Intel Core i7-8750H/ i7-8850H/ i9-8950H Coffee Lake

Putting a pad under the NVMe SSD fixed it. All is well.

 

I have somewhat mixed feelings about Intel, NVIDIA and AMD dictating what the OEMs have to do. This is somewhat like the notion of trying to mandate and regulate things such as morality and values. When mere compliance is the end game, excellence is almost never achieved. They should do what is required without it being required. Some would not do it even if it were required. But, in terms of achieving the same desirable outcome I agree philosophically. The problem is that very few--few as in none--of the notebook manufacturers seem to be particularly concerned about quality and delivering full performance from the products they sell.

The bolded part is just common sense, which seems to be something none of them possess. The fact that we see it happening simply proves they don't give a rat's ass about functionality.

A notebook with a powerful CPU that has been crippled by firmware and inadequate thermal management and overall shoddy engineering is about as worthless as the mammary glands on a boar. That is actually more sinful than mediocre build quality. Excellent build quality is basically worthless when the machine malfunctions and under-performs. I think there are a lot of folks that would wink at mediocre or average build quality if what they purchased actually functioned correctly. Price seems to have less influence on the outcome than it should. Poor results plague the notebook industry in all price ranges. Claiming that it "functions as intended" when the outcome is clearly crappy is just an excuse. Like my daddy used to say, "Son, an excuse is just a reason surrounded by a pack of lies."

 

Ugh, another one. i9 laptop that under all core boost can't even reach the all core boost of the i7-8750h due to firmware power throttle. Overclocking the K SKU CPU literally had no effect on performance.
At least because of the constant power gimp it runs cool... yay?

 

Very nice. Why not try -0.200mv for the 5GHz all cores Cpu OC?



Same DELL model but with the cheaper 6 core i7. Same tragedy.
Dell G7 15 (i7-8750H - GTX 1060 Max-Q) Laptop Review

 

Oh come on guys... It's so cute, and it is such a pretty color of blue. And, it has Max-Q, for Pete's sake. Is the excitement giving you goose bumps yet?

If I had one, sticking with the trendy Spanish theme, like El Cazador and Machete, I could name it El Vómito Pocito.

 

Don't mention vomit! You don't want @/Papusan to catch wind of it.

I have yet to have seen a laptop with the new 6C/12T mobile CPU's that isn't entirely crippled in some silly way or other.
There probably is one, but have yet to have seen it. 5GHZ in a laptop with a mobile cpu is impressive if it can keep itself cool.
Annoying that OEM's try to get a 4GHZ 6C/12T CPU to fit in a 45W TDP

 

I was specifically thinking about Brother @Papusan when I invented that special nickname for it. Anything that ends in "ito" has got to be cute, even if it is truly disgusting.

Speaking of puke, this is fitting for the G7 el cheapo 'throttlemaster' gaming turdbook.

 

I've mentioned it elsewhere, lemme just slot it in here:

Might be nice to know that the Precisions are amongst the highest stock clockers (reach 3.6 GHz, break 1200+ CB15 point barrier, 1300 in the case of the 7730) of the 8750H, 8850H, Xeon E-2176M, Xeon E-2186M and 8950HK. The 8950HK in one case actually runs at 4.15 GHz on all 6 cores (150 MHz away from stock all-core Turbo clocks) after a 100 mV undervolt, and no other cooling mod, including thermal paste.

Of course, it's no P870DM, but it's decent enough for me, it's decent enough for a chassis half as thick and half as heavy

Sent from my E5663 using Tapatalk

 

Thats still just kind of meh

The 8950HK was advertised as being much more capable, but I guess I really shoudln't be complaining

Of course its not P870DM, but I probably won't know what that feels like, mine was DOA and is getting send back the seller right now

 

Yeah and the laptop manufacturers continue... Whats the point with warnings ( It’s time to call out laptop manufacturers for their ******** CPU throttling) and posts that nobody is interested in follow up?

Gamescom 2018 | Next generation Asus Zephyrus S GX531 will be even thinner than the Razer Blade 15
The new "thinnest gaming laptop in the world" is just 15.75 mm thick at its thickest point to be even thinner than last year's ROG GX501. The GX531 introduces narrow bezels and 144 Hz displays to the Zephyrus family with plenty of...

