Poll: 17R4 Repaste Temps, Raised vs. Flat on Desk

Some of us are experiencing massive temperature differences when raising the rear of the laptop or running on a laptop cooler. I don't think many people have reported this factor in their stress test results, and I think many of us are curious to see if our repaste efforts have been successful or not. Hopefully this poll will shed some light on this matter and set a benchmark to compare against.

I set the cut off at 80c because it seemed like a good threshold based on reports - below is good, above is not so good.

Please only vote if you have at least repasted your CPU/GPU, this is to help us benchmark against other people's results.

Also, let me know anyone has suggestions on how to improve this poll and I'll make the changes.

 

flat on desk - 100% load 2hrs 91c CPU, 86c GPU
raised back - 100% load 2hrs 75c CPU, 66c GPU

I wish they just made the rear feet thicker and higher... would solve these temps just like that. Right now it struggles to suck in air... just a 1/4 inch height increase would help tremendously.

 

Wow that is a big difference. Is this only a thing on the 17 or also 15 and 13?

 

Would be nice to know what test, clock speeds, undervolt, pl1/pl2 unlocked. Many factors that all effect temps. Throttling taking place? Clocks able to hold?

Typically under load I haven't see mine go above 81c. I also see no point and stress testing for 2 hours as no game or software would run CPU and gpu at 100% for 2 hours. I found after about 20 minutes temps don't increase vs 1 hours. I would think unless ambient changes 2+ hours should yield similar results.

Just checked on aida64 at 20 minute flat on desk LM repad repaste 82 c was my highest on CPU.


Sent from my iPhone using Tapatalk

 

my cpu differentials are small too. 1-2c difference. 91 90 90 89 and 75 74 74 74

im fully stock in terms of voltage/oc. latest bios. only did liquid metal repaste and repad.

 

I would imagine it's similar for all models. The bottom clearance is the same.

 

Same here, at most 3c difference between cores, but I used traditional paste so probably makes sense. I am at 95c when flat on desk, 76c when on laptop cooler.

I'd be interested to see @Freitz do a 20min test with the back raised, to see how much lower it gets.

 

I'm beginning to think I was wrong to not test on a laptop cooler. Before, it seemed like my repaste efforts were a failure, but running further tests on an old 14" Zalman cooler with the fan set to LOWEST setting:

Firestrike - 17091, CPU (4.1ghz stable) core temps 76/74/74/70, GPU (+100/+100) temp 64c.
TimeSpy - 6812, same as above

I tried to replicate iunlock's OCCT test at 40x, and it seems like my core0 and core2 are still 12c higher. Does anyone know if this means the heatsink needs to tilt down towards the 2 screw side, or the 1 screw side?

 

Why are the Temps so different when the rear is raised? Why wouldn't Dell add slightly larger feet on the rear? You mean to tell me a bunch of randoms on a forum can figure something so simple out and a group full of engineers can't? Changing the size of the rear feet looks like it would save them a good amount of money on service calls and repairs.

Sent from my SM-G955U using Tapatalk

 

i have a 6700hq model with a gtx 1070 (17"), i made a custom stand for mine, no extra fans or anything, just something that allows me to sit this beast on my lap while giving 2 inches or so of air flow underneath, also have a -0.150mv undervolt in place. ran prime95 small FFTs for 8 hours, cpu peaked at 66c, but averaging 60c or so, i did also run a test of both heaven benchmark and prime95 together, that made the temps peak to 76c iirc on the cpu, which when down to 71-72c when fans kicked in, gpu was at 63c. all in all, i do believe that raising the laptop helps alot, the temps were higher for sure when on a flat surface, however i dont remember exactly how much.
but that was an unrealistic test, when im using this thing for gaming, with a cpu heavy game like gw2, the cpu hovers around 55-58c, and the gpu about 59c

 

Which test would you like to see? Aida64?


Sent from my iPhone using Tapatalk

 

When the system is repad and repasted and heatsink is balanced out properly. The only reason to raise the back is if you are chasing temps or benchmarking and or overclocking. The temps are fine and great


Sent from my iPhone using Tapatalk

 

Thanks @Freitz, a consistent benchmark will help some of us determine if the repaste was a success. I really appreciate the help. It might also help others decide if a repaste is needed.

Can you please run a test with these settings?
BIOS:

• Reset to default, restart.
• Performance CPU and Fan
• Custom OC Profile - cores at 35x, power limit 92000/92000 (to test CPU at its advertised max turbo boost clock)

Software:

• Undervolt -0.100 (seems like every CPU can handle this as a baseline)
• High Performance Power Plan

Aida64:

• FPU/GPU test
• Start with back lifted by at least 1", run for 10 minutes, record max temps.
• Then while test is still running, put it flat on desk for 10 minutes, record max temps.

What we're interested in: max core temps (individual), GPU temps, both while lifted and flat.

I'll post my results next.

 

17R4, 7820HK/GTX1080 results after repaste/repad with Kryonaut/Fujipoly 11w/mk

Lifted: CPU - 82/76/82/79, GPU - 69
Flat: CPU - 94/89/93/90, GPU - 83

No throttling while lifted. Thermal throttling started soon after laptop was laid flat, fluctuating between 2.9-3.1ghz on all cores.



I know that these temps will likely never be achieved during use, but it should show the limits of the cooling system of each laptop for comparison purposes. It also illustrates that a laptop cooler can have as much impact as repaste/repad, if non-technical persons do not want to open up the system.

 

Here is mine. I don't have my 1" risers with me so this is flat on desk. I am a little dumb founded today because I was about 8C cooler at home then I am at the vacation home. Ambient temp is about the same so trying to figure out why is running hotter here. Mine is Also LM'd

 

After some more testing and conversation with @Freitz , I think we need to establish a safe baseline test that everyone can easily replicate (with as few steps as possible) in order to determine and compare systems' heat dissipation capabilities. I'll try to keep it as simple as possible, and start a new thread.