BGA Venting Thread ;)

Another point on why I starting to support non-BGA laptops:

Apparently those BGA Clevo laptops nowadays have at most 4 USB ports! I need more than that...

 

Sticking to old standards is bad. I'd rather have more USB3.1/TB port rather than more USB Type A

 

Well I can agree on it with you it isnt good. I cant say for sure if a 4 screw solution on the CPU site would fully eliminate it. Because like I said, I had uneven core temps with such a heatsink. Chinese manufacturers are just not precise enough and a lot of bad heatsinks slip out through quality control even though they are not up to standards in my opinion. This is a problem with the entire industry in my opinion, Quality control issues.....something that I see in pretty much any tech device type nowadays.

Apple controls this very well though with a laptop design that is extremely easy to manufacture without issues. Almost everything is mechanized and only the final assembly needs to be done by people. But with these more complex gaming laptops its a whole different matter. I rather pay 100usd more a laptop for better quality control right out of the gate. Or they could even ship me the laptop in parts and I assemble it myself.

Luckily I could acquire the parts myself of my current laptop to make it perfect for my uses.

 

People need info!! And some of us can give them correct info.... A lot of new buyers of notebook's don't know there are other and better options than BGA machines. See example below.

http://forum.notebookreview.com/thr...nware-17-r4-15-r3.797653/page-5#post-10408676

http://forum.notebookreview.com/thr...nware-17-r4-15-r3.797653/page-6#post-10408714

Edit. A new one as well. Info out to the people is importent @Mr. Fox http://forum.notebookreview.com/thr...cussions-lounge.797128/page-145#post-10409061

A straightforward statement. 4 screws around the processor heatsink makes no 100% guarantee, but the chance of uneven temperatures between the cores is less. And pay 100usd more a laptop for better quality control right out of the gate is also absolutely correct mindset. +rep

 

Ah yes, that too. Apparently TB ports are near non-existent in BGA laptops!

(And yes, I did include such combo ports when counting the no. of USB ports per laptop, be it 2.0, 3.0, 3.1 and be it Type A or Type C. I did so since there are such things as converters; would be really cool if resellers or even Clevo includes them with the laptops sold!)

 

I believe ASUS includes an adapter with some of their ultrabooks.

 

I'd like to throw in a wrinkle - the free market may save the socketed design.

While my sample size is extremely limited, I'm seeing more and more posts from the 'gaming' ilk starting to just now understand the limitations of their BGA designed systems - thermal / power issues, repair issues to name a few. Now unfortunately, the marketing of Clevo / Sager / MSI is the proverbial drop in the ocean compared to the mainstream players with BGA systems. However, if there is enough of a grass roots effort led by the likes of @Mr. Fox and @Papusan, there might be enough of those ppl willing to abandon high end BGA gaming laptops. The market could then shift and offer a viable alternative to high-end BGA gaming machines in socketed design. In turn this would help those niche machines become mainstream. That is why I support explaining the limitations of this design to any/all who will listen.

However, with the entire PC industry seeing a decline in overall sales, who's to say long term what the future may hold, but you can plan on me helping the grass roots cause of bringing the BGA gaming crowd into the light.

 

Excellent community service! Very helpful indeed!

 

I do think enthusiasts create markets - pretty demanding markets at that =)

 

OK. Obviously some of my points are flying over many heads and under many radars, so I'll literally put them in points. That way they can be addressed easily as I'd genuinely like to hear back on these and not just dumped with "I don't care about X/Y" as that literally helps nobody. I'm not trying to convert anybody to my way of thinking (it's not a religion, nor is is even remotely life-alteringly important to be frank). At this point I'm mostly curious as to how this kind of thinking developed over the past few years.

1) To set things straight, I did not mean to equate Mr Fox's PoV to a literal hate group or hate speech (as that technically requires it to be targeted at a person or persons). The wording and structure has parallels which for a reader will have implications. I've shown some of this thread to colleagues and we all get the same impressions. Maybe it's just because we're Australian and reading more into it? I know this is an international board so English may not be a first language which complicates things as even written words can carry many implications beyond the literal interpretations of the words.

I'm not particularly sensitive on this matter at all, but the rhetoric could indeed be counter-productive to the entire cause as from where I stand, it seems overloaded with aggression. To say that some of the attitudes here are very hard-line and aggressive is perfectly reasonable to me. Even worse, when any counter-points are brought forward they're just dismissed out of hand or drowned in posts. This is not how you convince people of your cause (rightly or wrongly).

2) The 6700HQ and 6820HK have proven to make for perfectly fine gaming CPUs. For a very long time we've had a plateau in IPC performance. If people are after a gaming experience, then I see no issues here nor do I believe they're getting "duped" or otherwise tricked into getting them.

3) I see many arguments about the fact that the 6700HQ throttles under load. Well of course it does, it says it right in the spec. 2.6ghz base with 5/7/7/9 turbo multipliers and a 45W TDP.

4) Now you may be thinking, it should be able to maintain a 3.1ghz turbo (2.5 + .5 multiplier) across all 4 cores permanently right? No, and I don't know where people began assuming this should be the case.
Turbo Boost by definition is entirely dictated by power, current and thermal limits and applied on top of the base performance (hence a "boost"). They certainly don't owe you that and the ability for a machine to maintain it's turbo should be the mark of a GOOD ODM design and nothing to do with bad CPU design.
I've not seen a recent example of a H series CPU that throttle below it's base clock (which is a REAL problem) outside of software bugs.
This same mentality seems to persist into GPUs as well with Nvidia GPU boost. That being said, NV don't do themselves any favours by making the current Pascal generation vary significantly in clockspeed.

5) On a similar note, I don't see ANY marketing material which appears to claim otherwise. All the marketing material I can see from the big manufacturers always lists base clocks and sometimes boost. For example, all of Clevo's spec sheets don't even actually list boost clocks, only base clocks.

6) Some laptops with the 6700HQ can actually maintain their 4-core boost clocks. Which brings me to a VERY important point.
If the BGA CPU itself is perfectly capable of maintaining it's base AND, boost from a technical perspective, doesn't this put the design failures cases of other BGA-H CPUs squarely on the shoulders of the ODMs? People keep pointing the finger at Intel, but as far as I can tell, the current Skylake generation is one of the best we've had yet with some serious efficiency improvements.