If it's Cute... It's Disgusting.


Why give a score of 88% if the performance and cooling is heavly crippled? The reviewers push people aka the buyers into Trash!!

Lenovo ThinkPad P52 (i7-8750H - P1000) Workstation Review

Processor
Intel’s Core i7-8750H is a current generation Coffee Lake hexa-core CPU manufactured in the 14nm++ process with a TDP of 45 W. As such, it is perfectly suited for workstations and gaming notebooks. According to Intel, maximum turbo boost for all six cores is 4.1 GHz while the base clock is at 2.2 GHz.

In reality, Intel’s claims have to be taken with a grain of salt. With all four cores running at 4.1 GHz the CPU would require much more energy than just 45 W. Even just approximating the maximum turbo clock speed results in much higher power consumption, which is why Lenovo enabled a maximum of 78 W for a few seconds. In this configuration, all cores started out at 3.9 GHz when running our Cinebench R15 multi-thread loop until core temperatures reached 97 °C, and then subsequently dropped to 3.4 GHz and a TDP of 58 W.

During subsequent iterations of Cinebench R15 CPU temperatures remained around 97 °C while clock speeds slowly declined even further. They eventually settled at 3.1 GHz and a TDP of around 48 W. Consequently, the cooling system was running at its full capacity already and the faster Core i7-8850H only makes sense if you need a higher single-core performance or require the better GPUs. Considering the fact that the Dell XPS 15 also maxed out its cooling system with the Core i7-8750H already the ThinkPad P52 is in good company.


Temperature
Initial frequencies of 3.6 GHz at a TDP of 56 W resulted in core temperatures of up to 97 °C, and the system started to throttle after around a minute. Instead of 56 W the CPU’s thermal envelope was reduced to 35 W and clock speeds of 3.2 GHz. Temperatures oscillated around 90 °C for the next 10 minutes, after which clock speeds were reduced further to between 2.8-3.1 GHz. Temperatures eventually settled at around 95 °C.

The reason for this rise in CPU temperature at the end can be found in the P52’s cooling system: instead of two independent cooling circuits for CPU and GPU the P52 features a single connected cooling system for both. And since the GPU is prioritized over the CPU and its clock speeds are not reduced the CPU must throttle down accordingly. The GPU started out at around 65 °C and rose to 70 °C over the course of the stress test. Combined stress was simply too much for the combined cooling circuit and CPU clock speeds had to be reduced in order to prevent overheating. However it is worth noting that the CPU never clocked below its base frequency of 2.2 GHz, and thus did not throttle in the truest sense of the word. Keep in mind though that our review unit was equipped with the entry-level CPU and GPU and the more powerful components might bring the cooling system to its knees much faster.

Verdict
The revised cooling system and the P52’s overall performance can best be described as a double-edged sword. On the one hand CPU performance has been improved dramatically... On the other hand the cooling system was already overtaxed with our Core i7-8750H and an Quadro P1000 equipped entry-level SKU. This is as much Lenovo’s fault as it is Intel’s - almost no hexa-core-equipped notebook is capable of fully utilizing Intel’s latest chips to their full potential. DON'T PUT THE BLAME FOR TRASH COOLING AND THINNER AND THINNER CHASSIS ON INTEL!!

 

88% - because its just as brown and smelly as the other turds. Gold stars for all!

This is the ACCEPTANCE OF MEDIOCRITY that @Mr. Fox mentions all the time

Notebookcheck ratings have been stupid for a very long time now.

 

So, how does such a badly throttling system get such a high rating?

Whatever rosy glasses the reviewer had put on, I want some too. It's plain mind-numbing.

 

It's called Group-Think like being in the Matrix. Many people don't have a clue that they are settling for mediocrity. Fanboys and the hollow-promise of a better tomorrow make convincing arguments. I am reminded of the old motto, "Too cheap to meter." Turns out, that wasn't really true.

The moment you take the blue pill (or is it red?), and you wake up to find we are nothing more than batteries...

 

People buy after seeing "nice" recommendations (I call it fake reviews) who showing high verdi/value total score and not how the systems work. The reviewers out there are one of the main culprits for the sad situation we have jumped into in the notebook land today.

