon limited; you need LN2 to realistically get 4.2GHz+ in almost all their chips. Not really voltage limited, I think. But yes, that's the gist of it. And no, it is not typical of every single mobile process. @Mr. Fox can tell you about his 4.8GHz 3920XM endeavours and the many other people who got very high OCs on their Sandy/Ivy/Haswell chips with notebooks that could deal with them. Skylake was piss-poor for binning, and so is Kaby Lake. You could say Kaby Lake does fine and clocks high for a mobile chip; sure. But since it only matches Skylake when Kaby Lake desktop chips can go much higher and use much less voltage... they are comparatively terrible.

power draw is voltage, ampere, and heat related. 4.4GHz on a 6820HK is almost certainly 1.3v+ range; 6700Ks can do 4.6GHz on 1.25v or less. The power draw on the 6700Ks would have been a lot less at a higher clockspeed. Anybody who has access to a 6820HK in a good system would probably be able to show you maximum power draw for them, but I know of no such person who still owns one. Actually... maybe @bloodhawk can tell us?

Yeah, that rating is useless. All it really is, is the default TDP limit of the chip. Out of the box these things have 91W TDP and 112W short power. I didn't care and set mine to 300W each, not that the CPU will ever, under any circumstances, draw that much. And once more I'd like to remind that just because a chip is rated for a certain thing doesn't mean it won't use more if it needs... here's that nice old screenshot of my 4800MQ sucking a cool 84W because Linpack. At stock speeds. UNDERVOLTED.


Ignore the stock ratings on the chips. You sell an unlocked CPU, it's fair game. The TDP is so low on these things because they want small, weak power bricks like 150W-180W for gaming notebooks. The boost has a massive range also because of this; even if a CPU is claiming it's TDP throttling OEMs will still say they aren't. If courts had an actual IT user in there to cross-reference, we could try them all for fraud, let me put it that way.

 

Ah, yeah, I remember flashing old microcode onto my 4710MQ, cranking it up to 4 GHz and the CPU momentarily drew 90 W before thermally throttling (lol, 90 W on a W230SS).

So TL;DR implementations of the 7820HK

a) have poor/insufficient power delivery, leading to TDP throttling,

b) Are binned worse than desktop CPUs (this is the part I honestly don't get—why the hell would Intel give worse bins to notebook CPUs given that they want to maximise battery life?)

c) Have worse thermal apparatus.

In this case, how can we accurately compare a 7820HK and 7700K (say, P650HS vs P750DM3) besides generally saying that one system is overall not as good as the other?

 

More laptops don't have the power delivery module (VRM) covered by a heatsink

 

While we're on the topic...

I'm actually working on a bone stock 17R4 now...this is under full load at 44x, ran just fine on benches and held clocks... 1.2V stock before any tuning...I'm positive I can get that into the 1.1V's



::iunlock::

 

You mostly have it, and that's the point. Mobile chips cost more, are worse for performance (not that IPC is lesser, but the binning is terrible compared to others in its line, starting from Broadwell, and stock speeds are SIGNIFICANTLY slower than desktop chips starting with Skylake) and are all soldered to the board and thus disposable. Then people will charge $9000 for them in a ****ing Acer Predator. Well done, everyone.

The socketed nature is only really for in-generation upgrades/swaps/replacements more than anything else, truly. Users with the PxxxDM series can use Kaby Lake... with Prema's help. That's the only upgrade I've seen. Covfefe lake is probably going to need a new chipset because intel are extremely stupid.

Either way, as I always say, mobile sockets are just needlessly expensive for no real reason. Even if the mobo is more expensive by design, it isn't $600 more expensive. So more people buying socketed units are not really happening so much in this current era. But it's important to understand the benefits and downsides of each route; even if the benefit of one is only the effective price the consumer pays.

 

Charging $9000 for a bga system is a Crime. I want to interview the people who buy these systems. I'm good friends and in contact with a well known YouTuber and even he is like, wtheck is this piece of poop? They could have easily fit a lga/mxm in here, especially for the money. It's a crime, a scam and outright highway robbery.

As for the CPU's, correct me if I'm wrong, but to further validate your point about mobile chips being the bottom of the barrel, these chips are cut from the same silicon. ie... K series, HK, HQ can all come from one sheet of the silicon and from there they are sorted. The ones that don't make the cut get labeled/classified as HK chips, then the next being *face palm, why do they even exit* HQ nightmare chips.

I have seen that a lot of people get efficiency confused with value. That perception is very wrong....oh how thick this smoke and mirrors have become... @Mr. Fox ... there's the magic formula... drug them with marketing candy, smoke them into a glaze with fluff and skin them of their wallets. Brilliant. @ole!!! , you're correct in that, we're not stupid!