7) As an example, the 2016 Razer Blade (which should be the pinnacle of "BGA Filth" even with soldered RAM) is actually one of few 6700HQ laptops which actually maintains it's 3.1ghz boost on 4 cores (according to the NotebookCheck review). I think they're overpriced as hell, but if you have the money, it's pretty damn compelling to get a <2KG 1060 equipped machine which doesn't throttle the CPU.

8) I think repair/upgrade issues are largely a moot point. It's all about volume and in these cases, replacing motherboards is not a big issue for most manufacturers. Most come with 2-3 year warranties now which is about optimal. Enthusiasts will likely upgrade on or even before those times are up because technology just moves fast enough to justify it. My own laptop history is just such a case, I went from a 8600M (XPS15)->260M (AW15)->660M (W350ET)->860M (W230SS)->970M (P750DM)->1060 (P650RP). Almost every 2 years like clockwork and I'm not even buying into the top end (my income is not THAT disposable as perf/$ generally reduces the higher you go). Each one was pretty much in time with big GPU generational changes as that was the primary driver of my upgrades as a gamer.

9) I think there is a fundamental divide here between people who just want a gaming machine and people who want no compromise performance. From a gaming perspective, a 6700HQ vs a 6700K makes bugger-all difference in most cases as it's mostly about the GPU. This is precisely why Clevo remains the only ODM shipping desktop CPU based models, they're just not popular or really needed for regular gamers. From a performance perspective these 2 can exist as separate entities. To say that one is "trash" and "garbage" is obviously not on coming from a technical level given most of them work as designed, assuming your expectations are set appropriately based on their specification.

10) There's a general apathy here toward lightness and battery power. These are truly valuable features to people and with the current Pascal generation we're finally reaching parity with getting this combined with high end gaming.

11) One interesting parallel happening is with 3-way and 4-way SLI users. In the same way that a mobile HQ CPU + GPU is "enough" for the large majority of gamers, NV decided 2-say SLI was enough for regular desktop users. 3/4-way pretty much only exist for people breaking benchmarks, not practical application. You could argue the same thing about desktop CPUs in laptops. For the majority of users, they would serve no practical application given that gaming generally the use case and we're seeing a large shift into server-based computation for the real heavy-lifting tasks.
It brought about some interesting questions and indeed some reality checks for many people. I can just imagine a board meeting at NV where somebody asks "why do we even support 3 and 4-way SLI any more if <3% of users even use it?" to which the response would be "benchmarks?" and nothing else.

12) I can definitely understand that much of this "hate" was created by the Haswell generation. Haswell had serious issues which I believe Intel really did address in Skylake. One thing that stood out to me, was that a large majority of the Haswell i7 mobile range was released with with a 47W TDP which was just...wrong. Even the base 4700/4710 processors dumped out more than 47W and it was completely unrealistic for the higher end 4900 chips.

Furthermore many HQ processors performed differently to MQ processors. However, this had little to do with the CPUs being BGA or LGA in nature. It was more likely down to binning or other factors simply unknown to us as the mounting method makes little difference. If Intel was using lower binned chips for the HQ series then that's obviously no good but they have to go somewhere right? Would everyone be erroneously directing their anger at socketed CPUs if they were binned the other way around?

13) Finally, I would love to see what a no-compromise notebook truly looks like. Even SocketR (2011-3) if it was possible to handle ~150W cooling. I believe Eurocom have used the "mobile server" moniker before, but that's mostly just in name (really just a P700 with Xeon quad-core instead of i-series).

The Clevo section was a VERY different place years ago and while my input was and remains sparse, I have been around here much longer than many and find more posts about how BAD things are instead of how GOOD they are. Go back in time 5 years and tell somebody we've got near parity in GPUs and CPU performance has basically exceeded gaming requirements and you'd be told you're crazy!

 

Wow. A lot there. I'll share on point 8. I think you're underestating the gravity of that situation, but do admit, my use case is different than the gamer. When there is a problem, with CPU,or soldered GPU or soldered RAM, what is down time? It is critically important that my downtime is limited.

This is a bit contrived, but this summer, I had one of my SO-DIMMs go bad. So, i took it out, ran on half Mem, and ordered a new stick. Now, if this would've been a machine with soldered RAM.... Down time waiting for a board replacement? Unacceptable!

Also, all your eggs are in one basket. Just when you think you got a winner with a great performing CPU or soldered GPU , one little glitch in means you are back at the luck of the draw for new silicon. Not good

Perhaps some don't mind waiting for a repair where the board needs swapped, which takes a lot longer than changing a CPU, GPU, or memory, but for others uptime is of the most utmost importance.

Or perhaps they wouldn't mind getting different silicon, but my guess is these types of problems with a repair which requires an entirely new mobo are more important than stated.

 

Here's one for @SRSR333 since I did not agree with his first car analogy.

When you special order what you expect to be the PC equivalent of a GT500 Shelby that pulls a 11.6 second quarter mile at 125 MPH totally stock off the showroom floor only to find a base model with a V6 "wheezer" shows up in your driveway, it's not a happy day.

Or, how about the tranny craps out in your tree-hugger Toyota Prius turd-mobile and the dealer tells you that you'll have to replace the engine as well (for double or triple the cost) because they are welded together and only serviced as a complete assembly?

 

For downtime, that's certainly true and for things like that I'd certainly prefer to replace my own parts rather than wait for replacement. For the layman gamer who doesn't know what goes where and is liable to break things by opening a laptop up, it's very different. It's just a different mindset.

For business oriented systems where such downtime is unacceptable, often parts are replaced same-day. I know this because it's precisely part of my job. For mission critical systems in our geographical area we can have hardware turn-around within a day. Doesn't matter if it's a CPU, ram, drive, gpu (in some cases with compute clusters) or even an entire server. Costs money though and you'd be surprised at how the desire for 100% uptime evaporates when the bill arrives.