 

Oh, the P52 was dismissed as not having a good thermal dissipation system by @yrekabakery easily a couple of months ago, NotebookCheck just confirmed it. Half a heatpipe for the CPU... What were they thinking?

Sent from my E5663 using Tapatalk

 

Nothing like trash journalism. Reviewers know that they will be blackballed by the vendor if they paint the reviewing sample in anything but the best light. Why is it everytime I do a search for best gaming laptop, I always find the Alienware 17R5 included, and nowhere in the review is anything hinting that thermals are a problem. Heavy. Pricey. Look at this load of crap. The uninformed believe this because why would journalists deceive?

 

Post #1000 IS IT POSSIBLE GIVE ZERO STARS?

 

The fact that such blatant lies are perpetuated by debase people that know they are deceiving the gullible is very disheartening. The notion that some might actually believe the reviews that they have authored gives a whole new meaning to professional incompetence. It's probably safe to say that some of these Facebook jockeys are legends in their own minds. Monkey see; monkey do.

Even if you could give negative star counts, a multitude of muppets that believe everything they read would remain. Seeing a product that is defective by design earning a 5 star rating from those same muppets destroys the credibility of a customer review system. People don't know what they don't know, and many lack the technical knowledge needed to recognize that the product they have purchased, and are infatuated by, is malfunctioning out-of-the-box. At least those with ordinary intelligence can have pity on their ignorance rather than contempt for their lies and treachery.

 

AW17 R5 i9 is limited to 110w power draw (reality around 95w package due to other factors, even if temp is under 75c) short, and 60w package long / indefinite.

Adjustment of PL1 and PL2 is not possible, it's locked in the BIOS. Adjustment using XTU will fail or not apply at all.

The weird thing is that PL2 doesn't apply (bios + XTU shows 110, but reality is 60w package limit).




The XMP is also not working at all even though it's displayed in XTU, my 3200MHz G.Skill kit still runs at 2400C18 on the machine. Adjustment of ANYTHING in the RAM section requires a reboot and will always result in a failed attempt (changes for memory couldn't be applied).



On non AVX loads and with good thermal paste (I have LM applied right now) I can see that the maximum sustainable clockspeed with max fans is around 4.3-4.4GHz.



Edit: also this sample has the ESC key area sink down for whatever reason. Causing bad keyboard feel and weird sound from the keys.

 

Thank you for sharing the sad experience with their latest iteration of defective trash. Those CPU clock speeds were impressive 10 years ago, without all of the firmware throttling, thermal engineering deficits and other defects. Maybe some think having two extra cores throttling makes up for the rest of the mess.

That's how the monkeys at Dell roll now, so it is a self-fulfilling prophecy. Why spend extra for an Alienware when you can ultimately have the same broken garbage experience for a lot less money in a G7, and still get lots of "likes" from your friends on Facebook.

 

For the price that they're asking for, it's asinine.

 

For the Q6, I'll also be "downgrading" the current cooling system ( a fused GPU+CPU heatsink) to that of the older Q5 (separate GPU+CPU heatsinks).
Then I'll have an extra heatsink as a backup if that backfires.

I believe that the fused heatsink assembly is detrimental to the already crippled cooling in this notebook.

Key reasons being that a GPU stress test in this state steadily raises the CPU temperatures, and that should not be the case.

Even on the older MSI GS43VR with a separate assembly, stress testing the GPU never resulted in temp spikes for the CPU.

 

Yup, I agree with you. A unified heat sink is a defective engineering concept. Having the GPU and GPU puking on one another is not a good approach. The idea of "balancing" the heat distribution seems logical, but in practical application the execution leaves something to be desired. It makes mitigating high temperatures on one component more difficult because you are having to deal with it on both, and fitting both components well is more complicated. Nobody manufacturers notebooks with fit and finish to tolerances exacting and as precise as that would require to work well. The other part of the design concept related to keeping the thermal solution as compact as possible to facilitate a thin and light chassis merely underscores it's stupidity. The whole package is generally too small to work well because the chassis doesn't have enough room to accommodate anything awesome.

 

And there's the reason why the issue of poor heatsink contact (with the tripod mount) gets EVEN WORSE.

And related

There's the Q8 with the same cooling system backing up an i9. A f*cking i9! Which brings you to this.



