BGA vs PGA continuation

(Note: This was the best starting point I could find for this conversation so this is a thread split, hence why the first post doesn't make sense - Ethrem)

I was most shaken of the heavy throttling from the processor in 3Dmark11 combined test... I wondered if I had seen right I almost forgot that this was the newest BGA high end processor from Intel packed into the new modern thin gaming laptop from Dell

 

Just wanting to answer, yes. Or at least it should be a yes. The A02 wasn't released when Mr.Fox did his benching or the 15 and 17 R2, and even then, he did discover the wattage limiter (wattage limited by which PSU is detected at boot) the two laptops had, which then lead to the A02 BIOS to "fix" the problem.

 

↑
@Mr.Fox, I guess it was the 1st BIOS right? A00??
Game7a1 Just wanting to answer, yes. Or at least it should be a yes. The A02 wasn't released when Mr.Fox did his benching or the 15 and 17 R2, and even then, he did discover the wattage limiter (wattage limited by which PSU is detected at boot) the two laptops had, which then lead to the A02 BIOS to "fix" the problem.

Pretty amazing that Dell's engineers failed to perform such mandatory testing-benching as Mr Fox did with Dell's new gaming laptop they had just developed. What if no one had told about such an issue? Had Dell engineers have managed this? Most likely NO...

 

Most likely the reputable dell engineers who know what they are doing (or compal engineers) resigned from their jobs due to self respect when alienware 13/15/17 will be BGA only.

 

Should be exciting to see the functionality of new bios updates to AW15 og17R2 by Dell engineers... Maybe a new bios allows all hardware becomes even hotter with an even poorer fan profile. This is poorly workmanship. Seems this is beginning to be a farce...

 

LOL nobody "resigns due to self-respect." Given the current economy and job market, nobody with any amount of self-respect gives up a perfectly good job because of a few enthusiasts in a tech forum or some similarly stupid reason. If it did indeed happen, more than likely they were laid off or reassigned when Dell went private.

 

I think Notebookcheck.net should test AW15 and 17R2 all over again. They had probably got the same result with throttled processor down to 0.8-2.8ghz or/and battery discharges.

 

You know you're right on this... that would be a great example of cutting off your nose to spite your face. It's very difficult to understand, but some people actually think the direction things are headed is a good thing. Otherwise, nobody would be buying thin and light laptops with BGA components. It should come as no surprise that some of us hate it. When it comes to taking home the bacon you just suck it up, put your head down, and do whatever you need to do... like it, or not.

 

The difference between a socket laptop processor you can overclock in a thick laptop vs the new modern Intel BGA processor intended for thin laptop's in Cinebench R15...The numbers speak for themselves
Socket laptop processor vs a whole bunch BGA garbage

 

The only people who like BGA are the ones who have been convinced that thin and light laptops aren't possible otherwise (a socket isn't so large to make that statement accurate), the ones who have no idea the compromise that comes with them, and finally those that benefit financially from the change. Intel can now sell garbage silicon and blame the throttle on TDP limits that manufacturers wanted. It's a mass brainwashing campaign to give Intel even deeper pockets and there's nothing that is going to stop their domination. AMD can toss all the cores they want into a desktop Zen chip but not so much in the laptop space which means Intel remains king.

 

@Ethrem You have very very right about this. See what Asus is doing to throttle processors from becoming hot due to "thin" laptop. Tragic...

http://www.notebookcheck.net/Asus-G501JW-Notebook-Review.141745.0.html

 

Papusan, you are comparing an $1096 processor against a $378 processor, a difference of $718. About 2.9x more expensive.

If we are talking about cost/performance, and I am getting a 675 score from my 4700HQ running at 3.6x/3.5x/3.4x/3.4x Cores.

You would need to be getting a score of 2000 to scale cost/performance 1:1.

At 852/675 x $378 that would be an actual value of $477.

So you paid 2.30x what the added performance allows.