It's just a different world. To be fair, companys like Dell actually offer next-day parts on higher end systems (XPS/Alienware) if you option for the special warranties (came by default a years ago when I had an XPS then an AW15). I actually used it a few times as my XPS was one of the "bad" 8600M GT series (now THERE is a serious HW issue people could get angry about) which burnt out and they had a guy out to my house with a replacement motherboard the next day. The AW15 had a PCH failure late in it's life which got replaced on their next-day warranty as well. It actually still runs to this day.

Each time (motherboard replacement) it took about 30-40 minutes for the tech they sent to replace it and test. It's funny how fast it gets when you do it all day and know exactly what to do. The same tech also services Lenovo in the area and I've had him replace motherboards in various Lenovo machines (including a couple of Carbon X1s which are surprisingly repairable) for clients under similar warranty systems.

As for silicon lottery, that's just a bit of luck really. Technically they don't owe you that and it was luck of the draw in the first place. Could always get a better binned chip as well.

 

Well that's just a straight up mis-representation for a product which I hopefully addressed in my previous post (in relation to cpus/bga/etc).

Good to see your prejudices extend beyond computers.

For the record, I drive a 420rwhp LS2/V8 powered monster, but I'm also not naive enough to call a Prius a tree-huggers turd-mobile given it's an extremely well selling car which suits a purpose. Nor am I going to pretend my car is even remotely practical from a fuel efficiency standpoint. But that's a whole other debate.

 

Perhaps. Perhaps not. Most folks I know are more than capable of changing a GPU or memory. They could easily change a CPU if they only know how easy it was. One other issue I just thought about is the $$$ of a repair, especially out of warranty repairs. When one component goes bad, you'll need a new mobo, CPU, GPU, memory? Then you have to pay for all? Uggh.

For rural areas where I'm located it is definitely not same day, but rather 24 to 72 hours. Like I said, it was a bit contrived, but I was more than capable of making an adjustment and keep my machine running while waiting on a new, easy to change SO-DIMM. This was a lot simpler to band-aid, rather than if I had a computer which couldn't address the problem of a component that needed to be addressed because it was all soldered to the mobo.

Fair point.

 

@Stooj

(1) see https://en.wikipedia.org/wiki/Godwin's_law
EDIT: the version of this law I remember is that when personal comparisons to Hitler or Nazis is used in a discussion thread, the thread will suddenly terminate having degenerated beyond hope of any recovery.

(2) this depends on expectations. Some games and system configurations may run smoothly, but I'm sure you can find a case that won't (when you go to 4K resolution or 120Hz frame rates). The BGA CPUs will be more prone to throttle than there socketed counter parts und1er some situations. I think it's just a matter of understanding and accepting this prior to purchase.

(3) yes, I think there is general agreement that CPUs in general will throttle when they reach their TDP (and maybe under other condition). Those thresholds are simply higher for the desktop CPUs than for the BGA ones. @Mr. Fox also asserts that the binning for BGA CPUs is such that the CPUs end up at the lower end of the performance range which is a way to keep the cost down (less testing required).

(4) I think the point is that the BGA processor is inferior to what you can get with a socketed CPU. The 6700K has a base frequency of 4.0GHz. Quite a bit higher than the base freq of either the 6700HQ or 6820HK (even higher than the stock with turbo freq).

There are certainly appropriate places for the BGA solutions for example for Ultrabooks and other small or low power devices, but a large gaming laptop that has plenty of room for a socketed CPU and adequate cooling should be socketed at a certain price point (and at a certain price point you could justify paying a bit more to get the socketed model if the size and weight requirements are roughly the same).

(5) this misleading marketing material I've seen claims that the BGA systems have the advantage of using less power, generating less heat and using less battery. This is all true. What is misleading is that they don't tell you is that this is the result of running the CPU at a low freq. You can also run the 6700K at a low freq and use much less power and get longer battery life (I think I saw that someone could run the Eurocom Tornado F5 for 4 hours - about as long as I can run my BGA gaming laptop). The comparison is misleading at best.

(6) I don't think anyone is arguing that the BGA CPUs are not running to their specification. It is simply that these specifications are woefully underwhelming compared to the 6700K.

(7) I'm not sure what point your making here. My 6820HK also maintains its 3.6GHz turbo boost (in fact my temps are quite low when doing this and I can undervolt a bit), but that is still much lower than the 4.0GHz base rate of the 4700K. I have to admit I question a little bit my own purchasing decision - except that I'm not sure I could have fit the extra 0.5" of the Tornado F5 - and well I am quite happy with my BGA book (sorry guys - I refuse to call it a turdbook! =P)

The Razor Blade is a perfect example of where BGA is just wrong. The unsuspecting consumer who pays $3.7K ends up with a box that runs about the same as a thin gaming laptop for half that price.

(8) Well, for the poor guy who bought the $3.7K Razor Blade, the mobo replacement would be very expensive... Those 1080s aren't cheap. One would hope it comes with a nice long warranty. I would want more than 3 years at that price (or maybe some upgrade option a that price...).

(9) Yes, I agree with the notion of a divide. On one side you have high end gaming laptops with plenty of room for socketed components and priced accordingly and on the other size you have thin, light, adequate performing BGA gaming systems. I think honestly my own P650RS is somewhere on the border. To me the divide is determing mainly by price. There is a certain price where a BGA system does not make sense and granted that price is different for different people. What I think gets to people here is when BGA systems are clearly crossing over onto the wrong side of the divide and offering poor performing systems at the same price as a much better performing socketed systems would be (i.e. if they offered that instead).

(10) I have to admit I had the same general feeling, but if you think of the design constraints for a thin and light system you realize that dissipating heat is a real problem. This of course means that the BGA CPU and any discrete BGA GPU is going to throttle when the cooling system proves inadequate. The P650's are pretty big and they have very good cooling, but something like an Ultrabook with the same components is going to have real problems. An ultrathin BGA system with a 1060 might make more sense and that is something I would consider if I was restricted to a < 0.7" laptop that I wanted to play some older games on. The key is not paying for a BGA system with expensive components that will never be cooled adequately and will end up performing the same as cheaper components which ultimately equates to a marketing scam.

(11) enthusiasts are always a minority. They are folks looking for systems that meet a very demanding set of requirements. That is on that other side of the divide you were talking about.

