Clevo 2019

http://forum.notebookreview.com/threads/clevo-2019.826781/page-48#post-10869804

 

10% is going from 4.0 to 4.4Ghz, higher CPUs are never worth it from a lower one in performance per $ usually.

 

Hi, thank you for your reply and sorry for very late answer. Very busy during last weeks.

Intel says that i9-8950hk has 2.9 GHz base clock @45W TDP while i7-8750h has 2.2 GHz base clock @45W TDP.

According to what you said, in order to explain this behaviour, can we say that every 8950hk on the market @45W always operate at lower voltage than every 8750h on the market @45W due to better silicon and less overall defects, leading to higher frequencies because of more current absorption ?

P=I*V so if tdp is set equal for both cpu models, less V leads to higher I; higher I translates into higher frequency (2.9 vs 2.2 @ same 45w)

Is this correct ?

 

In theory, yes, the i9 s should be more efficient, but don't let the 2.9Ghz vs 2.2Ghz mislead you into thinking that this is the "normal" difference between them.

Remember that this is supposed to be under a worst case NON AVX scenario and that 2.2Ghz is the lowest is can get for 8750H.

If you'll study some reviews (those from notebookcheck for ex.), you'll see that most 8750H CPU are more efficient and even in prime95, most of them will stay above 2.2Ghz.

The reality is that the 9850Hk are more at the limit of that rated efficiency, while most of the 8750H are above that (2.2Ghz) limit, so the avr. gap between them isn't that high.

There is a lower limit of efficiency, but the higher limit will depend on the quantity and quality of the chips that are manufactured.

I believe that the demand is also playing an important role in this. The demand for 8750H chips is much larger than the one for 8850H or 8950H.

If most of the CPUs that are produced have a relatively high quality and the demand for 8750H is high enough, I suppose that it is possible that some chips which are supposed to be rated as 8850H, might be sold as 8750H. This would mean that the 8750H's efficiency can be as low as 2.2Ghz@45w or near 8950Hk's.

 

This is how binning has been working for a long time. The mainstream chip is always in high demand so it does depend on the binning bell curve.

 

https://cpu.userbenchmark.com/Compare/Intel-Core-i9-9880H-vs-Intel-Core-i7-8750H/m750169vsm470418

+

https://www.clevo.com.tw/clevo_prodetail.asp?id=1183&lang=en (the 17" one)

=

maybe maybe?

 

Probably

 

I think the i7-9750h is more likely.

 

Keep in mind the TDP figures offered by Intel are basically irrelevant.

The new i9-9980HK is 8-core, 16 thread, 2.4ghz base, 5ghz boost and you can be damn sure it's not sticking to it's advertised 45W tdp at all.

 

It will if the motherboard demands it. Just like you could set a 9900k to 45w and then it will "turbo"/operate in that range.

 

I tried that on my 9900K a while back to estimate a 45W limit ... any power limit that would have forced it to run below base clock was ignored (3.6 used approx 65W on TSBench so effectively 65W is the minimum a 9900K can be run at using power limits).

I did manage to emulate below base clock /sub-65W loads with trial and error on Speedshift/EPP settings.

 

Well true but then again the base TDP is max load at base clock.

 

thats giving intel too much credit bro. "always" would be the key word and have to define how long it goes back to.

1st gen core i7 920xm, then 940xm more refined silicon
2nd gen 22nm - 2920xm & 2600k, then 6mo 2960xm & 2700k
3rd gen 22nm - 3920xm & 3770k, then 6mo and ONLY 3940xm, intel reducing cost starts now.
4th gen 22nm - 4930mx & 4770k, then respectively 9 month to 1 yr later, 4940mx & 4790k.
5th gen broadwell, first generation 14nm it was so bad for efficiency at high clocks they made very few skus based on it.
6th gen 14nm which will be first refinement of 14nm, 1yr+ after broadwell comes skylake.
7th gen 14nm+ another refinement which is another 1yr+
8th gen 14nm++ another refinement comes 9-11 months after with limited supplies
9th gen 14nm++ i think its better than 8th gen but hitting diminish return, the next comet lake 10 core will most likely be little to no improvement except 2 additional cores.

so all of those 14nm suppose to just last 1 yr, with the 2nd half of that year being a refinement but since 10nm went down the drain thats what they got. reducing cost, prolong the 14nm and reduce more cost.

 

You can also refine a process more for yields than raw clocks or cherry pick for a special series ala 8700k vs 8086k

 

Can confirm. Clevo has set PL1 to 55W and PL2 to (wait for it) 107W!
Source: we have i9-9980HK in stock now.

OC greetings,
XMG|Tom

 

well if Clevo keeps the cooling system of the LGA system then 107W would be no biggie to handle for their systems especially as temporary PL2 limit

 

You do have to work within the dimensions of the unit of course.

 

Indeed, I have since acquired one myself an I can confirm that so far I've used as much as 140W! (Overclocking)
And there's no sign of the overclock slowing down.

The cooling system is quite compact in the PB51 I have. Relatively few heatpipes compared to the larger LGA systems. However, the fans are significantly more dense and the exhausts run out to the rear and sides (as opposed to just rear on older models).
I've got mine running 4.6ghz all-core at the aforementioned 140W at around 85C (Max fans mind you). More info and benchmarks are in the OC lounge.

I suspect that the lack of IHS actually helps the situation immensely as the die is otherwise identical to the 9900K AFAIK.

 

The cooling does sit between the desktop models and the slim models for sure.

 

Yes for a user, but NO for a manufacturer. There is always room for improvement. The airflow wasn't designed to handle the heat from a 16-core 14nm++++ CPU. If intel follows AMD, then we might be talking about 32-core for the next generation 14nm++++++. Will the cooling still be enough? The answer is NO.

 

Depends on the TDP and thermal management settings.

 

We need more watercooling solutions.

 

Or just make the machine a little bigger with larger heatsinks rather than lugging around water plus an extra box

 

It's been tried before. Makes laptops heavy and bulky.

 

Yep I have just finished building a full self contained unit after a hybrid in the same model, and this is one of the main outcomes.

It's a stone age unicorn in that the chassis even had the space to fit in tubing, pump, res, and two additional fans and radiators, to attain the cooling performance to make the final product worthwhile (I'm yet to update my project thread but with the help of the external cooling >500W consumption, to the voltage limits of all the silicon, with noise barely above idle).

A hybrid setup where the cooling is done externally avoids most of the weight and retains most of the cooling/noise/thermal performance, but the required skill and attention to detail required to keep a frequently disconnected open loop safe and reliable is fairly significant, and outside of the majority of laptop users' experience and knowledge.

 

Plus in your unicorn case extra heatpipes, radiators and fans would have been lighter and performed better.

 

Yes, but the braggin rights tho!

 

Eh the 870 series certainly does that too for being the fastest...

 

Why not have both?

Spoiler

 

I think at that point you are better off with making some sort of a bulky 2in1.

 

Because it's a lot more practical while still being king of the room.