*** Official Clevo P770ZM / Sager NP9772 and P770ZM-G / Sager NP9773 Owner's Lounge ***

Stock max fan
stock auto fan (caught it on a high, 77-78 avg on auto fan)

 

Those are some nice numbers

 

Wait until tomorrow! so long as I get up early, I'll post some links to raise some eyebrows!

 

thanks! mine are 9C higher than that

does anyone have this weird problem with autofan kicking in on GPU instead of CPU first sometimes? when only CPU is under stress?

 

The CPU can send heat over to the GPU side pretty quickly so that may be why.

 

FS listing is up!
http://forum.notebookreview.com/threads/fs-clevo-p770zm.783935/

 

the Gpu sensor shows lower temps at that point, so i wonder...

 

finally got similar temps with just Gelid extreme under the lid! maybe 2-3C higher! the key was in applying more pressure with fixing(gluing) the lid back!

 

If you delidded, you really should use clu between the die and ihs!

 

Yep, the heat travels over to the heatsink, the fan spins up and cools down the GPU with the two main GPU heatpipes.

 

i dont really want to mess with liquid metal.

btw, i have assambled another p771zm for a friend without deliding the 4790k. after 4th attempt i managed to bring the temps from 99 to 80C in XTU stress test on stock freq with autofan.

even more so, lowering voltage -70 helped to bring it down to 75-78.
i am gonna tear my own machine appart now, because i have realize that most of batman heasinks come slightly bent!

the bent does not allow the surface of the heatsink to fully lay on the cpu HIS. they are bent the way that the part that goes to the fan bumps into plastic casing and the power of the springs on the cpu plates is not enough to straighten it up. so before i realize it, i had to use a lot of thermal paste to fill the gaps in every p771zm i have assambled before.
this one was my 7th one built and lots of lessons learned for good!

 

Pretty impressive drops in temps there, even without liquid metal. Let us know if you can get it to drop any more by fixing the bend in the heatsinks.

 

that's exactly how i managed to achieve this massive drop on my friends p771zm, just but straighting up the sink!

i will do some more experiments with my own machines, which has a delided cpu already, but i assume the heatsink is far from perfect there. will need a few days, since i need the machine up and running right now.

 

I'm at stock multiplier, -60mV, 68-72C, auto fan, ambient 68F. Be careful you don't bend the heat pipes or dent them. I'm on my third one because I wasn't careful. But it sounds more like you are discovering what I called a learning curve on seating the heatsink. It's not new, but you are learning. There is what you are talking about, then there are warped plates. Verify it is flat for the contact with the cpu as well.

 

not wrapped plate, thank god, that was'nt the issue, but the radiator itself was bumping into the plastic casing near the fan, i did had to bent the pipes slightly to make it straigh and whoa! i got a thin and smooth layer of thermal paste after that under the plate. before that there was a really bad connection with the HIS.

 

well my friend has now same temps at stock as you showed me in XTU stress test, but with no clu and not even delided, magic!

 

Every chip and heatsink is different.

 

Plus, you said high 70s, which is still damn good!!!

 

btw, do you think one cpu can be hotter than the other(not considering HIS) at the same power usage level?

my best guess is that most watts go into heat, so that same voltage+same frequency+same manufacturing process(die area size) should result is the same heat distribution per square are.
i might wrong though.

so unless the cpu voltage control adapts to ASIC quality, the heat output should be the same.

 

First, W=V*A so watts depends on voltage times amperage for electricity. But watts can refer to a heat measurement, which is entirely unrelated. When you talk about computer components, know which watts you are referring to.

Second, that is what meaker just said. Silicon efficiency varies greatly with CPU silicon, especially with Haswell! So, how you are trying to measure it doesn't fully get how hot the cpu gets. A high voltage (relative, not absolute) on one chip may have low heat, all other settings being equal, than another with lower voltage having higher heat.

 

yes exactly W=V*A, where A represent the actual workload. since the "performance" effectivness of modern cpu, gpu is still far less than 0.01%, meaning 99.99%+ of power used, in the end, goes into heating up the crystal.

some chips may have better ASIC quality, and other qualities that affect power consumption, which means they can operate ot lower power with same performance output. but still all of the watts used go directly into the heat in the end, thats just plain physics. yes, some chips can use less watts for same performance. but they will still produce proportianal ammount of heat to the actual W used, there is just no other way around.

the only thing i can think of, that can effect the "hotness" of the component is it's actuall quality of heat conductivity, that can vary from piece to piece.

again, this is just my understanding of things, which can be far or close to what exactly is going on. i did study rocket science in my young age though, haha.

 

.

 

So you have the rough idea... Now you should have an idea of gates for transistors. In the process of manufacturing, some fail. Also, some gates are less effective due to imperfections. These can cause more of the watts to wind up as heat (sorry, I'm a little brain dead, Turkey brain). But that is why, as you point out, lower volts will cause less heat, but there are still differences due to yield imperfections in transistors. Then, you factor in the variance in surface connections with the IHS (usually concave where it connects to the heatsink), the heatsink, the spread of thermal paste between the die and ihs, the tolerance allowed for how high the ihs sits relative to the die, the tolerance on die height, the thermal expansion of the black (most likely a polymer) glue used to connect the ihs to the pcb, any dents or imperfections in the heat pipes that effect the flow of heat, the azimuth of the fins for heat dissipation and any gaps between the fins and the fan not sealed causing the pressurized air to flow around rather than through the fins, and you have your answer ....
Edit: how you mentioned the watts being relational is how it used to be calculated until Intel bastardized the formula and made tdp no longer as related to actual electrical wattage...