The majority however is prone to deceptive marketing - this is where I have a problem. If someone is going to buy a BGA system, they should not pay a lot for components that will never perform as expected (like the 1080 in the new Razor Blade Pro).

(12) I'm also coming in late and I agree that Haswell left a lot of people bitter. But if you look at the high end gaming laptops - the enthusiasts niche - the major players like MSI and Alienware are using BGA boards where they could easily use a socketed system instead. The price and size of the laptops is adquate for a socketed system. This makes no sense.

(13) Well, I'm beginning to think the Eurocom Tornado F5 (MSI something L something) is a much better compromise than my P650RS (though I do like my P650RS quite a lot actually). That is, the F5 has socketed CPU and GPU. While it is thicker and pricier it is still manageable. That system with a 6700K delided with liquid metal would be quite awesome. I think OC-ing to 4.5 or 4.7GHz is not unheard of (if you care about such things), but even the base performance of the CPU is quite a lot better than the BGA counterparts. It is just a much better CPU (and if you get a dud at least you can replace it).

 

If all goes as planned, 7700K and 1080 should soon be available in that little monsterbook for those that want even more performance.

 

Respectfully, that's a straw man response.
Do you understand the point being made?

Most of your points aren't arguments. Merely an essay broken down into paragraphs. You asked for your points to be addressed but failed to respond eg. just above.

Feel vs Fact.



I'm all for discussion but I want to be clear. If reason and evidence hurt feelings it's not my problem.

 

Whoah - I take it the upgrade path from the version with the 6700K and 1070 would be pretty straight forward. That is a really nice thing! (Um unless they have to rework the cooling or what-not).

EDIT: Jeez ... I'll be lucky if I can last a year without wanting to trade up lol

 

Basically the short of it is that the longer a discussion grows, the probability of a comparison (and more specifically equation) with Hitler increases. More specifically that this happens:
https://en.wikipedia.org/wiki/Reductio_ad_Hitlerum
Which I specifically address and more to the point, to address Mr Fox's assertion that expression of hate towards BGA components is "healthy". Boggles my mind given the triviality of the subject a feeling as strong as "hate" should even be bothered with. I think we can all agree that ascribing hate to something as mundane as a product is in fact, not healthy at all. Or maybe I just missed a joke in there somewhere. It's not unlike the pcmasterrace movement where a few take it as license to hate on consoles (and by extension and inevitability, their users).

Agreed. There are definitely some games which will be CPU-bound but I daresay there are very few of them and are an exception rather than the rule. It's about making sure expectations are met. A good example is ARMA 3. I'm convinced we won't have a CPU that runs that damn game properly for 10 years!

Agreed when you compare said mobile CPU with a higher TDP chip. Perhaps a more apt comparison is the i7-6700T, it has very similar performance characteristics to the 6700HQ with a low TDP and has a system designed around it (Clevo N350DV).

Yes that's true. The 6700K base frequency is much higher than many actually, but it also has double the TDP which dictates large amounts of the design. Yes, the HQ processors are "inferior" to a 6700K in that they're just slower, but not in a different way than say a i7-6700T is "inferior" as well. It's just a different tier of CPU. I don't expect a 6700HQ to perform like a 6700K.

No doubt a socket would be preferable at higher price points assuming it's a thick laptop anyway. But that's assuming there's not some other reason to do it.

Have you got an example? All of the spec/blurb pages I find will clearly list the CPU clockspeed. I can't see anything which suggests the comparison between a BGA CPU and a Desktop CPU should be made in the first place either. It's usually just embellished with generic stuff like "excellent performance" or "faster than before" and things like that.

For example, reading the blurb for the Gt83VR-Titan SLI (which is probably the ultimate example of a massive laptop which should just have a desktop processor in it) the CPU information reads "lower power consumption but deliver better performance than before" (before being the 4720HQ explain further in with graphs etc).

As for the battery life thing, it's really more to do with current draw if you want an apples-to-apples comparison. The funny thing about that, is a large amount of the "gap" between BGA laptop power draw and desktop CPU power draw is actually whether the system uses Optimus/iGPU or not. The NV GPU in most cases draws nearly 10W by just being on which is quite significant when you're talking about an Optimus machine idling at 15W and a 6700K machine idling at 25W. I have no doubt if you could utilise the iGPU in the 6700K then machines like the P700 would get pretty close in battery life.

And I believe this is at the very core of this problem. People are setting the benchmark as the 6700K which is a 91W part. Hardly seems like a fair comparison if you want to get objective results.
Other than the number 6700 they're very much different (other than clockspeed, cache is different etc). Maybe it's just the name? Is it really that simple? If the 6700HQ was called a 6620HQ or some smaller number would that be enough to stop the comparison?

And are they really that woefully slow? So slow in comparison as to warrant saying they're junk? In most cases a 6700HQ will bench about 10-20% slower than a stock 6700K. The 6820HK shrinks that gap further.

According to NBC's recent benchmarks of the Aorus X7 DTv6, the 6820HK which comes pre overclocked to 4.0ghz is actually quicker thana their Eurocom 6700K test model. Of course, the 6700K still has great potential, but you've gotta admit that's impressive out of a BGA chip.

This same thing happened with the use of the i5/i7 monikers on ULV CPUs. To this day I think it can mislead people into thinking an i7 ULV should be "fast" because the i7 name has become synonymous with being "top end".

Mostly the same as point 6. But it's just an example of one such machine "doing what it says on the tin" price aside.

I'm referring to the 14" 2016 system with the GTX1060 which currently retails in the US for $1799 for the base model. Even the top 3K/1TB model is $2699 USD. Money aside, as a piece of equipment it gets the job done, if people want to pay extra for it then more power to them.

I'll certainly concede that the Razer warranty is crap (mostly because their service is pretty bad), even on their 1060 system.

It's hard to say how much a motherboard replacement outside of warranty would cost though. For example, I can see a Razer Blade i7-4720hq/960M motherboard on ebay now for $120 but most other manufacturer refurbished ones float around the $500 mark for a similarly spec'd motherboard. The problem is you basically the use case for replacement cost is somewhere beyond 2 years which means you have to "guess" either way.

As another example where there's a socketed CPU in play, rjtech has a W230SS motherboard for $425 and a presumably new i7-4710MQ for $300.

