Clevo Overclocker's Lounge

My laptop is 13", it has an rPGA socket. Even the great, tiny W110ER had an rPGA socket (3610QM). It all depends on what the OEM wants to do with the hardware they have.

That's not true. The MQ chips were plenty more powerful than their HQ equivalents; they also throttled less. Furthermore, the
4940MX was rPGA. Let loose after overclocking and with a good thermal dissipation system it absolutely wrecked everything else that wasn't a 4770K in terms of performance.

LGA isn't cheaper, it's actually more expensive when everything else is considered. The socket is too big and the CPUs used have too high a TDP to be used in properly portable notebooks. Not to mention the annoying IHS that Intel has slapped on the LGA-packaged CPU.

On the other hand, the P6 chassis could easily fit the rPGA socket in the hypothetical scenario that we had a 6700MQ, or even better, the 6820MK, or should I say, 6820MX? It would still eke out more performance, and more consistent performance than the 6700HQ while still having reasonable proportions. Note that I am not talking about desktop replacement notebooks here.

 

Refer above to his running the chip at the same rated speed as the mobile counterpart. @Papusan, repeat and show wattage draw with hwinfo to show exactly how little wattage is needed!

To address the ihs, they could sell it without the ihs. Only reason they don't was because over a decade ago, people were over tightening and heat sinks were getting larger, so the ihs protected from die cracking and distributed the weight. As to the form, whether the socket is larger or not, doesn't matter which they use for pins. The inconvenience comes from locking the cpu in place without the ihs. A delid die guard for laptops could deal with that. But trust me, the wattage at that low of multiplier is equal our lower than the bga crap out there!

Edit: just looked again. 28.43W, well below the crappy bga chips!

Sent from my SM-G900P using Tapatalk

 

It's not cheaper when you consider the absurd premium Intel is free to slap on rPGA CPUs. Look at the 4930MX. What was it, like $1K? One advantage of using LGA processors is that Intel can't just say "it's mobile so we're going to charge you three times the amount a comparable desktop CPU would cost." When not locked down by firmware, TDP is also just an arbitrary number saying "this is the sort of operating environment we expect," not what you would expect it to draw running full tilt when fully unleashed. See @Papusan and his run at 6700HQ clocks with a peak draw of 28W in CineBench. I believe @Mr. Fox had like 130+W pumping into his 4930MX at some point.

 

You're running Skylake, I'm on Haswell. Also, I'm talking about rPGA here, not BGA. Skylake draws way less power than Haswell at low loads. Show me the power draw of an i7-4700K vs an i7-4940MX.

They could, but the fact is they don't. It doesn't matter what Intel can or cannot do, but what they actually do. What they've done now is make potentially great, upgradeable systems like the P650 series, redundant after six months to one year.

Someone with a Haswell 4700MQ + Maxwell MXM GTX 970M P650SE machine could've upgraded his/her CPU and GPU to a Skylake 6820MK and a Pascal GTX 1060/1070. But no, now the machine has gone through four iterations (P650SE -> P650RE -> P650RP -> P650HP), or from Haswell -> Skylake -> Pascal -> Kaby Lake. All because of no more rPGA.

This is a logistical pain. No point in fantasising about the ideal, utopian world. The real world is messed up as hell, we need to fix it first.

 

A 4790K owner can't just do a straight swap to a 6700K. The socket and chipset are all vastly different. Haswell uses a FIVR whereas Skylake doesn't. The only difference here would be that you could pop in a Kaby Lake CPU instead, and even then, there would likely be a new release due to new chipset features. Now if you're arguing that motherboards should be backwards- and forwards-compatible whenever technically feasible, you'll find no arguments from me here, but that has nothing to do with LGA vs rPGA vs BGA.