Putting things in perspective it really isn't a matter of Socket vs BGA, it is a matter of $'s.

Of the available choices, the socketed CPU laptops are just too expensive. Configuring any of them for my needs would generate a cost of over $4000 - $5000.

Having paid $2300 for a G750JH, it does what I need for 1/2 the cost.

When the G750JH isn't doing what I need, I recover some value, or pass it on to some deserving person, and buy whatever is current - much faster and hopefully cheaper.

For most people spending less money, getting value for their $ invested is more important in the short term, and the long run.

Maybe if the Vendors making/selling socketed CPU's had kept the prices competitive, things would have turned out much differently?

I don't think we can lay all the blame all on Intel. Intel provide similar cost / performing CPU's in both FCBGA and FCPGA.

The first MQ processors didn't have VM bits enabled / presented in the BIOS of most of the first Haswell laptops, and the Asus G750 was the first one with HQ processors that did have all the VM bits available - it really was the only choice for hosting VM's.

BTW, why is it that the FCBGA gets 1364 pins, and the FCPGA gets 946 pins?

Compare Intel Products - 4700HQ vs 4930MX
http://ark.intel.com/compare/75116,75133

 

I paid the same as a i7-4900mq The problem with BGA is that you can not choose what you want to buy or upgrade later. Freedom of choice. You can get an MX processor for just under $ 600 if you want to upgrade. But however; you can not do that, as you have a BGA processor..... What do people when the processor / graphics card breaks down after the warranty? Buy a new processor / graphics card or buying a brand new laptop? Cheapest option is to buy a new motherboard with all hardware soldered on to a costly price. Freedom of choice
Edit: You said If we are talking about cost/performance, and I am getting a 690 score from my 4700HQ running at 3.6x/3.5x/3.4x/3.4x Cores.
Not everyone with a Bga processor manages this(your) score because of processor throttling when the temperature increases or get the max tdp. Because this garbage is in a thin laptop.
Some choose the mother while others choose the daughter. Such is freedom of choice.

 

@hmscott can your 4700HQ maintain full Turbo Boost beyond TDP like MQ/MX chips can? Well there's your answer re perf/price.

 

@hmscott Not everyone manages 675 cb with Bga processor in Cinebench R15 during a good day. Maybe from 490cb in Cinebench R15

 

octiceps, neither can most MX/MQ laptop installations, there isn't adequate power or cooling

Back to the CPU cost / value performance equation, an $1100 4930MX processor isn't something I would choose to upgrade to from a 4700MQ, if I did have a socket.

There is likely a better choice along the curve of price vs. performance.

The platform would need to be much better than those available right now, or recently, to really take advantage of the full power, don't you think?

 

Last I ran R15 I only got 675, so I went back and redid the numbers for you at 675 vs 690.

Not much improvement really...

 

Next time maybe this score .. Or even worse

 

Shall we question your score as well?

I used XTU and Asus GPU Tweak to optimize performance. Most people won't.

You are making noise, but no improvement in your position

 

Is not help by owning a more powerful Bga processor @hmscott You are lucky that manage such a Cinebench R15 score with your BGA processor

 

Papusan, again most people don't know to use XTU to set multipliers to maximum. Those scores can be much higher if they tuned their CPU/GPU.

 

Power and cooling are variables which can be controlled. HQ CPU's locked TDP limit is a constant which cannot be controlled.

A 4910MQ can be overclocked to 4.3 GHz across all cores and is significantly less expensive.

 

Xtu and Trottlestop is very known software. When people see that they have throttle problems on the processor, so they learn quickly to use this software However Bga processors are not particularly well suited The best part is that you say "overclock both". As Winnie the Pooh says.

 

Papusan, your numbers are based on XTU tuning of your processor and tuning of your GPU - you don't run at stock do you?

It is an unfair comparison against other processor scores unless they have also tuned their CPU/GPU.