I an see your point about pricing there. We should also not be quick to dismiss the monetary value placed on saving weight

Depends on how you look at it. The LGA1150 socket would be too much for the current larger BGA boards. If there was a Skylake MQ I suspect they would indeed use that in the thicker machines. But for stuff like the GS40/60/70 and other "thin" machines, they simply could not exist with a socketed cpu. Many of those chassis (MSI GS, Gigabyte P34/35, Aorus X3 etc) were designed when Haswell was still in it's first generation and available in sockets, you'd think at least one of them would have done it if it were possible within the height restrictions.

Respectfully, pointing out a straw man is in fact a straw man in itself.

I didn't say they were arguments. I said they were points. Did you read my post? Says right at the top.

An argument can only be made after a point is established, which is what I'm trying to do. That's how a debates and discussions work. First you pose a question, one party makes point A and another challenges that point etc etc.
I establish a point and I'm trying to get others to engage and refute my points with arguments. That way we all learn something from parties on both sides of the fence especially with something so subjective.

That post was also responding to a different point brought up earlier, not by myself. I've addressed anyone who replied directly to my post.

That video is a formless rant about misogyny and not upsetting people? I don't think anyone here is upset are they? Relevance? Now who's straw manning.

 

Yes. If the system had a socket then one could choose to configure the system with a 6700T instead of a 6700K.

I have run out of possible reasons to have BGA in a very expensive high performance laptop that has room for good thermal properties. Even for a very thin laptop one would have to consider the price.

Yes, from Eurocom's site "The on-board, surface-mount design of the CPU and GPU keeps these laptops’ weight down, and allows us to pack more features into a slimmer form factor. The onboard components also minimize power consumption, which means less heat when you’re gaming on the go!". None of that is untrue, but it is certainly spinning lower performance as a benefit.

The 6700K is up to a 91W CPU. It can be run at much lower power levels to "minimize power consumption" etc...

I'm assuming that is a Eurocom that has not been OC'd to 4.7GHz or so... Comparing apples to apples and all that.

Well, if the motherboard has a 1080 on it the cost could run quite high. That part alone is quite expensive.

How so? Early on in this thread there is some discussion of socket real estate. My recollection from tat discussion is that the sockets take up more vertical room but not more horizontal room.

If the box is thick enough then a socketed CPU and GPU shouldn't be a problem. That not to say the heat sink and cooling system design is in any way trivial.

 

<oops - double post sorry!>

 

@FredSRichardson I have to tell you that I get a kick out of your Mugatu avatar - "and your little dog, too" - every time I grace this thread. It's a nice bit of levity here.

 

In big machines definitely, no arguments there. Machines like the P600s begin to straddle the line, particularly the P670. But something around the P650 I think is a great balance and not at all unwieldy. Smaller than that (GS60/GS40, Aorus x3, P34 etc) would be functionally impossible without it. There's also the big issue of how much portability is "worth". There's a point where simply being smaller or more portable creates value in itself which cannot be ignored. The Blade *was* very controversial in this way (they have dropped their prices a bit) but now that every manufacturer has jumped in with competing models we now have a better image of just what that portability is worth to people.

That's a tough one. It doesn't explicitly say anything about performance when you break it down. It says "pack more features" (which I take to mean literally put more hardware in like HDDs/SSDs using said "saved space") and "less heat when you're gaming on the go" which is pretty reasonable. On the flipside I don't think it's unreasonable to assume using less power means less performance though as that's something most enthusiasts will already know (basically the law of conservation). It'll be a cold day in hell before manufacturers start talking negatively about their own products though.

This is certainly true and I did that with my 6700K whenever I ran it on battery. But it would be equally (if not more so) disingenuous to build a system that couldn't handle the entire 91W TDP. An example mentioned being the N350DV. You could theoretically put a 6700K in it but I assume it doesn't have the thermal capacity or power delivery to actually support it.
The 6700K also has the minimum clock being 1600mhz vs 800mhz in the H series (lower voltage as well I think). I believe that's dropped to 400mhz in ULVs. That could very well be a yield problem but in the other direction.

That's true. It's the stock 6700K with 2/0/0/0 multipliers. The Aorus and many other 6820hk systems come pre-overclocked out of the box as a warranted product at 4.0ghz though which is worthy of merit. It simply proves the point that you can purchase a 6820HK machine, preset to 4.0ghz and it'll perform very much like a stock 6700K equipped machine.
I still find it interesting that the 6820hk at 4ghz outpaced the 6700K in the single-thread Cinebench tests which shouldn't happen. Could come down to ram speeds though.

I don't envy anybody who has to replace a 1080, MXM or otherwise. Either way you'd have to sell a kidney. For the time-being it's a non-issue though since you'd assume all 1080s are still under warranty in most countries.

If comparing the older MQ (which were actually PGA and not LGA) to HQ series, the BGA package was significantly higher density (granted, there's lots of reserved I/O there). The actual package dimensions are 37.5mm x 32mm for BGA and 37.5mm x 37.5mm for the PGA MQ. So basically 5.5mm x 37.5mm. When you look at how much is packed onto some of these motherboards, that's not an insignificant amount of space.

For reference the LGA1151 package is also 37.5mm x 37.5mm but the socket itself occupies more space by virtue of it's clamping system (PGA just gets away with a pin clamp which mechanically much smaller).

 

@Stooj - I think the discussion became somewhat fragmented through the bullet points, but I thought of a good example to try to figure out the points we generally agree on.

I'd like to take my current purchasing decision to get the Clevo P650RS BGA system. I am actually quite happy with this laptop, but I do have some slight upgrade envy for the Tornado F5. Here are the main differences between the two:

The F5 is about 0.5" thicker. I would definitely have to buy a new bag...
It is also between 1-2 lbs heavier. That's not a big deal.
I would spend up to $400 more on it. But I might be willing to do that given the tantalizing benefits of the socketed system (and to get the 4700K delided liquid metal applied CPU).

So if you were deciding now and looking at these two systems what would you be thinking? You have compute power and potentially upgradeable components vs a smaller foot print and weight. (Oh yeah, rumor is you can put a 1080 in the F5 - now that's an upgrade).

