Without delid/CLU (ICD on heatsink), removed clips, added springs and good thermal pads it would hit 98c (temp throttle point) on stock clocks and mild load.
Now I can run shorter benches @ 5Ghz before it hits that point.
-
-
I'm good with the temps I am seeing from Kryonaut for the moment. My main concern with conventional pastes other than IC Diamond is poor durability. I hate having to be inconvenienced with reapplying thermal paste several times a year. I am seriously spoiled by CLU temps being awesome and the fact that it never wears out... CLU has always lasted me more than a year with no changes in temperatures on machines that I was not constantly taking apart for experiments.
-
Tried to play Phantom Pain.
Absolutely no throttle. 1440P, everything on absolute max, GPU max 75C, CPU max 85C. CPU 4.0 GHz, -100mV -
Did something prevent you from playing it, or was that "tried to" a misinterpretation on my part? Everything was good for you?
It is nothing short of amazing what value the delid with CLU on the die adds. It's like having a totally different computer. People that live with insane temperatures because they are worried about voiding the warranty on the CPU or scared of the delid causing damage don't know what they are missing. It's worth all of the risk (which is truly minimal) and I say screw the warranty. A CPU that runs right without the misery of overheating and throttling is a whole lot more valuable than the warranty.Last edited: Mar 15, 2017 -
Played it for a while actually. The game is really nice!
-
-
4.4/4.2 cache, no throttle
4.6/4.2 cache, throttle back to 4.1/3.8
Edit: bloody hell tapatalk monstered those pics ... fixing soon -
Nope haven't had throttling at anything under 4.9Ghz.
For 4.6Ghz I run a -77 offset on core and 4.2 Ghz cache at -160 offset.
4.8Ghz on my chip is stable at about 1.375V. -
Meaker@Sager Company Representative
Each chip is going to have a low voltage up to a point and then each 100mhz will start becoming more expensive. If 4.8ghz is into the expensive region in your chip it's going to be a lot harder to push around there.
-
I'm aware this chip is average and am not expecting big things - its not volts (or temps) that's stopping my chip at x45 though. It can run certain workloads like CB r15 at 4.6 no probs (scored 989 multi / 197 single btw which is low but not surprising given I haven't optimised my windows install at all and all I did was close open programs)
I'm trying to understand the throttle coming from (what HWInfo says is) System Agent / ICCMax that happens in avx/cache/memory heavy stress tests. I focus on getting absolutely 100% rock solid stable before moving on to the next multi - these are clock profiles I will use every dayGeorgel likes this. -
Meaker@Sager Company Representative
Note in the 7xxx series people are now putting in offsets to slow down the CPU when AVX instructions are used as they are that much heavier.
-
Which is a great idea and I'd be doing that if I could (I forget if that's all kaby lakes or a Z270 only feature?) but at x47 this throttle happens just idling at desktop...
-
Well all the stuff is here, just wish i get the time this weekend
Johnksss, Prema, Papusan and 1 other person like this. -
Didn't you have your set up all sorted by removing those C-clips? I was happy with my temps until I saw what you were doing at 4.8 with automatic fans.
-
-
-
-
-
Ionising_Radiation ?v = ve*ln(m0/m1)
I'm not happy with CLU, for all the hype it was given here. It dries up way too fast, and I'm getting temperatures on my GPU that I'd never usually get—90°C and above. The mounting pressure on my GPU heatsink is already rather low (it has @Papusan's hated 'triple screw' configuration), and when I opened up the heatsink to check, there was just this silvery dried powder. Any other suggestions?
hmscott likes this. -
CLU Needs perfect die to heatsink contact. Otherwise that dry up thing takes place.
For low mounting pressure you will have to either add a shim and make sure it makes perfectly tight contact and wont slide around or just stick to ICD / Kryonaut / Gelid GCE. -
To be fair it's not really the fault of CLU though. The hype is justified (mainly on CPU's) but you need to have good contact for CLU to do it's job - As you found out it will dry up as air gets in and then it's useless. You could look into a shim to improve contact pressure but for a GPU I wouldn't worry about it. You don't get the same kind of drops as you do with a CPU. Most of the other reputable goops do a good enough job.Ashtrix, bloodhawk, Georgel and 1 other person like this.
-
Ionising_Radiation ?v = ve*ln(m0/m1)
The thing is that both my CPU and GPU are bare-die—I have an rPGA CPU. I remember the temps of both very clearly when I first got my notebook nearly three years ago in July 2014: the GTX 860M wouldn't get higher than 71-72°C even on a FurMark run. The same makes my GPU jump to 80°C immediately, and it climbs ever further until thermal throttling kicks in. And I've probably changed the heatsink something like five times since I've purchased this notebook. Ninety-five degrees C is just pushing it way too much on a GM107 GPU.Georgel likes this. -
Based on the history you are describing, along with the CLU drying up, it seems almost certain that the heat sinks are not fitting correctly. If that is the case, nothing is ever going to work great for thermal paste. The best way to check is with pressure sensitive testing film, but that stuff is really expensive. Copper shims are cheap and even an ordinary thermal paste like Kryonaut, IC Diamond or Gelid would work much better with improved contact. I would order some shims from eBay and see if that helps. I bet it makes a huge improvement.Papusan likes this.
-
-
Most definitely liquid metal will not work under these conditions, you're 100% correct on that.