You are fooling yourself if you think that 490 is the R15 score for a 4700HQ, I just showed you it is a 675 for a run on a 4700HQ at maximum multipliers unlocked with XTU.

The comparison still stands. Your score improvement is inadequate to the expense, and BGA processors are a better value and provide great cost / performance.

 

I haven't seen laptops magically get cheaper since they moved to BGA CPUs. Which makes the "better value" argument a hard sell since BGA is already limited by its crappy power restrictions which reduce performance compared to PGA, e.g. 4700MQ > 4700HQ.

 

octiceps, the platform fixes the cooling / power to static configuration. You can repaste, improve power by getting a higher wattage adapter, if one is available.

For the most part the only thing you can vary is the socketed CPU / GPU - the upgrades can come with larger heatpipes and fans, but still you are pretty much not going to get more heat handling out of the same frame. If you do you push it to get more airflow, you are going to get more noise.

That is why the 880m had so many problems. It went past the heat handling threshold of the frame.

Just like the 4940MX heat problems. You just squeaked in under the TDP wire with that 4930MX.

I am looking forward seeing the disclosure of the 4910MQ tweaks that Dufus made to get that performance

 

The point is that you can overclock socket laptop processors. 490 cb Cinebenchmark R15 scores are from a more powerful processor than what you have. Did you read the link in the previous post? The test are from the more powerful Intel Core i7-4720HQ...
http://www.notebookcheck.net/Asus-G501JW-Notebook-Review.141745.0.html

 

octiceps, go back and re-read my first post, the numbers are there.

The BGA laptops didn't get cheaper, the PGA laptops and CPU's got unreasonably expensive.

 

I agree, I wouldn't upgrade from a 4700MQ to a 4930MX either, the cost difference is huge! The performance difference is not that much different either because it's hard to cool down an overclocked 4930MX. But, other people in this thread have said that the problem with the HQ versions - you can't unlock the max TDP of those chips, so they throttle above a certain load, which I imagine could prevent full use of the turbo bins during gaming if the CPU is pushed. What kind of TDP can you sustain on your 4700HQ? Do you maintain full turbo during all your gaming, are they demanding CPU games?

 

octiceps, yes, and the processor is detuned from maximum performance as shipped by Intel, they need to use XTU to increase the performance. You do know that, right?

Just because the potential is there, it doesn't mean they are going to see it.

The 675cb score I got is the same others get that use XTU / Tweak tool to tune their CPU/GPU, I have helped lots of people do that on the ROG forums.

Maybe you don't know that.

https://rog.asus.com/forum/member.php?131395-hmscott

 

Run a Wprime 1024 (with the 8 threads) ver 1.55 test also when you get time.
http://hwbot.org/submission/2849851_papusan_wprime___1024m_core_i7_4930mx_2min_47sec_657ms

 

Exactly my point. You can improve cooling and get a bigger PSU so your MQ/MX can draw more power under load to maintain stock or OC clocks. But since BGA has locked TDP limits, you can't make your CPU draw more power even when it needs it, thus it throttles, even below stock TB frequency.

4700MQ actually.

But Dufus was only able to accomplish such a massive overclock by increasing TDP significantly, something which you can't do with your HQ. So at the end of this day, it doesn't really matter.

Those numbers are irrelevant to me and besides, Extreme Edition has always been a halo product and commanded $1000 since the days of Pentium 4. That's why I said I would get a 4910MQ instead.

Where is the proof that PGA laptops and CPUs got unreasonably expensive?

OK let's cut the crap. If you can show your 4700HQ running Prime95 on 8 threads @ 3.4 GHz for longer than 5 min, I'm all eyes and ears.

 

FWIW this was my best CineBench R15 score when I had the opportunity to play around with a 4700MQ.



Note I was using 2133 ram with decent timings, so that helped the score a bit as well. Probably could've squeezed past 700 cb had I bothered to tweak the BCLK. (I know for sure it was bench stable at 100.5MHz)

 

n=1, 695cb for a 4700MQ is awesome, and likely if I had 2133mhz memory my 4700HQ would get a similar score.

