*OFFICIAL* Alienware m15 Owner's Lounge

Hey! I'm currently repasting and changing the thermal pads of my r1 for gelid gp-extreme. The yellow thermal pads are a pain to remove, so sticky!
Is it me or all thermal pads are 0.5 mm except the 2 long bottom ones that are 1.5 mm?

 

Can someone direct me to a diagram of all the thickness the pads SHOULD be? I haven't been able to find them, the ones near the CPU should be 0.5 mm right?

Sent from my Moto Z (2) using Tapatalk

 

That's kind of what I measured, although the yellow ones were 2mm for me.

 

screw #3 on the heatsink still doesn't want to go down all the way.

 

Neither it did for me, nor for at least one m15 repaste YouTuber.

 

Not sure if I did this right, it was the first time I repasted a computer. With phobya nanogrease xtreme spreading technique and gelid gp-extreme thermal pads here are the results of the cinebench r20 with a -150 undervolt on my m15 r1 8750 rtx2060:
with speedshift at 128 had 2759 with max temp on all cores being at around 80 degrees.
with speedshift at 0 had 2943 with max temp on each core being 96, 85, 100, 90, 97, 90
second run: 2997 with 95, 85, 100, 90, 98, 90

Seems like two of the cores are running more hot than the others.

 

Yes, idle temperatures didn't improve much with the repaste, I suppose that you can dissipate 6W just over the board without even using the fans. Using battery the GPU fan is totally stopped and the other one is rotating very slowly and I don't even hear it in total silent. I'm trying to improve the situation when on AC, I suppose that I'm not long enough below 40C to see it stopping. Let's see once the ambient temps go down.

 

What was the ambient temp? Any improvement vs stock paste at all? Could be that the 3rd screw is a factor. Could try the pea method I guess. How is the GPU doing after the repaste?

 

Seems that the X application method could be a good option too:

https://www.pugetsystems.com/labs/articles/Thermal-Paste-Application-Techniques-170/

 

Ambient temp was a climatized 21 degrees. haven't tried the gpu. Feels like it is almost the same as stock paste.
It is weird how different cinebench runs can have very different results. Got from 26XX to 2997 with speedshift at 0.

I will try repasting with the X method. I don't have to change the thermal pads everytime I repaste right?

 

Doesn't look great Some variance is normal, you can set higher process priority to manage this. Also the first run from cold could be better if any throttling is involved.

 

Can anyone share a HWInfo trace file of the first 5 minutes of prime95 "blend" test without undervolt?
I am especially interested in the resulting graphs of laptops that were repasted

 

Did this couple of days ago latest p95 version, AVX disabled, however this is on Large FFT test (>4 hours). I did it to test stability of my undervolt (i'm on conductonaut). Kudos and virtual Rep (since i've already repped him) goes to @kahuna0k (have a look at his post at page 315 of this thread). I followed his feedback and did the same on my m15 R1 and managed a stable static voltage of 0.8203 V . Here below my CB20 score too. Not to shabby

 

Hmm, your core voltage is showing as approx. 1V, not 0.8?

 

Yes i know, don't know why, it may have something to do with LLC being a little too aggressive (I don't know if we have that on laptop and as far as i know there are no way on laptop to adjust that) ? these are my TS settings:

 

Thank you.

 

Two Qs:
* Any reason not to apply the same static voltage settings to the cache and iGPU?
* Any known downsides? Higher power utilisation / temps on idle I guess.

 

Now.... those are some smart questions, that i didn't ask my self
I'll do some test this evening and will see.
i'll call in @kahuna0k for him to chime in to see if he has some answer to them too.
Etern4l, several pages back on this thread there was a talk about Throttlestop doing nothing regarding the TDP limits settings on m15.
Please, do one test. Run Prime95 with stock TDP settings on TS TPL tab (they should be PL1 45W and PL2 90W as default). You will see that during P95 test on main throttle stop window the power draw from the CPU will be 45W and the cpu will be power throttling (mine was a 3.7ghz) Now with P95 still running change PL1 at your liking and apply. Frequencies will boost at the intended 3.9Ghz on 6 cores and power draw from the CPU on TS window will increase. In my case, given the voltage i've applied, my CPU will draw aprox 52W.
Seems like Pl1 settings on TS does indeed work.

 

Good weekend homework! Will try to do it, and who knows, perhaps even go through with the Phobya repaste if stars align!