 

Edited

 

Well firstly, Skylake doesn't slot into Haswell. You would be stuck at Broadwell. The pure fact that CPU architectures over generations barely improve *AND* you were getting at most 2 CPUs per socket type was why I don't find soldered CPUs the stupidest thing in existence. It's pretty far up there, but even on yours or my system in peak condition we couldn't use Skylake, and broadwell is give or take worse (hotter, more power hungry, only 3% faster per clock).

The machines definitely have had revisions, but Clevo always does those. PxxxHM was Sandy Bridge/Fermi. PxxxEM was Ivy Bridge/Kepler 1st gen. PxxxSM was Haswell/Kepler 2nd gen. PxxxSM-A was Haswell/Kepler 3rd gen (though this was to fix a mSATA bug on the second slot from what I hear, it still was a new revision that had other changes like the back cover and got updated software etc). Clevo would be making new revisions anyway, honestly. I don't *LIKE* that it is being done to such frequency; but that's how they work.

Don't get me wrong at all. I still think BGA is terrible for all-out designs. I still also feel that a proper mobile chip in a socket with a good cooling solution is preferable to desktop chips in a notebook (especially since we don't have the IHS to worry about). But the chips being BGA isn't what makes them suck. That's BGAtel doing what they want. Skylake was a godsend to small notebook making ODMs. It was cool, it ran much more slowly than previous gens, and it doesn't have a socket. They simply never bothered to care about enthusiasts since they went full BGA. 6820HK is too bad of a chip for me to consider it acceptable as an enthusiast, especially considering how much I see CPU usage skyrocket in games in 2015 onward and how 120Hz is becoming a thing again. It's not even a matter of getting a notebook that can cool it, they simply don't pass 4-4.2GHz reliably, when the LGA models START at 4GHz (whether or not LM TIM and delids are required).

What you should be looking for, is an overhaul of the entire laptop ecosystem.

• No more intentional crippling of *ANY* notebook part for ANY reason. Make chips for the occasion. The 4702MQ, 4712MQ, etc were lower clockspeed, lower-power-target chips for thinner notebooks. Bring these back, and do NOT make them the default.
• Officially state that if boost is not held under sufficient load, the chip is throttling and the laptop needs tweaking. If it cannot be fixed, then it is defective and should be replaced or refunded. This obviously will not apply to xxx2Mx chips or U chips as they are designed to be lower power first.
• Raise the TDP of any unlocked chip beyond this silly 45W crap. 65W minimum, let cooling solutions be designed around 70-100W for higher end chips at the least. Unlocked chips should have the same TDP the desktops have, and should not be put in thinner notebooks with minuscule power bricks. Stop giving idiots the wrong idea about what notebooks can and can't do.
• Kill the prices of notebook CPUs to match desktop ones or cheaper. The mobile unlocked chip should not be over $350 USD. It lacks an IHS and TIM inbetween the die and IHS, and the tech is exactly the same. It should never cost more. Give us equivalent binning between mobile OCable and desktop chips... with a raised TDP barrier, the coolers will no longer give up the ghost 2 seconds into SuperPi and Chill.
• Clear differences listed in the chipset for people to read. No more confusion between machines like with HM170 and CM236 where the former lacks 4 PCI/e 3.0 chipset lane slots and has either slowed NVMe performance or can only use one NVMe drive at a time etc and people wonder what's going on. Make it so manufacturers need to show the chipset names and people can look it up before buying. Desktop boards need to show the chipset they have; laptops should be similar. Have it easily available; let people do their digging otherwise.
• Give us a bloody mode where we can be using the dGPU only but have access to the iGPU for quicksync hardware acceleration. This is NOT HARD.
• And finally, bearing on the previous chipset point, give us quad channel memory support in a chipset. Mainstream desktop CPUs don't need it, people can buy enthusiast-line CPUs if they need quad channel (and those CPUs will also likely suit whatever the user is doing better anyway). But laptops don't have CPUs or chipsets designed for it, and we should have it for our top end systems. It's not impossible. People got plain old DDR3 SODIMMs to run in quad channel on the X79 chipset in the P570WM... granted it needed Thaiphoon Burner to do so, but the fact that the chipset supported it on a basic level meant it was possible to work. So we can get some quad channel memory for high performance workstations in a mobile form factor.