 

what blasphemy, these soldiers should know better! clevo is less evil!
oh and i agree on razer being trashy. i will probably never buy their product lol.

well yes and no. D2 ultima mentioned is more of a specific comparison. ultimately these cpu and silicon are cut from batch. basically, any quality of silicon will have their "best efficiency" at a given clock, and once past that clock the amount of voltage needed increase. mobile chip at very low frequency dont actually have better efficiency than desktop chip, its the feature thats set by OEM such as things like speed shift, power saving etc. desktop may not have those additional tweaks built in but they make very little difference tbh.

whats the point of not using cpu past 3ghz at their best efficiency right? seems silly when i can pay for same cpu and overclock it to 5ghz. use 4x more power? bring it!!

well its greed. look at this, best quality silicon goes to xeons (server/enterprise), failed server/enterprise goes to HEDT i9, then move onto mainstream desktop i7. im sure in the early history, some are even thrown out as they dont pass quality standard but once intel has hit 95%+ of market share they have to think new ways to increase profit. so the worst ones aren't thrown out unless they are complete unusable, they are given to i3 and nowadays goes into mobile CPU.

reason they cap i3 clock speed is because intel KNOWS that these silicon are trash which even at low clock they will use high voltage or chip won't operate. thats why you get bunch of these chips. i remembered my 2630QM use so much voltage i thought it was normal, which i later replaced with 2960xm then realize 2630QM had junk silicon quality.

you know whats also funny? WD and seagate do similar things with their hard drives, the crappy ones have lower read/write speed goes into their external "backup" device with USB 3.0 controller purposely controling the limit of read/write. thats why you find these external backup drives already in an enclosure done for you, usb 3.0 cable, box, power supply, all provided for you for less cost than their "bare drive".

well in a way WD/seagate is a bit better because they provide less performance devices with less cost and more in the package, where as intel charges more for BGA while giving less brainwashed consumers.

MOBIUS COME BACK TO CLEVO BROTHER

 

Looks like the first Coffee Lake H benchmark has surfaced, thought intel was skipping Coffee lake for mobile and jumping straight to cannon lake.



https://www.notebookcheck.net/First-Coffee-Lake-6-core-mobile-CPU-benchmarks-are-in.245804.0.html

Seems to be quite a hefty jump in performance compared to a 7820HK, even in single thread.

 

Intels hit a homerun with ulv quads and now with the HQ series cpus. I'm still not convinced on the desktop side.

AMDs 35W Raven Ridge will not be able to compete with cpu perf with Intel hexcores nor will it be able to beat rumoured 1040 in gpu perf.

not sure what AMD has in store apart from ravenridge for laptops :/

A tweaked R1700 to battle Intel HQ series undervolted vega 56 on the gpu side ?

 

?

I have owned the P870KM1 for months now...

 

Looks like someone didn't read the article:

 

I don't know about their 7700HQ, but my 7820HK doesn't reach 4k single thread even OC'd to 4Ghz

 

Regardless of perf increase (which would probably be negligible in real life) it would take a while to actually get the higher end parts in anything, including Clevo

 

On the notebook side they likely intend to rely on the overall single chip package for being reasonable and cheap to produce.

 

The benchmark was v3.4.1, single-thread, 32-bit.

NotebookCheck's average 7700HQ score is 3600. The 6700HQ achieved 3400 on average. This looks like a very typical marginal 5-10% IPC increase that we've come to expect from Intel over the past 5 years.

Nothing to see here except the extra 2 cores. Useful, definitely, but it was a reactionary measure to AMD Ryzen.

 

Extra cores does not tend to impact gaming too much at the moment either.

 

+rep for your post
From Hwbot.org
[email protected] max in Cinebench R11.5 http://hwbot.org/benchmark/cinebenc...Id=processor_4390&cores=4#start=0#interval=20

[email protected] max in Cinebench R15 http://hwbot.org/benchmark/cinebenc...Id=processor_4390&cores=4#start=0#interval=20

Havent seen fully working 6820hk passed 4.5GHz in any importent benchmarks.

You forgot to read the text inside the parenthesis in your Core i7-6700HQ and i7-6820HK Review link "The two Skylake quad-cores could even run as 35-Watt chips under practical maximum load for the CPU, and the specified TDP of 45 Watts should not be a problem for the top model i7-6920HQ. Our K model only consumes more with 54 Watts when we overclock it to 4.0 GHz (3.8 - 3.9 GHz in reality, probably limited by IccMax) – the advantages of the 14 nm manufacturing process are getting smaller in this frequency range."
Expect the power consumption would be higher if the 6820Hk BGA chips could run stable 4.0GHz in Cinebench R15 and not the confirmed 3.8 - 3.9 GHz !!

Forget 60w power consumtion with [email protected]. Voltage will pass 1.4v and the power will be out of control!!


Regarding Notebookcheck tested max power consumption for Intel's next best (Skylake) mobile i7-6700hq in Cinebench R15 (31,4w Cpu Package Power)... I tested same clocks with my [email protected] and the power consumption with equal locked down clocks was a bit over 28W in Cinebench R15. Of course will a cooler chips pull less wattage but still... 28.1 vs. 31.4 is a noticeable difference aka over 11.5%. Say that BGA chips is better binned for low clocks + efficiency is a bit off. Maybe for idle clocks but not for fully load.