I'd rather get struck by lightning before buying any other laptop from Eurocon.

 

PSA: Eurocon is not misspelt.

That was deliberate.

 

Well, I am not an advocate of ad hominem slandering of any vendors in a forum like this. Especially when they have a social media presence here in our community. That's just not cool in my personal opinion. If a person has a bad service experience and they want to post feedback about that and it is unflattering, well that's totally fair game in a forum. I've done the same a few times and accountability is good for everyone. Even then, it needs to be done in a deliberate and factual manner... just the facts ma'am.

And, to be totally fair this is a Clevo product, not a @Eurocom Support exclusive offering. It's not their fault that it is not a great product. All of the Clevo resellers and Sager shops sell it to the best of my knowledge. People that want it are going to buy it from one place or another, so this falls into caveat emptor bucket.

 

Stop spreading false information. The heatpipe connecting the two heatsinks in your Q6 (picture of this assembly here) has no impact on CPU temperatures as it is transferring heat from the GPU VRMs and MOSFETs to the CPU radiator. That heatpipe is NOT transferring heat to the CPU die. This heatsink assembly is more than capable at cooling a CPU with PL1 set to 45 W. I have the P950HP6 and I get temperatures of 80 C when encoding a video using x265 in Handbrake (see below for screenshots). This is with the CPU repasted with EK-TIM Ectotherm, which performs like MX-2.

Spoiler

If you added a third heatpipe to the CPU heat sink as well as increasing the pressure the heatsink exerts on the CPU, you could set PL1 to 60 W without much of a problem.

 

That seems to be the case.

What I cannot explain is why stressing the GPU only (furmark, to be exact) also raises the CPU temperature considerably.

And another observation: Your PL1 and PL2 are set to 45W and 56.250W respectively.
On the Q6, with default BIOS settings, they are set to 45W and 65W respectively.

You also mentioned that

(currently using MX-2 at the moment until I transition to liquid metal pads) , so perhaps that would explain the temperature difference between your setup and mine's.

The liquid metal pads are primarily meant to overcome the current limitation of low heatsink contact pressure with the CPU die. Hoping it holds up.

And I stand corrected.

 

Here's another example where an OEM cares more about looks and power-life rather than performance:

https://www.pcmag.com/news/363189/r...-2018-macbook-redesign?source=SectionArticles

I guess fat and chunky notebooks that prefer being plugged in don't rate much for news. Not like "best gaming laptops with good thermals" is trending. Just thin and light. Should be thin, light and throttled.

 

Thick and chunky is always better, but some people don't know the difference.

 

It depends on what the user's needs are. If the only thing you intend to do is browse the internet and use MS Office, you don't need a i7 8700K to complete these tasks. You would notice no difference in page loading times between a dual core i5 7200U and an 8700K. The main bottlenecks is the internet connection speed and the web browser itself. A notebook with bad battery life is not a notebook, it's a portable desktop with the battery serving as a UPS. If you're a student, you don't want to be lugging your charger around in your bag all your day and you don't want to be tethered to a socket all day if there even is a socket available. It's also pretty difficult to fit anything bigger than a Dell XPS 15 in most lecture halls.

Buying an ultrabook for its performance is like buying a Ford Focus for off-roading.

 

Current model (Eurocom Q6).


Current Cpu Package power draw 27.7w or about +30% lower Cpu Package Power (heat) vs the old model.



Yesterdays model (Eurocom Q6) with no pipes from GPU VRMs and MOSFETs to the CPU radiator.


Current Cpu Package power draw +36w or about +30% Higher Cpu Package Power vs the new model who ships with the extra heatpipe who tax the cpu grills.



The bigger die and +30% less Cpu Package Power Heat should mean the today's Coffee lake processor could run a lot colder? More Wattage = More Heat. There is no major changes (same graphics) outside the extra added heatpipe going to the rear Cpu grills for the new today's Coffee lake model.

And you own one?

What bios version? If you have tested newest and older bios... Did you see a change aka could you max out the 110w TDP for sustained max load before? Or is this tragedy... 95w package power limit just for sustained max load or short? See... You can reach 110w (95w) limit with short time max load but not sustained max load?

 

Typically for gaming you prioritise GPU over CPU. Metal pads are not suitable for direct die from what I remember.