When I compared benchmarks against an MSI 4700MQ vs the 4700HQ I got the same performance for both. The only reason I couldn't keep the 4700MQ at the time is that the BIOS didn't offer VM options.
.

 

For your HQ processor what TDP can you maintain? For instance under Prime95? Would also be interesting to know what CPU clock you can maintain when gaming on CPU intensive games? This would provide some balance.

 

As Octiceps said, this is utterly pointless, since even if you attached a 330W PSU to those things, the CPU is going to stick at 47W and all the extra power is wasted

LOL. No. Not even close. The 880M had other heating issues and power issues and voltage issues and throttling issues not present in any other product prior to it. I can clock my 780Ms RIGHT NOW to 1006/6000, beating an 880M's optimum operating efficiency at stock, and I won't overheat or throttle like that card does. End of story.

The 4930MX and 4940MX had heat problems for two reasons: 1 - On the whole, haswell chips were hotter than SB and IB chips. Systems were designed for IB type of TDP, as intel specified the same level of TDP for the chips... actually they specified 2W more of TDP. No need to change heatsink design for that. What happened was the chips were RIDICULOUSLY hotter. Alienware machines still cooled them just fine, mind you. Clevo's and MSI's at-the-time barely passable IB cooling was not much improved at all for Haswell, and it showed. Clevo fixed it a little with the SM-A series in the single GPU models (vent layouts, making better heatsinks which remove tape mod requirements, etc) and MSI attempted to share the heat to the GPU (which failed as they paired it with the 880M... smart people, MSI, I promise) and then they discarded the models and properly overhauled the cooling into the GT72/GT80... which for all intents and purposes are cooled amazingly, and are severely held back by their TDP-locked mess of CPU configurations.

 

Drowning out the original point doesn't make it any less true.

Bringing up endless comparisons of limitations between an old low end processor and a newer or high end processor doesn't make the original point any less true.

The newer processors in BGA also have more tuning options available in XTU, including Power Time Windows and TDP. I don't have a screenshot handy, but they exist in the ROG forums.

Papusan, mouthed off that BGA was "Garbage" and gave his benchmark results in R15 as the example.

Comparing his $8k 18" laptop against 15" slim laptop benchmarks as justification as to why BGA was crap was silly, I called him on it.

If you think Papusan picked a poor example, that's fine, I am not going to endlessly compare other benchmarks to continue to prove the same point I already made.

BGA and PGA are form factors, features and tunable options are different and more available in both form factors as the processor release dates are newer and further up the cost chain.

A PGA processor in a 15" slim notebook wouldn't make any sense. Neither would a BGA processor make sense in an 18" highly optioned laptop.

There is no need for trashing one or the other, they are both proudly purchased, owned, maintained, and enjoyed by their respective owners.

Please start showing more respect for your fellow members.

End of Discussion.

Papusan, you are comparing an $1096 processor against a $378 processor, a difference of $718. About 2.9x more expensive.

If we are talking about cost/performance, and I am getting a 675 score from my 4700HQ running at 3.6x/3.5x/3.4x/3.4x Cores.

You would need to be getting a score of 2000 to scale cost/performance 1:1.

At 852/675 x $378 that would be an actual value of $477.

So you paid 2.30x what the added performance allows.

Putting things in perspective it really isn't a matter of Socket vs BGA, it is a matter of $'s.

Of the available choices, the socketed CPU laptops are just too expensive. Configuring any of them for my needs would generate a cost of over $4000 - $5000.

Having paid $2300 for a G750JH, it does what I need for 1/2 the cost.

When the G750JH isn't doing what I need, I recover some value, or pass it on to some deserving person, and buy whatever is current - much faster and hopefully cheaper.

For most people spending less money, getting value for their $ invested is more important in the short term, and the long run.