Tested 7820hk in Notebookcheck showing that the hk chips don't run correct with higher overclock(+4.1-4.2GHz). Either due crippled firmware, power settings or thermal throttling. The scores with overclock literally sucks!!

 

No, at very idle situations it's difficult to make a desktop chip use under about 10W. I have gotten my 7700K undervolted to do so if I don't use static voltage, but it's still usually only pitfalls that do not stick. I've seen 7700HQs sit at about 1W total, both 7700K/7700HQ downclocked. Make no bones about it, they design those chips for power frugality at lower clockspeeds. Hell I can get 2.7GHz out of an i5-8250U at 15W in TSBench if I undervolt the thing a little (CPU-only testing mind).

i3s are actually pretty fast all things considered. They used to be very slow, but since Haswell Refresh they've been 3.7GHz and all that. Not bad for such a low end chip, really. Though they can't boost and excepting the 7350K can't OC, but no idea why you'd buy one to OC it anyway. As for your 2960XM vs 2630QM, that was probably just chip to chip lottery. I've had two 4800MQs and my second one used less voltage (sub-1v at 3.5GHz to be precise) but could barely OC and needed an entire 100mV for stability just going up about 300MHz. The first one drew less power while using more voltage and was a much better and cooler chip, I miss it to be honest. I never got LM before I lost that one.

Don't ask me, people are stupid.

Maybe. They reuse a lot of silicon. There's been chips that have dual names. 4810MQs read as 4900MQs AND 4810MQs in CPU-Z, and some 4710HQs acted like 4710MQs and had no innate power limits.

Wot? Where? You mean the CPUs that get throttled to 1.6GHz just by playing Hearthstone without a dGPU or eGPU plugged in because it's 15W limited and the iGPU shares that power limit with the thing, rendering it pretty much useless for anything that requires any load on the iGPU as well as CPU loads, so it looks ok in CPU benchmarks and not in anything else? THOSE ULV quads?

I don't know how those H hexacores will work either. 2.6GHz base clock is still rather low, who knows what they're gonna do with boost. Even moreso, we don't know if Covfefe Lake is going to use the new mesh cache design either. If it does, you can expect anything that uses the cache to perform SIGNIFICANTLY worse than previous gens. Know how the 7900X beats the 6950X in things like cinebench that are completely parallel (AMD does well there too) but suck at things like games? Needing somewhere around 5GHz to match 4.5GHz from the previous gen in the same titles? Yeah. That. You might find that the majority of games will perform WORSE with Covfefe Lake if that is what it's going to be using.

Good luck with that one.

7700HQ and 6700HQ have the same IPC; the 7700HQ is just a bit faster in clockspeed. Also can use 2400MHz memory out the box unlike skylake which actually needs a mobo to support something at times.

See above cache design possibility.

Six is the limit. A lot of new games won't max a quadcore but will benefit from a hexacore. Even Overwatch, it gets very easy to hold higher FPSes with a hexacore on it. A lot of other games like it too.

 

@Papusan ..want to see something funny?

Buckle your seat belt...LOL..

I've been looking for the folder with my 17R3 benches, but for the life of me I can't find it...I did find this and thought it'd be funny to show the voltage ROFL...



I remember testing this unit..it was my 2nd 17R3 at the time and I was seeing how much gas it consumed bone stock. I believe this was before a repaste and of course tuning...if I remember correctly, I sold this system for top dollar to a hungry chap that couldn't wait for the back orders at the time....gee seems like forever ago...

Anyhow...thought you'd enjoy the bbq grill...

 

This from the fully locked 7700hq with equal Os. See single-core score. Almost equal. Where is the hefty jump in performance in single-core? This ain't even a 7820hk.

Any better? Of course... Almost 4.7GHz


You see... Fully throttle!! http://hwbot.org/submission/3182795_mostafa_abbasi_cinebench___r15_core_i7_6820hk_946_cb

 

All performs differently because of minor increase in BCLK frequency. 101MHz make a huge difference in clock speed.

 

With a fully locked down 7700hq you can reach 38MHz extra with increased BCLK. But what OEM will let you run 101MHz with a locked down BGA?

 

I'm not familiar with Geekbench, how come a 7700HQ scores higher than a 7820HK at 4GHz?

Just did a quick run:

http://browser.geekbench.com/geekbench3/8437011

 

It's benchmark. Scores can differ slightly, and you may have cancer firmware or trash aka bloat running in the background. Fully locked down 7700hq https://browser.geekbench.com/geekbench3/8307935

 

Oh, just noticed that the scores for the memory segment are basically double than mine, probably becuase I'm running a single 16GB stick, that would explain it, the scores for integer and FP are higher on mine.

Disregard what I said about "hefty" increase in single thread performance then, you're right, sadly it's going to be the usual increase as the last few years.

 

what i meant to tell him is at usage, each chip has their most power efficient frequency and thats based on how well silicon are binned and correspond to its quality. mobile BGA loses out pretty much at all time. i wouldnt run my cpu all the time at 800mhz to just 1.2ghz.

 

http://www.hwbattle.com/bbs/board.php?bo_table=news&wr_id=85751&ckattempt=1