 

There are multiple reasons for this difference.

1) Heat soak - the i7 8750H produces a lot of heat when it is in its PL2 state (67 W, significantly more than the cooling system can handle). That extra heat is going to remain in the heat sink and heat pipes and will not be removed by the fans and radiators and will result in higher temperatures over time.

2) Different thermal pastes and application of that thermal paste as well as different mounting pressures (depends on what torque value the screws have been tightened to)

3) The Eurocom Q5 (Kaby Lake version) is based on the Clevo P957 and Q6 is based on the P955. The P957 has more cooling vents on the bottom than the P955, which would lead to the Q5's better results vs the Q6.

4) Voltage - the 8750H is set at a rather high voltage - 2.2 GHz @ 0.79 V is extremely high considering that many 7700HQ's can do 3.4 GHz @ 0.9 V.

Your theory states the CPU radiator is saturated with heat from the extra heatpipe that is connected to the GPU. This is rather dubious because it is transferring heat not from the GPU die itself but from the GPU VRMs and MOSFETs which do not produce much heat in the first place. How could such a small amount of heat saturate the GPU radiator? It would be very easy to test whether this is true or not by increasing the fan speed. This will result in more air passing through the CPU radiator and therefore more heat would be removed from the CPU radiator which in turn should result in a lower CPU temperature.

 

As you said. “Can” be several reasons. But add pipes from other heat sources into Cpu grills won’t help on the Cpu cooling. And 22% bigger die + 30% less power consumption “in current real time - test over prolonged time” aka less heat should still lower the temp a lot. I hear you talking about voltage but Wattage will still convert into heat. More wattage from heat sources means more heat. Less means the opposite.

 

Exactly. But, what happens when you are looking for a Ford Explorer, the marketing claims that vehicle will meet all your off-reading needs. You show up to pick up the device, and find out it is an AMC Pacer. The point I was making is that the OEMs should stop claiming their devices can do anything, but in reality, they cannot. Another analogy, unlimited data caps. Not really unlimited when there was actually a 5 gb cap. Whoops, they didn’t expect the end user to recognize they were being manipulated.

 

Did you read my post? There is no evidence to suggest that the heat pipe connecting the two heatsinks together is the primary reason (or is even a factor) as to why the i7 8750H in the Q6 runs hot despite it's lower power consumption vs the Q5.

Well, it depends on the laptop. The i7 8750H is marketed as a 45 W CPU, so consumers should expect PL1 to be set to 45 W. The all core turbo ratio is set to 3.9 GHz. With PL1 set to 45 W, there is no way in hell any 6 core cpu, even an i7 8086K, will reach that clock speed under full load. A more realistic target is 3.1 - 3.2 GHz, which corresponds to a sustained Cinebench R15 score of between 1050 - 1100cb. Most thin and light Coffee Lake laptops seem to achieve this target either out of the box, or with a - 120 mV undervolt applied. Oh, and by full load, I mean normal programs like Handbrake and Cinebench R15, not power viruses like Furmark and Prime95.

 

And you posting the evidence? Aka you have correct and the other wrong? Oh'well, thats ok. For me will higher power aka more heat pushed in current real time over prolonged time + (colder chips due bigger die size vs. about same amount wattage) who same time put out less heat speak more for it's self. But I shall not bother you more with this

 

You were the one making the claim that the heat pipe connecting the two heatsinks was the primary cause of the Q6's higher temperatures, so the burden of proof is on you. I agree with you that the Q6 should run cooler than it does compared to the Q5 because the CPU power consumption is lower, but I dispute the fact that it is the connecting heatpipe that is causing the issue. I provided 4 much more probable reasons as to why this is the case.

 

It will always leak heat as long the whole metal is connected/linked together. And the added extra pipe will suck it up and push it to the Cpu rear grills. Put it this way... Add in a second engine on same old single radiator ain't the best.

 

Maybe, probably, possibly that I own one :O


Newest bios as of 17 Aug 2018.

Yes, short is 110 and sustained is around 60-65 which is weird.