 

Thanks

Would anyone mind testing with 35W PL1, Balanced fan profile in AWCC, Performance power profile and no undervolt?
I have the same test results (did by myself) for m15, razer blade 2019 and lenovo P1.

I am curious to know if a repaste of m15 can get the temperatures to be a bit closer to the razer blade.
The razer blade 2019 Base model (supposedly no vapour chamber!) with 2060 ran 10+ degrees cooler than my m15 with virtually no noise

 

How the... I am getting without undervolt VID values of 1.42V! I must have lost the sillicon lottery. Because the CPU really wants that juice from the PSU. Is there any way to see the real VCore values?

Adding a static core voltage in Throttlestop does not seem to be doing anything for me.

 

I have the same issues, Throttlestop will set the voltage but I understand that the LLC is increasing the delivered voltage over the predefined value.

I don't look at the voltage itself, I look at the power dissipated by the CPU, that seems to be accurate in Throttlestop. The goal is to drive that below 60W while running something like Cinebench R20 or mprime (I was using custom with 192K size and in place FFT to avoid unrealistic loads). If you look at that you'll see that you have much more room setting the static voltage than with the adaptive voltage, as at least in my case, you can go down to -200mV or even -210mV but only during load, the machine will crash at that levels as soon as there is no load, that's why I think that the static voltage is better.

About setting the cache and the iGPU, I suppose that you could do the same, I didn't even try as I was able to go below 60W just touch the core voltage. Will try later today for science

 

Sorry for the basic sanity check, but could this be due to a clash with other OC tools?

 

Could be the Cashe voltage settings who hamper the Cpu voltage from going lower.

If you undervolt iGPU. You should undervolt iGPU Unslice as well. Normally won't offset voltage on only one of the iGPU domain have any effect.

 

Thanks for the feedback Papusan, I'll do some more testing as soon as I'll have the time. For science, of course

 

I've tried to lower the Cache voltage using static and I've been unable to get even the same results. It didn't change the fact that the actual voltage is higher than what I set and, as others have noticed, this CPU is much less stable when you lower the cache voltage.

At 0.8750 both mprime was dissipating 70W while with my older settings it was at 60W, and when I tried below 0.8750 I got crashes. Leaving it as it was, as it works well for me.

The note about the iCPU unslice voltage, I made it equal to the Intel GPU following your recommendation, but didn't notice much change, I suppose that my iGPU is not doing much

 

After testing I confirm Kahuna0k results. I still suspect LLC being too aggressive. Shame we cannot adjust that as on desktop. Rised my static voltage to 0.850 (to give it a notch up from my previous stable tested 0.820. Max W under prime95 is around 52.5. also sync Unslice with iGPU as per Papusan advice. Gonna leave it like this for the moment being. I'm satisfied. Next step (not in an hurry to do that) will be trying to undervolt the Gpu using Afterburner voltage curve to see if I can make it boost higher. Just need to do my homework to fully understand how it works.

 

Vapor chamber on a laptop is actually probably gimmick, from notebookcheck base Blade 2018 Review and some user posts Blade base actually runs a bit cooler than advanced with vapor chamber. However base Model came a lot later after advanced they probably doing a lot things improving heat sink pipe design from previous Blade 14 which was bad.

 

So I repasted with the X method still with Phobya nanogrease extreme. I also replaced the 0.5 thermal pads for 1.0 ones because there were no dent in them.
It feels a bit worse. Got 2808 with max temp at 97 83 100 89 99 89 for run 1 and 2914 with 98 83 100 89 97 88 for run 2.
The third core was staying at 100 degrees the whole time.
I also noticed that there was a small dent in the cpu.

 

Sorry to hear, not clear where the dent could come from.

As for the temps, is it possible you are using too much paste? How thick is it? Did you heat it up before application? I think I will go through the exercise myself shortly so we can compare notes.

 

I might be using too much paste. Not sure how much I should use. But I feel like all the repasting and changing thermal pads didn't really have any effect vs stock.

Max GPU temp went to 87 during heaven benchmark. Not sure if that is good or not.
Ambiant temp is 24.

 

What if you continue with Adaptive volt for Cashe and instead increase negative offset voltage.

From post above...


I expect if you use most part dGPU, so there is little to be gained by adjusting the iGPU.

See also http://forum.notebookreview.com/threads/the-throttlestop-guide.531329/page-1004#post-10894096

 

Hey guys, I just got one with the i9-9900. When I looked in the bios at first boot up it said it was @ 3.10 GHz. Isn’t it supposed to say 3.60 GHz??