EDIT: just saw your follow up post - yes all very good points. Unlike me it sounds like you have actual experience with socketed systems. Also good point - if a 1080 parts fails it's no clear how much the relative cost of the board is to that part alone.

EDIT2: I think one service folks like @Papusan and. @Mr. Fox offer is helping n00bs like me navigate the marketing BS. The claim of desktop performance in an 0.7" gaming laptop is ludicrous once you understand the issues but all the marketing material implies just that.

 

Thanks! It took me a while to settle on that one.

And you're right that it may be especially beneficial for rhis thread.

Sent from my SM-G900V using Tapatalk

 

Funnily enough, I've grappled with this very issue a couple of times now and fallen on both sides of the fence.

For buying a new system I weigh things up using a few categories:
Performance
Upgrades/Overclocking (and implied "logevity" I guess)
Weight/Size
Battery Life
Cost

A few of these are subjective or inflexible. For example, cost is usually a fixed "budget" a buyer allows. Weight/Size while largely subjective, but often impacts performance and upgrades. I also consider form factor over weight. e.g. the P650RP is surprisingly heavy at 2.6KG, but since it's dense and has a fairly uniform weight distribution it's not all that unwieldy.
Everyone's use cases are not only different, but also change over time. Either way, I think Cost is probably the biggest driver. 90% of "what laptop should I get" questions usually start with "I have $ABC and want to play games, what do I get?". Anything under $ABC is savings which can go in the bank or toward bigger/more drives or ram etc. Anything over $ABC is pointless since the budget is set or even cause financial harm.

When I bought my P750DM-G I was leaning heavily toward the performance and upgrades side of things. My laptop at the time was the W230SS, so the P750DM-G presented a pretty big jump in both CPU and GPU power. My desktop was already running a 980Ti, so I was willing to take a GPU performance hit (even with my insanity 970M overclock below, not even half the performance of my 980Ti).
Comparing the P650RG/980M of the time and the P750DM-G/970M (both at around $3200) I would lose about 2-hours in practical battery usage, about 1KG extra weight and maybe 15% GPU performance. Alternatively, I could get a P650RE and save myself about $500 which was very compelling since that could be spent on bigger/faster SSDs.

But I'd gain upgrade options, faster processor, Thunderbolt3+USB Type-C and overclocking headroom. With some research I found the 970M should be able to overclock very well and, with the help of Prema's fantastic modded vBios I did exactly that. I actually OC'd it right past the stock 980M comfortably. I couldn't go any further because the power delivery literally wouldn't let me. Here's my final FS on that system if you're interested: link

My use case was that it would also replace my ageing 2600K desktop (with ideally an eGPU enclosure for my 980Ti), which while functionally fine, was a very hot running system. It ran anywhere form 4.2ghz->4.6ghz since I'd downclock it in summer or it'd turn my study into a boiler-room.
My practical use at the time required very little travel, most of the time would be near a power point and I fully intended on having that machine for at least 3 years. The upgrade and overclocking abilities also extended my budget somewhat as the extra cost is offset by the extra longevity (or so I thought).

Things have changed with Pascal though. For less money, my P650RP out-performs my P750DM in every way that matters to me. Since it's more portable I can also find more ways to use it on the go. My i7-6700K that was in my P750DM now powers my desktop paired with the 980Ti as the P750DM is now a dead-end machine and literally a year on from eGPU being announced we still don't have a proper standardised solution.

I think at that mid-high level you're at, with Clevo/Customizable machines at least, there's a significant enough price gap between the P600 series and P700/F5 systems which creates a dilemma for many users. Unfortunately, there's no MSI chassis builders in Australia so I haven't seen/used an F5 specifically but I assume it's just a more plasticky P750 type thing (it's .5KG lighter somehow, which kind of worries me). One thing I noticed, is the F5 has much of the cost of the 1070 built-in to it already. Dropping to a $1060 only saves you $96. Whereas here in Aus, it starts with the 1060 and you have to pay $500 to upgrade to the 1070 which creates that price gap I speak of.

So, the problem comes with what your budget is. Lets say it's $1600 but you could stretch a bit:
P650RS-G with i7-6700HQ and GTX1070 for $1499
or
F5 with a i7-6700K and GTX1070 for $1666 or dropping it to a 1060 puts you down to $1570 so you're in budget at least.

$167 for the F5+1070 isn't insignificant but it's also quite a bit lower. You'll also need to add $67 to get the 330W brick to ensure you actually can upgrade to the 1080 later.

You'd also be crazy to downgrade the F5 to the 1060. At which point the P650RS-G wins simply with the big GPU jump. If gaming is your primary goal there'd be almost 0 downside in that case.

So it really boils down to the P650RS vs F5+1070+330W brick and a difference of $234 which is a pretty big number.

The 1080 would be a good upgrade in 2 years if you can find it at a reasonable price. But what's to say that in 2 years Intel and Nvidia (or AMD for that matter) haven't released new architectures by then which just exceed what you get now. Particularly if you can sell your old system (kinda like trading in your car to recover some costs of replacing it) which is what I do. Laptops are tax deductible where I am so I can write it off and sell it after I'm done.

For most people, "potential to upgrade" is literally a gamble and a hard pill to swallow since it costs you money right now and only *might* save you later. There's no safety like with desktops. Clevo is sure to refresh all the P600s and P700s again with Kaby Lake processors and chipset which may bring other benefits which bring up questions about whether you just replace the whole thing for new stuff. PCIE4.0 is around the corner which could throw another spanner in the works.

All that being said, I just spotted that Eurocom is listing their P650RS with a 120hz 1080p panel option which is mighty enticing if I were you (maybe you can swap for it or something?). If it was G-Sync as well, damn...I'd almost sell my machine right now for that. Might ask my local reseller about it. I've got a 144hz/g-Sync desktop monitor and it is just on a whole other level when it comes to gaming IMO. If you haven't used a 120/144hz monitor I highly recommend it. 60hz almost looks bad to me now.