However, if the fit is extra sloppy, then no thermal paste is going to do a good job. That's what I was trying to communicate. That has to be corrected or the machine will always run too hot. A shim between the heat sink and the CPU and GPU dies may be an easy and affordable way to address it. Won't be as good as a proper fit, but could still be a massive improvement. -
Ionising_Radiation ?v = ve*ln(m0/m1)
I can find something like 15 shims for US$3 at a local reseller. How would I go about using them in the first place? -
I use 30*30*0.5mm shims. They would be larger than the bare-die laptop CPU and GPU, but it's OK if the copper shim is larger (smaller is not OK).
Very simple to use. Clean everything up to prepare for a normal thermal paste application. Put paste on the CPU/GPU, set the shim on top, then add paste to the top of the shim. Assemble. You're merely making a shim sandwich, LOL. That's all there is to it. You can polish the shim if you want to, but I've never done so.Papusan likes this. -
Ionising_Radiation ?v = ve*ln(m0/m1)
Heh, should've thought of that. However, wouldn't the two layers of thermal paste slow down and worsen thermal conductivity, compared to just one layer? I'm also considering lapping my heatsink.TBoneSan, Georgel, Papusan and 1 other person like this. -
Yes and no. A good fit is preferred, but the loss of efficiency is negligible compared to the improvement gained by plugging the gap and achieving good contact. I've seen massive improvements using a shim.
Also remove the c-clips on the heat sink screws that hold the screws and springs captive on the heat sink. Sometimes the c-clip can cause interference and keep the heat sink and processor from coming together properly. Removing the c-clip eliminates any possibility of contact interference. The clip is only there as a convenience to keep everything together while the machine is disassembled. Once assembled, it serves no purpose and can actually become an impediment.
Watch this short video...
bloodhawk, ChenZenn, Georgel and 1 other person like this. -
-
Yeah, I hate this TRIPOD design. Same also for the unified HS design with shared pipes, heatsink or worse everything shared.
You need plugging the gap to achieving good contact. And if the gap is to big as I think, lapping your heatsink wouldn't work. Remember to check the contact gap also for cpu when you use a shim for gpu.bloodhawk likes this. -
Doubling the number of paste-metal interfaces does double the potential for air gaps but that is affected by the flatness of the surface of the heatsink and application technique.
The thermodynamics of it are that when calculating thermal conductivity across a gap, far better to have any amount of 400W/mK copper in there over 14W/mK thermal paste -
5.0Ghz O.C. , scores look right?
http://www.3dmark.com/3dm/18655409?
http://valid.x86.fr/9i804a
5.2Ghz Validation:
http://valid.x86.fr/h5q5u9 -
Teensy bit low for 5ghz, i think most of us are hitting 16k.
Whats your voltage like for 5Ghz? Also Amazon or Newegg? -
Gotcha.
Though im not too sure about how much of a difference RAM makes on the FS Physics bench. -
I think most of y'all are also running 2800Mhz+ ram in dual channel so that should also make a difference.
-
@Mr. Fox or @Johnksss@iBUYPOWER or @Papusan Might know better.
-
-
Forgive me if this seems like a dumb question but as a new NP-9873-S owner is there a way to keep the cpu and or gpus in general to run cooler? ie: Laptop Cooler ect.?
I read up on delid but i dont feel comfortable doing that just yet. Any feed back would be great thanks.Jon Webb likes this. -
Will do Thanks.
-
Can you link me the U3 mod video? cant find his youtube.Jon Webb likes this.
-
Also, is there a cooler that i can buy that would work best?
-
I'm trying to find the mod pics. when I switched my c drive I lost all my favorites. And no, I don't know of ANY stock coolers that will work like this one. Give me a minute and I'll find the pics of the u3 Mr Fox modded.
-
@Solo0071 go to this notebook forum (Best Notebook Coolers for 18" Alienware Laptops) there's @Mr. Fox mods in there, @Papusan mods are in there. Check pics from some posts on this forum from @Johnksss@iBUYPOWER, you'll have to scroll back, his posts are a few days old but these guys will show you how to keep the laptop cool.
-
Meaker@Sager Company Representative
Re-pad, re-paste, de-lid and use of liquid metal will also contribute to a tuned system.Jon Webb likes this. -
so i just reinstalled W10 on my notebook only with bare minimum drivers (chipset,wlan,gpu,sound) and CPU still deactivates it's turbo feature in GW2 and Overwatch, looked a bit for some bios settings but none seems to make a difference whatsoever, World of Warcraft tho doesn't trigger that "disable cpu turbo" tho and when it occurs there is " NO" Limit reason. So the GPU it is then? I mean after unchecking "disable turbo" in Throttlestop 4.5ghz sticks, till i close the programm and the game, then i have to restart TS ofc, if i leave TS open 4.5Ghz is almost permanent.
My brightness fix still worked tho 376.60 install > 378.78 device manager install.
Sorry if i sound a bit vague, sitting here with the flu i already wait for the day it's over
-
@Meaker@Sager is 100% correct. My system is de-lided with CLU between both, the processor and IHS and between the IHS and heat sink and it runs COOL. But I watched @Mr. Fox You Tube videos several times before I even thought about doing that myself. Definitely educate yourself before you try that, and ask the guys on the forum if you have any questions. There's a lot of experience here.
Clevo Overclocker's Lounge
Discussion in 'Sager/Clevo Reviews & Owners' Lounges' started by Spartan@HIDevolution, Mar 4, 2016.