 

Dang, you have jump on Azorz's TRIPOD wagon

Tried previous bios versions to see if it may help? Of course if Dell alow you to roll back. This is how Dell let you overclock nowadays. For the 2017 linup AWbooks. See also how Dell play game/ball with Micro$h4ft

 

The average consumer will not have a clue about power limits. Even I don’t have much experience, though I’m slowly learning. Let’s assume a typical gaming laptop has a 45 w TDP, such as my current device. Let’s also assume that the manufacturer even includes the TDP info as part of the sales pitch. What they won’t share is the actual power limit the device is constrained so as to not exceed it. On my current device, MSI is keeping me from getting close to the 45 w TDP. I can go to XTU and change PL1 to 6o w, but the computer never gets much above 32 watts or so.

Most consumers aren’t going to use benchmarks either, or know anything about undervolting. We are aware in order to optimize the devices we have. I personally don’t like being the Eskimo someone is selling ice to. Hence, why I am here to increase my awareness and make informed decisions. When a trusted publication uses words such as beats the competition, I would like to be able to take them at face value. Unfortunately, sucker is what they see in me.

 

With that in mind, I feel like I've made the right decision.

 

pl1/pl2 setting always fail in xtu, or doesn't apply at all

I can set current (amp) limit but the PL limit is going to be triggered first rather than the amp limit

 

I'll try to rollback later on.

 

That is weird (unless things changed in the latest firmware update)!

Did you set "performance mode" in the BIOS or did you select an OC profile in the BIOS (these are the only options to set both power limits to 110w and XTU can't increase them, it can only lower them)?

http://forum.notebookreview.com/thr...5-owners-lounge.815492/page-181#post-10772725

Did you reach thermal throttling temps (even if those temps were attained for a short duration)?

With stock paste, the dynamic power limitation begins right after thermal throttling was triggered and the power draw is dynamically adjusted so the avr. temps don't exceed 93-94C (quite a big difference from 75C):

http://forum.notebookreview.com/thr...5-owners-lounge.815492/page-136#post-10760367


Here is a LM'd AW 17r5 i9 that can hold a CPU power draw of +100w (it was current limited before it could reach 110w):

http://forum.notebookreview.com/thr...5-owners-lounge.815492/page-145#post-10762121

The same user can now sustain 110w (he probably managed to find the optimum undervolt for his unit), but his CB scores are still lower than avr for 4.3Ghz.

 

I set performance mode for fans in the bios I think, and overclock mode to "customization" but left all of the cores to 0 (zero/default).

I have liquid metal applied on a heatsink that has been rebalanced and all arctic thermal pad on the VRM to eliminate possibility of VRM throttling/overheating.

I am on the latest bios, and do not have any of the Alienware / Dell application installed. Only use XTU and Throttlestop.

 

I don't think those dell OC apps are really needed since doofus99 doesn't have them (AFAIR), so you should be able to set PL1/PL2 limits to 110w by selecting the proper performance or OC settings in the BIOS.

I think that there is also a performance mode for the CPU in the BIOS, but I don't know how are the OC profiles or OC settings displayed in the BIOS.

There is definitely a difference between your unit/firmware and what the other users have, because with standard performance mode, the limits should be 45w PL1 and 90w PL2, while you mentioned that you get 60w PL1 with 110w PL2 (which is further limited to 95w).

From what I've seen in the AW 17r5 thread, the limitation that were encountered so far (if the proper performance settings were set in the BIOS, so PL1=PL2=110w) are:

-the dynamic power throttling mechanism that is only triggered if the temps reach the thermal throttling range (doofus99's AW),

-current limit throttling (Aristotelhs2060' LM'd AW).

 

Seems like you are spreading false information yourself, @RampantGorilla
How come on a MSI laptop, which has separate CPU and GPU heatsinks, when the GPU is put under heavy load, the CPU temperature increases even though the CPU is barely used (like only 1 core) e.g. Valley/Heaven ? GPU *CORE* temp increasing by 30C raises CPU idle (very light load) temps by about 10-15C. At 100% maximum fan speed too. And this can be tested by manually reducing the GPU fan speed while keeping the CPU fan speed constant to make the GPU even hotter.

And on a MSI turdbook, the GPU VRM heatsink doesn't even transfer heat to the CPU radiator! It has its OWN radiator that is cooled by the CPU side fan!

Can't make blanket statement answers like that without scientific testing.