 

And all brand offering next-day parts and service as Dell/DEllienware? Like OEM's that sell ASUSBOOK, MSIBOOK, ACERBOOK, LENOVOBOOK, HPBOOK, RAZERBOOK, eMACINEBOOK's and so on? What about extended warranty? Is it cheap? Intel BGA chips is much worse binned than LGA chips. Aka much harder to obtain a top noch binned BGA chips vs. LGA. Already at the outset you're screwed.
What with the acquisition of motherboards due faulty soldered hardware Trashware... New or used with just the hardware you need. Change moterboard isn't for everybody. Most people can replace a i7 or MMX graphics. Put all your eggs in the same bag will give an increased risk of smashed eggs, than if they were individually packaged in each box

 

The "big 3" which a Dellienware, HP and Lenovo offer it here (assuming you've got the extended/XPS warranty or whatever they call it now). MSI, Asus and Acer all have service centres you'd have to send your machine to (or Metabox/Clevo for that matter). They may offer similar warranty services in other countries.

As far as cost, Dell and HP include it in some of their higher end models. Usually it costs around $150-200 AUD. Dell's goes as high as 4 years if you want. For popular models on a business level, usually at somewhere around AUD $200/year you can have onsite replacement. Any sign of a problem and they just replace the entire thing and transfer the drives/ram over. I think 3 years is probably the optimum level as usually if something is going to fail, it will either do so before that or long after. Hopefully after 3 years you'd look at a new machine.

As far as binning, it really only matters with overclocking parts. Since the 6700HQ which is ubiquitous in BGA laptops is locked it doesn't really matter. Not sure about the binning methods for the 6820HK though, as they're often sold pre-overclocked at 4ghz which would suggest they definitely use higher bins for those.

I certainly agree that changing a motherboard isn't for everybody. Many of us here would probably do it if we had to, for everyone else I imagine they'd just get somebody else to do it (repair shop etc). If it's under warranty and a relatively painless process then I don't think it's all that much of a problem. Out of warranty is a whole other problem. Fortunately Australia has some very strict laws about warranties. Even if you buy a laptop with 1-year warranty (legal minimum) you can often still claim after that because a computer is reasonably expected to run for 2 years or more. That's assuming it's not a battery or HDD which are known to fail over time.

 

Not all places have cheap support. EG Premium Dell support is not cheap here at home. And most brand use service centres and you have to send in your machine. Dell is probably the only brand with ok support the day after here.

3 years Dell Premium Support on site which is quite normal for people = 581.947 (USD)
If you can not afford to replace your notebook so often as you want and rather choose 1 year extra extended warranty on top... Aka 4-year Premium DEll Support onsite = 1238.671 (USD). Because of the high price on support will a lot of people skip the extended warranty. This can be a costly experience if you own laptops with soldered hardware Repair shops and spareparts as MB isn't cheap either. Nor easy to obtain. It takes a maximum 15-20 min to replace the CPU or graphics card if you have selected the correct machine and have the spareparts. Even if you have 5 tumbs you can do the job yourself Time is money in a high cost country. Downtime can Cost You Money and you risk a lot if you select the wrong machine.

 

My P650RS-G through HID was about $2400 - a bit pricier than from a Sager reseller, but I wanted the CLLU paste job and given the thermal issues people are having with this laptop I'm really glad I went through HID (mine runs quite cool). A pretty loaded configuration of the F5 is also about $2400 for a very similar configuration minus the Evo 500Gb ($170 from Amazon).

So price between the two there is not a big difference and I would have the desktop 6700K in there.

The main cost is that it is significantly thicker (0.5") which would bug me a bit.

 

Well, if you want to delid the CPU and and use CLLU or other liquid metal it could take longer (and with CLLU you are taking some risk of course).

 

Without a big price difference you're really looking at the trade-off between size, desktop CPU perf and a bit of weight. All thing's that are really up to you.

With the BGA bandwagon however, there's been an interesting leak of the Acer Predator 17X with a new Kaby Lake 7820HK processor on Notebookcheck:
http://www.notebookcheck.net/Acer-P...K-Kaby-Lake-specifications-leak.187823.0.html

Yes it's a fat laptop with 1080 and should probably have a desktop 7700K in it instead, but what grabbed my attention is the spec sheet for the 7820HK clockspeeds. It lists a 2.9ghz base and 3.9ghz boost which is a rather big jump on the boost side. Given Skylake and Kaby Lake are pretty much identical in IPC this makes me wonder if they've manage to increase yields significantly in the past year or have better binning procedures.

Would certainly explain why the number of pre-overclocked 4.0ghz 6820hk machines increased recently.

 

Actually the Tornado F5 now offers the kabylake 7700K and 1080 as a configuration. I'm not sure the kabylake CPU is a huge advantage though (I've heard it is essentially pre-overclocked 6700K).

The real advantage of the F5 right now is the possible upgrade paths. One could get it now with a 1070 and 6700K and then upgrade either of those components later on. It's not a cheap option but it's much less than upgrading the whole laptop.

But you are absolutely right. The weight is not a big difference for the F5, but it is 0.5" thicker which unfortunately is a pretty big difference.

The only thing I don't know much about is fan noise for that machine. I guess in low power mode it would be okay. I'm not sure if there is any option to run only on the iGPU - which doesn't make a lot of sense to me.

EDIT: Yeah, that Acer Predator 17X should have socketed components... it is 1.8" thick! And it is offered at 4,000EUR? IMO that is BGA abuse...

 

Just a kind reminder. I am fairly certain notebook manufacturers have little say in the matter regarding socket vs soldered CPUs. Intel now offers only soldered laptop CPUs and manufacturers have to go along with it. I am fairly certain they complained about it just as much as we did. The only solution was for them to adopt desktop CPUs in laptops which is far from an ideal solutions since desktop CPUs take a lot more space, both in area and hight, not to mention the more expensive cooling solution.

If you really want to vent, vent yourself to Intel. They have been in monopoly in the notebook market for so long that they can pretty much throw anything down our throats. Our only hope is AMD. Maybe petition them for socketable laptop CPUs and AMD MXM boards and who knows, maybe they will start to cling back high-end notebook market from Intel and Nvidia.

 

@Blacky - From what I can tell, size is not a problem on these big thick laptops like the Acer Predator 17X. I guess having the option to put in a socketed 6820HK might be nice, but I can't see why anyone would be unhappy with a 6700K. Those can be run at low power if needed. The only thing I see missing is mshybrid or Optimus though I don't think that is a big loss.