We start doing all this, and leave desktop CPUs in laptops to things like Skylake-E, and we'll be good.

 

I tested a few benchmarks with [email protected] BGA Filth clockspeed.
Wiping out the the 6700BGA Trash bench scores on Hwbot with same clock speed. Wprime 1024M maxed out at 31.1W and max 33C degrees. Best Wprime scores for [email protected] in Hwbot (206,273 sec and 6sec 745ms) was smashed in a jiffy
http://hwbot.org/hardware/processor/core_i7_6700hq/

This was a weird experience. I could drink up my cup coffee, go out with my dog before the Wprime 1024M stress test was finished, LOL

4940MX was some of the worst that came from Intel. If you wanted to switch from 4940mx to the first Hotwell revision 4930MX you were almost ruined. Mobile Extreme was ancestor pricey.

And you should compare 4940Mx with 4790K. Not the trash 4770K
Devil's Canyon would eat 4940Mx for breakfast. With rPGA socket you would be screwed.

Picture this.. If you've got a useless 6700K in the new machines and wanted to change to a better one... This would cost you around $ 350.

Another example... If you upgraded from 4700mq to 4930Mx this would cost $ 1,000. Current situation with LGA and the newest laptops... If you'd upgraded from 6700 to 6700K, this would cost you $ 350. rPGA socket for powerhouse laptops is a clear NO from me!! Replace BGA with rPGA, but let LGA as it is. We don't want it anymore for the most powerful laptops!!

 

Yeah, totally agree. 4940MX was the worst example of an Intel Extreme Mobile I have ever seen. They are very poorly binned and I have never seen even one example of one that overclocked worth a damn. It gets raped by 2920XM, 2960XM, 3920XM, 3940XM and 4930MX. The 4930MX is almost identical specs and performance as 4970K, but costs $1000 more new (when I last checked a 4930MX was nearly double cost of new 4790K if purchased used on eBay). If I owned a laptop with 4940MX the first upgrade it would get is a downgrade to the far superior 4930MX. Using ~$350 unlocked desktop K series CPUs in laptops is the ideal choice for price and performance, and I hope the use of them increases. They are no more power-hungry or difficult to cool than Mobile Extreme processors were. Any excuse that cites those as reasons for BGA are just lame cop-out excuses offered by losers in a feeble effort to conceal their lameness.

 

Famous last words... or prophecy of doom? Ask Alienware's General Manager, Frank Azor, Anything!



This flawed mentality is why we are overrun with turdbook filth. Unfortunately, he is not alone in his flawed perceptions... he is in the company of other like-minded trash-peddlers. As long as we are dealing with stupidity among decision-makers and there are too many consumers willing to open their wallets for junkbooks, we will see BGA garbage continue to flourish. And, try as they may, the eGPU crutch has never been a legitimate excuse to forgive the rest of the mess. Sure, an eGPU good for some geeky tinker-toy fun, but it has not and will not make up for an otherwise severely compromised product.

 

@Mr. Fox, in regards to FrankAzor, I really wish I was Gartner or some 'industry' group which had all the numbers to *know* where the real market for DTRs (read DT CPUs in a lappy) currently sits. If there are more offerings from MSI and Clevo, at least that is some indication there is money to be made.

 

There is some money to be made, but the effort is also quite high. A big company could do it, but they fear stocks and fear the inability to sell those.

Intel seems to not care much about it, on a desktop PC, changing the CPU is ~400$ for 6700K, but on desktop changing the motherboard is also much cheaper than on laptops, and this allows for infinite theoretical upgrade paths.

In laptops, there is a brick wall for upgrades, after which we cannot upgrade anymore.

Guess this is why they avoid making those.

Even so, there are people who need the power and need it now. Those people are also probably the ones who would pay for those configurations.