Maybe if the Vendors making/selling socketed CPU's had kept the prices competitive, things would have turned out much differently?

I don't think we can lay all the blame all on Intel. Intel provide similar cost / performing CPU's in both FCBGA and FCPGA.

The first MQ processors didn't have VM bits enabled / presented in the BIOS of most of the first Haswell laptops, and the Asus G750 was the first one with HQ processors that did have all the VM bits available - it really was the only choice for hosting VM's.

BTW, why is it that the FCBGA gets 1364 pins, and the FCPGA gets 946 pins?

Compare Intel Products - 4700HQ vs 4930MX
http://ark.intel.com/compare/75116,75133

 

PGA CPUs have these options too. The difference is that on PGA they actually work.

His laptop is neither eight grand nor 18". Exaggeration is doing you no favors.

BTW take a chill pill, you're taking this way too personally...

 

They also work on the BGA processors, having seen it work I can only assume you haven't actually tried it for yourself?

Thanks for correcting me, I assumed he had an Alienware 18" SLI. If he blew $1100 for a 4930MX I assumed it was something like that.

Here are the specs, sorry I didn't look. So it was likely at $4k-5k laptop instead of an $8k laptop? $5k was my original number in my first post.

Alienware 17r-1 Ranger
[email protected] (With Liquid ultra)
gtx780m@slv7 vbios
32gb ram
512gb Samsung 850 pro
256gb ocz vector
64gb m-sata
120hz 3d monitor
Win 8.1

Of course people take things personally, that is why I called out Papusan for trashing BGA's, given any comparison it's BS, but given the particular off balance comparison it needed calling out.

Please stop doing that

 

My laptop is a 17. I paid under 3900$ for my Aw17 +I Buy 1 pcs 850 Pro ssd...

 

Papusan, nice laptop.

Now please stop trashing other peoples choices for laptops

BGA CPU's aren't trash, they work quite well.

 

I have a 4720HQ inside a P650SG. But like I said, increase the two TB power settings to something like 65W/80W, run Prime95 on 8 threads, and see if it holds steady at 3.4 GHz.

Read. He didn't pay full price. He said he spent as much as a 4900MQ which is less than $600.

 

I don't trash other people's choice. I'm only pointed that BGA processors is trash. 2 different things... See under my avatar... This is my own opinion..

 

octiceps, it isn't the CPU's fault the platform it was built into doesn't provide enough power or cooling resources.

I actually see the whole complaint more of a size/power/cooling platform complaint than a BGA/PGA problem.

The BGA's are often put into a smaller lower power less robust cooling platform, and that then gets compared against PGA's in large frame 17" or larger with higher resource limits available.

If you put a PGA in that P650SG it would like perform the same, given the same power and cooling available to the BGA.

It doesn't matter how good of deal he got on his particular CPU. I was comparing CPU retail msrp's listed on the ark.intel.com site - a normal and completely valid metric.

 

Papusan, please stop trashing other peoples choices for laptops.

BGA CPU's aren't trash, they work quite well.

Now you have my request and my opinion

Update: thanks for the rep points Papusan.

 

The golden test of whether a CPU will pop its TDP cap is to run the wPrime 1024M benchmark on all 4 cores. If the TDP is locked down internally the CPU will stop boosting somewhere around the 20-30s mark. I'm pretty sure I posted a few screenshots of my 4900MQ holding steady at 4.2GHz throughout the entire run, running at around 65-70W TDP and not throttling once.

 

Why are you beating around the bush?

 

The most BGA processors I have seen Trottle in Wprime 1024. What score you get with your 4900 in Wprime?

 

Yeah good point. On hindsight recommending Prime95 28.5 to someone with a mobile Haswell probably isn't a good idea, else I'd be an arsonist.

 

My 4700HQ is locked at 28 seconds, I can't change it, newer HQ processors default to 8 seconds but it is adjustable, and they allow many additional tweaks. Someone posted a newer XTU controls view recently, but I can't find it, yet.