Sent from my SM-G900V using Tapatalk

 

Cause it's more expensive to make. I also think you need a different Mobo chipset and you will be missing on some power saving features that laptop CPUs have. Also you need to tinker with it to get a 55W TDP such as the 6820HK. Not saying it's not possible, but most people want things to work out-of-the-box without any tinkering. All in all, I am not saying it's not possible, but Clevo is rather an exception when it comes to desktop CPUs in laptops and we should appreciate them more for that!

 

Hear! Hear!

 

None say you need to buy their trash!! Buy the cheapest JokeBook you can find and Intel and OEM's will receive less money from your wallet. Pretty simple. Vote with your wallet. Only this way can force a change!! To the better for all

 

I'm not sure I buy that. The systems were talking about are already quite expensive. The socketed Tornado F5 is very close in price to my BGA P650RS if you get it with the 6700K and 1070. You can also upgrade that beast to a gtx1080 and 7700K. The only downside I see is that it is 0.5" thicker.

In fact if someone could not travel with an extra 0.5" then the P650RS-G would probably be the best they could do.

Sent from my SM-G900V using Tapatalk

 

So here's a good question for this thread:

If someone cannot travel with a laptop thicker than 0.7" - that is it has to be an Ultrabook - what is the best tradeoff if they want to play some older games?

They are pretty much limited to BGA systems. Should they get one with a discrete 1060? Or should they give up on that idea and get something cheaper and live with iGPU performance?

What would be the maximum price? The razor seems too expensive. There is a Origin EVO15-S for less but you are still talking $1600. For the record, I gamed on my Ultrabook using the iGPU for years though they were very old games

 

I think the GS43VR is a good candidate. Its got lots of power for its weight and price, and still has decent battery life. Thunderbolt 3 is present for possible eGPU, and you can also custom order it from HIDEvolution with CLU / ICD to help with temps and upgrade components, and so they can assist with warranty support if needed. Technically, it even somewhat fits into a $500 limit, if you consider it has a 3 year warranty for $1500 right now (from HIDEvolution) thats essentially $500 / year.

 

Nice reccomendation! I think a razor would $2K or more? I guess for that you get a fashion statement =P

Sent from my SM-G900V using Tapatalk

 

Have you considered the Gigabyte Aero 14? It comes in at $1700, but it's thin, it doesn't throttle, and it has amazing battery life.

 

To be clear, I'm not in the market - this is more of a hypothetical question, but it gets at the root of what different people think is reasonable for price and performance of BGA Ultrabook for the expected performance. For example, hardware that is radically throttled (i.e. the gtx1080 on the RBP) most people would agree is not worth the price, but for everyone there is a threshold.

When I first looked into this I was considering the Aero and Razor and I am thankful someone convinced me to get the P650RS-G instead, but it is quite a bit thicker than an Ultrabook. I don't think a 1070 could function properly without enough room for cooling, but perhaps the 1060 can? Or is even that going to be a marginal performer (though maybe you need a dGPU to render some games...).

 

I take issue with the hypothetical in its entirety sadly.

If I can travel in foreign countries while still attending to my fiance's parents who don't speak a language that I can (english/korean), carrying a full Durango of luggage, and I mean full, barely could see out the back, and still bring the following:

M4600 w/psu
Y510p w/psu
Ge40 w/psu

In one backpack, I just don't see why people "cant" carry a half inch thicker laptop.

More to the point, even the ge40 has a socketed cpu and its thinner than 1 inch.

My own personal opinion of that scenario isn't that someone can't, but won't, or lastly just don't have any real concept of weight when reading it on paper. I'm a fat dude, and enjoy spending my time gaming and though in the past few months I have been going to the gym even before my choice of promoting my health began I had absolutely no issues carrying my m6600/m4600 both filled with hdd's

I understand there is always the exception to the rule, I just dont quite understand the premise in the hypothetical

Just my thoughts, I am biased but am open to a rebuttal

 

Yeah, the notebook is too heavy nonsense has been going on for a long time. I don't get it either. I've never owned a notebook that was "too heavy" so to me it's really an overly-exaggerated non-issue.

The world has more wimps than I ever could have imagined. Or, maybe us fat dudes (yes, me too... and old on top of fat,) take our strength for granted.

 

@TheReceiver. @Mr. Fox
Yeah, this is where my opinion diverges. You may not have constraints on size and/or weight but I do and other people do. With all due respect, asserting that other people should not have these constraints is utter nonsense. I know people that can't carry more than a few pounds with them due to injuries or disabilities. Should we deny them light and/or thin gaming hardware?

The real issue is where is the cost threshold if one accepts all the limitations (i.e. buying new every 3 or 4 years and accepting low performance, what is a good price threshold?) I think $1200 is perhapa on the low end, if the is a dGPU.

EDIT: BTW if there is a socketed system that meets these constraints then that is a solution.

Sent from my SM-G900V using Tapatalk

 

Let me clarify since I did not already. Huge high performance laptops are not "too heavy" like some people say they are. If they were not so big and heavy they would malfunction. I do understand some people actually need something abnormally small and light for a specific business purpose, or because of a disability, and those can be tasks that the bigger laptops are not well suited for. Nothing wrong with that. But, it still doesn't mean that monsterbooks are "too heavy" or "too big" and I like them because they are. Where things get sideways is trying to use the wrong tool for the job and expecting a wimpy turdbook to run like a full-sized high performance machine is not a realistic expectation. Expecting a real high performance notebook to be thin and light is an equally unrealistic expectation. People that don't have a specific need for compromised performance with something thin and light and maintain that other machines are too big or too heavy are the wimps.

 

@Mr. Fox. Okay, we are again on the same page.

I mean specifically.for the niche of users who are legitimately size or weight constrained down to an Ultrabook and they would pay a bit more to play some games. I think for that group even there is a price beyond which they see no gain in performance and a price where they pay too much for a little better or inconsistent performance . I'm guessing that is above $1400 but it may be lower especially if you buy the previous generation.

Sent from my SM-G900V using Tapatalk