Making the entire laptop = much higher costs than making every component, unless someone makes the components, but there is an industry mind set of not making the components resulting in one - two companies still making those on their own.

Quite sad...

 

See http://forum.notebookreview.com/thr...6-owners-lounge.797457/page-265#post-10424647

http://forum.notebookreview.com/thr...6-owners-lounge.797458/page-201#post-10424577
Awful reading http://imgur.com/a/e8Ius

It looks like the soldered mobile i7 filth are not as energy efficient as some claime And Intel want more $$$ for their mobile i7 BGA but offer less aka no TIM and IHS

I have not seen anyone like AW Ceo Mr. Azor @Alienware-Frank turn so totally around!! Not only put into soldered Cpu... The man wanted the whole BGA package at once aka Everything soldered on the motherboard in a bang. MSI didnt tourn around as fast... They have still MMX graphics

 

Not sure if it's been said/known yet, but it seems the 7700K starts at 4.4GHz all-core turbo boost up from the 4.2GHz base clocks.

So I guess a "baby" overclock is now 4.8GHz+? xD

 

I'm still waiting to see what the temps are like.under sustained load @ 4.8Ghz. Or if it will be exactly the same as the 6700k .

Just hating the comparison to the 4790k on every freaking goddamn website.

 

https://t.co/ZuOOqQO7Ns

 

Excuse me while I drool

Is that the 3200MHz or the 3000MHz OC'd?

 

Looks like the memory controller is a lot stronger on the 7700K

 

Well, probably because if they compare it to 6700K it's less impressive and represents a smaller improvement. Comparing it to 4790K is just a whole lot better for marketing.

He and I both have actual 3200MHz G.SKILL modules. @Prema is able to use his. I'm still trying to figure out how to get mine stable above 3000MHz. The screen shots I posted a page or two back show the AIDA64 benchmark results, but they are too unstable to use until I figure out what needs to be tweaked.

Yes, it could also be that @Prema's 7700K CPU handles the faster RAM better than 6700K is able to.

 

I know 6700Ks handle 3200MHz easily on desktop boards, so I would suppose that it's a combination of mobo and CPU in this case

 

Well, searching for solutions for myself, I see more than a few desktop owners struggling with even 3000MHz. Some can, but others are having issues. I think a big part of it is BIOS related, which would be consistent with hit or miss results in desktops. Based on what I read in other forums, MSI and ASUS enthusiast boards seem to handle overclocked memory better than some of the others. Multiple things can come into play with memory overclocking (XMP or manual settings) as well, like System Agent voltage. If one piece of the puzzle is missing or out of place it can lead to instability.

 

@D2 Ultima - I guess it was more than a page or two back, LOL. This thread keeps growing and growing...

http://forum.notebookreview.com/threads/clevo-overclockers-lounge.788975/page-835#post-10423416
~46K read / ~49.5K write for dual channel mode is pretty nice... IF I can get it stable

 

Yes. MSI had serious problems a while ago. I had to stop recommending MSI boards for Skylake because I kept finding issues with high speed memory there. ASUS seems to handle them just fine. Gigabyte is really hit or miss for me in my research.

In this case I suppose it really is a combination.

There's 8467 (as of this post) posts in here lol. I'm pretty sure "a couple pages back" will never again work xD

I did see your posts about it. What I'd really like to know is if a delidded 7700K is cooler than a delidded 6700K, assuming their IPC is exactly the same, and if it works on W8.1 without any issues. If yes, then I'll officially want one hehe. I know it should use a lot less voltage for something like 4.6-4.8GHz and that should automagically impact temperatures, and by default give me a more decent acceptable overclock temp-wise. Plus 3200MHz memory *drools*

 

If 7700K only works properly with W10 it might be a deal breaker for some. I'm not sure that I am willing to put up with that nonsense, personally. The insinuation of an intent for that really turns me off, and no amount of fancy schmancy marketing will make up for such a horrific breach of decency.