 

Hm, all energy goes into heat, it is it's last state phase. when you have a certain watt usage in the closed die, eventually it all goes into heat, well 99.999 of it, the rest may go out in the form of radiation of other kinds.

so yes, one die can be more efficient than other, but if it consumes 50watts, it gives out 49.999999+ of watts into heat.

which leaves us with 3 most temperature effective variables:
1st - automatic software voltage adjustments according to chip quality(though that also lowers raw power consuption per calculations, but that the proportion raw power/heat still remains),
2nd - heat conductivity effectivness of the die, less effective ones tend to build heat up.
and 3rd -speed of heat distribution on the surface of the cpu.
i really cant think of anything else.

do you know if Haswell cpus auto-adjust voltage to the ASIC quality, like modern nvidia gpus do?

quote "Edit: how you mentioned the watts being relational is how it used to be calculated until Intel bastardized the formula and made tdp no longer as related to actual electrical wattage..."

i am taling about actuall watt usage from the wall, not the tricky marketing figures

 

I just said 2 and 3. Yes, number one is used. So all three are in play. I then stated other variables in the system.

Edit: 2 isn't discussed in those terms, but higher failed transistor rates (lower yields) make more heat build up since it must work around other transistors for the current. This causes inefficiencies which cause more heat. Unless you are talking surface imperfections. 3 is effected by efficiency per core and the thermal interface material from the die to the ihs, the ihs to the heatsink.

 

ok, good to know we understand one another perfectly

 

See my edit above...

 

good one, details are nice to learn. theoretically it is still covered by the 2nd.

"round horses in vacuum", you know.... lol

anyways, the ammount of heat produced per watt should be the same for all dies, the only difference is that some tend to build it up more, from what i understand.

 

A couple things cause that: 1) contact with the ihs above, 2) the spread of the TIM between the die and ihs (sometimes it doesn't spread right and leaves air pockets), and 3) arrangement of the cores in the die (think being sandwiched between or on the outside). You can Google search the die layout for the third.

 

The sensors can be calibrated differently so yes they can read as different.

 

Is it ok to use Isopropyl alcohol for cleaning fingerprints from the rubberized surface?

 

I only use a microfiber cloth with a bit of warm Walter.

 

Mildly moist micro fibre cloth is the most gentle way of cleaning things.

 

Yes it´s better without any cleaner

 

@Chrack, Meaker: Ok, thanks!

 

Be gentle and it should last a long time

 

I have this red light in my 1st audio port... The one before the 2 headphone ones... Is there any way to turn that light off? @ajc9988 ?

 

That is the optical audio out... Not sure how to turn it off, but now you know what component it is...search for a way to disable it...

 

You could try disabling the digital output in the sound devices.

 

@ajc9988 , I'm gonna install Premamod... I'm just not sure if my BIOS EC version will cause issues... It's the latest 314 BIOS with 309 EC...

Edit: There were no issues, it works perfectly!

 

Prema's BIOS comes with a compatible EC that gets flashed during the flash process, so your EC should not matter at all.

 

May I ask what the difference between NP9772 and NP9772-S is? What does the S stand for?

 

The -S models are basically the same system as the non -S models. They just have some commonly picked upgrades done at a discount. If you started with the standard model and added in the same upgrades it would be more expensive. So the -S is that it is on a special.

 

Makes sense, thank you pat

 

Oops, posted in wrong thread. Please ignore / delete this post!

 

@Prema or anyone else:

I installed the prema bios and the .09PM EC. I did the "unplug it, remove battery, hold power" deal.

Sometimes in windows, no matter what driver, no matter if X-Fi is installed or not, my earphones would have a super silent isue and my mic would not work. Sometimes the stuff works if I restart, and will work until I shutdown or restart again.

By silent, I mean it sounds like my earphones are being considered as rear two speakers, so some things can be heard faintly, some not at all, but windows error sounds are fully clear.

My mic would sound like there's a loop wire and just buzzes with beeps nonstop.

I went back to the original .09EC and 0.15bios, and now I haven't had the issue.

I've also noticed the PM bios says P750ZM in CPUID, but I don't think that's a problem.

Did anyone else running PM bios run into this issue? I suppose I can try to re-flash and see if it was just a small anomalous flash

 

I would try that flash back, it could have also been an odd driver issue.

 

Hey I have been incredibly busy lately finishing up my dissertation. I plan on running through all the drivers and updating them. I noticed that prema has released V2 of the bios for our system. It seems the only thing that has changed is G sync. Since I purchased mine before G sync was available should I even bother updating to V2?

 

Premamod has been updated with NVME support in BIOS I think and the vBIOS has better voltage control performance etc...

 

Hi.
I just done a firestrike 1.1 bench, what is your optinion about the score, temps...score, is this ok? win 10 pro