Yeah, unstable memory is not fun... I did not enjoy this part...

 

Oh believe me, I saw it

 

It's native 3200Mhz RAM, but it wouldn't even boot with with stock firmware...

RAM & CPU: http://valid.x86.fr/0tla6b

Ambient temps won't allow for proper testing (temp throttle) atm, but I don't think you will see better temps from 7700K compared to 6700K, both are hot (delid pending on both).

 

Well, I have ways of dealing with hot. I like that it did that 5.0GHz validation using 1.310V, which is typically less than 6700K uses for 4.6GHz, LOL. That part is what got my attention. If 7700K can handle 1.500V or more, it may be benchable at 5.5GHz or 6.0GHz with the right tuning. Now, that would make me give up W7 if I absolutely had to (but I would still hold a bad grudge about it).

 

The news on 7700K sound good, but doesn't that mean that it is also hotter at the same voltage? Which would make lowering voltage a much more problematic thing (?)

BTW, @Papusan @Mr. Fox

IF you feel you want a 9000$ laptop, Acer have you set up. No idea if it's BGA or no yet.

http://www.anandtech.com/show/10963...tor-21-x-21inch-curved-screen-gaming-notebook

 

If you delid and the lower voltage for ~4.6GHz on 7700K over the 6700K means it runs cooler, do let me know please =D

Full BGA

 

I expect a bit lower temp due lower voltage with same clock speed. What core voltage do you need for Wprime 1024m with 4.8GHz all over. Absolutely minimum!! Same for Cinbench R11.5
You could probably lower the voltage vs. what you use now if you had @Mr. Fox AC unit under the beast. Thanks

 

We don't know how hot 7700K will be yet. If I get one I will tell you. I bet it's no different than 6700K, which is much better than 4790K or 4930MX Haswell blast furnace CPUs.

Even if the Predator 21X had an X99 CPU, there is no way I would pay $9000 for any laptop... that's just stupid. Desktop, maybe... but no way on a laptop. From every authoritative piece of marketing from Acer I have seen, it will be worthless BGA filth. The fact that Anandtech did not even bother identifying the CPU is a dead giveaway that there's nothing to write home about. I left a comment for them.

 

7700K temperatures depend on the sub voltages, the early hot results seem to be with early BIOS files hitting the core with 15% more voltage than needed.

 

Actually, I thought the BGA question was already settled. So now we have an insane price tag to go along with poor engineering decisions.

The confirmation regarding BGA design-

http://www.techradar.com/news/mobil...-created-the-outrageous-predator-21-x-1327838

Except for pricing, nothing new. This was already covered in an earlier thread in the Acer forums - http://forum.notebookreview.com/threads/acer-predator-21-x.795578/

I see @ole!!! opened a different thread as well - http://forum.notebookreview.com/threads/acer-gaming-laptop-predator-21x.796195/

 

And ACER's engineers said overclocking can put the whole machine into flames Confirmed by themselves Worlds biggest Jokebook. Not the thinnest trash but certainly the biggest. I don't know if I should laugh or cry

 

It's a concept that if tackled with finesse could be quite interesting.

A 21" machine with 4x large slower spinning fans, curved high refresh rate 3440x1440 display, hybrid mechanical switches, dual 1080s, 7700K, high quality 4.1 sound system.

I'd set a target price of around $5500 for an -S system.

 

Take out the dual GPUs, put a price tag of about 4000$ and that would be awesome. (Most of my pro software doesn't like SLI profiles)

You had to post that

Makes this hobby more the fun!!!

Why did they solder the CPU and GPU?

At 21" and that thickness, it would probably have been possible to use some desktop grade Motherboard, some custom design fans and adapters and use full CPU desktops and pretty much everything desktop...

 

You don't need finesse when you have a 21" chassis lol. If Clevo can make a 6700K and two MXM 1080s fit comfortably in a 17", there's no excuse for Acer except laziness and being cheap.

 

They want to force you over to next revision of this BIG smartphone. And people are stupid enough and jump on this BGA train next time as well

 

And novelty. To avoid issues, and to get nagging enthusiasts off their ass.
Easiest solution, avoid that end of the market and price it ridiculously. Problem solved.

 

Dual 1080s or single Titan-x class GPU at the listed price as the only option for that one IMO.

 

Yeah. Not sure how they expect people to buy that machine at that price point.

 

Singe Titan X would sound much healthier.

We were discussing a 4500$ price point for a 21" machine, not a 9000$ price point

EDIT::: There is nothing at a 9000$ price point worth of consideration.

Even 3000$ feels a bit much for a laptop.

 

Sorry of topic
Some think it is beautiful, but I do not like the snow due my not so good health. @Phoenix + all of you, come visit my home so you can go skiing. This is something other than going in the sand dunes under the sun. You're all welcome This pict is taken now tonight. Right outside my front door. As well You see my best friend...
Ivan the Terrible

Spoiler: SNOW, SNOW, SNOW and more snow

My dogs name is from https://en.wikipedia.org/wiki/Ivan_the_Terrible





The summer

 

Anyone seen my Rockit 88 delidder? Have been searching everywhere...

Hope it fits the 7700K...well once/if I can find find it.

Edit It does: https://forums.anandtech.com/threads/7700k-delid-tool-compatibilty.2495636/

Now where is it!!?

 

Use Google translate http://www.tek.no/artikler/test-intel-core-i7-7700k/366407/5
As in the review 4.9 GHz, which proved to be stable * with 1,321 1.312 volts.


Edit. http://www.guru3d.com/articles_pages/core_i7_7700k_processor_review_desktop_kaby_lake,22.html

 

That is beautiful!

Down with a bad cold right now, but I wish Romania would have some snow this winter...

 

Take the storm headed my way!!!

Sent from my SM-G900P using Tapatalk

 

These articles are kind of annoying me now. Nobody seems to be willing to actually compare the processor to anything substantially.

"Here is the CPU running at stock. It's running at 4.4GHz because of turbo boost, but we only list 4.2GHz base and 4.5GHz single-core turbo to the end-users, and don't tell them the frequency of the processor unless it's been overclocked, and then we refuse to compare it to any other overclocked processor at the same speed whatsoever".

Like, is it that hard to shove a 4.4GHz 6700K next to a 4.4GHz 7700K with the same RAM and run Cinebenches R11.5/R15, 3DMark Physics, and maybe something else like SuperPi? Is it that hard to list your cooling solution and show whether or not there is any significant differences in the CPU due to the supposedly far less voltage Kaby Lake needs for clockspeeds like 4.4-4.6GHz? I mean you can literally drop a 6700K in the same setup in the same motherboard and just cross-check.

And Guru3D needs to be slapped in the face. They just listed that a 6700K/7700K clocked at the same speed as a 2600K gives about the same performance. I didn't know that "about the same performance" is somewhere in the vicinity of 30% more than the 2600K. You'd have to clock a 2600K at somewhere around 5.8GHz to match a 4.5GHz 6700K in raw compute power, and this isn't even accounting for RAM in non-wholly-CPU-constrained benchmarks, since Sandy Bridge can't use above 2133MHz DDR3.

I can understand that they probably want to make a point about "oh CPUs aren't that different through the generations" but that's the absolute wrong way to make it. AND THEN THEY LISTED THE BROADWELL i7 AS A GREAT CPU. WOT?

/rant

Edit: I also want to know if it works in W7/8.1 with a Z170 chipset board or there's some artificial lock in place.

 

Well I mean... There should be 0 gaming difference between two equally clocked processors Skylake vs Kaby Lake. No IPC change. Heck even the stock 6700k vs 7700k is almost nothing for gaming benchmarks with current gpus.

 

This is more correct. 7700K@5GHz use less Cpu package power in Wrime 1024M than older [email protected]. This is the number I will relate to(Cpu Package power = 102.805W)