 

You don’t have the unlocked K version. 3.1GHz is correct.

 

Hi Skolar,

Your temps are way too high both on GPU and CPU. Never saw my GPU going over 71 °C (spike) and this is only because of the stupid shared heat pipes design of nowaday laptops, during gaming session both CPU & GPU works and heat up, adding to each other heat that won't be there if they were on dedicated heatpipes cooling systems (Dell has to think that they feel alone, so they heat up each other to feel cozy ) . Typical temperature for my GPU for example is 55 - 65 °C under load. CPU seldom goes up to 80 °C during gaming sessions (hottest core), but those are just spikes. During stress test that uses only the CPU, the temps are as per my previous post)

Looking at your CPU temps it might really well be too much TIM used. Also, next time you repaste, try put your heatsink upside down (heat pipes down) on a flat surface (a glass surface is my favorite) and see if it sits completely flush over the surface. It is not uncommon to have warped heatsinks, and this can be fixed by slowly and carefully bend it so that it will be perfectly flush. Take care to the VRM area (above CPU & GPU). I've used K5 Pro with pretty good result on those areas (where the incredibly stupid sticky yellow pads dell put on those machines are, and yes, they are a pita to remove, isopropil alchool will help you big time removing the sticky stuff that will remain attached on the heatsink. Alternative? Zippo lighter white gasoline. Does wonders on leftover stikers glue)

How much undervolt have you applied on the CPU?

 

This is with a -150 or -155 undervolt. I have replaced all yellow thermal pads with Gelid gp-extreme. Paste is phobya nanogrease extreme. Had about the same results with the spread method and X method for the paste. Frankly at that point I'm not sure I can do anything about the temps.

 

Try to reapply the TIM.
Keep in mind that:
the amount of TIM between the heatsink and your components (CPU / GPU) must fill a really, but really thin gap. Actually is just to assure proper thermal transfer (heat exchange) of heat from the chips and the copper surface of the heatsink and for micro imperfection on the surfaces.
With thick thermal greases (IC Diamond 24, i'm looking at you ) you want to heat them up before application (put the siringe on a ziplock, boil some water, put it in an let it warm up while having your heatsink ready to be reinstalled right away, you want to put that TIM as warm as possible, it will help spreading it
On CPU i suggest the thin (really thin) line application. On GPU either the X or the dot in the middle (i do prefer the dot in the middle)

Remember that you can apply the paste, screw down your heatsink. Unmount and see how it has spread over your components and then adjust quantities as needed. By any means, while repasting take your time to do it.

Oh! and Never Ever Give Up!

 

I'll also add that having too much thermal paste isn't really a problem, having too little paste or bare spots is a problem.

 

Well with a re-paste, re-pad, undervolt and a little tweaking of the heatsink the m15 r1 is not a bad machine:

 

What heatsink tweaks have you had success with?

 

I used the tinfoil method on the cpu side of the heatsink for the 2 top screws (VRM side). This allowed me to tighten it down a little more and get even core temps. For pads, I used 1.0 and .5 mm arctic thermal pads. For paste, I used ICQ diamond.

 

Thanks, could you elaborate on the tinfoil method? Google is kind of coming up empty for me.

 

Automatic hibernation stopped working for me - opened a dedicated thread since it might not be m15-specific: http://forum.notebookreview.com/threads/m15-fails-to-automatically-hibernate.830202/

 

@hodgeMN Yes please, can you share some more info about it?

 

Didn't know that trick! How long do you leave it in the boiling water? couple of seconds, minutes?
Might try IC diamond. I probably need another heatsink or at least another screw as one of the screws is now stripped and impossible to screw all the way down.

 

You don't need to keep it in the water while is boiling. Just boil the water and put the siringe in, inside a zipplock. Regarding the timing. What i usually do is removing the heatsink, one of the most time consuming step of a repaste is cleaning out the old paste. Once everything is clean, i heat up the TIM, usually keeping it inside the hot water for 10 minutes, during this time i get everything ready so that it can be applied the fastest as possible.

 

Hi everybody! I have an m15 rtx. and battery has failed. I have a warning screen every boot that battery is not detected. Now the laptop cpu clocs are at 0.79GHZ and won't go higher is there any way to avoid it and make laptop run full power untill i will get new battery?

 

Check that you have the "Power" profile set to High Performance in the Alienware Command Centre and the Windows powersaving profile is set to the default Balanced

 

Hmm, I remember booting with disconnected battery once, but have no recollection of this emergency throttling effect. Weird.

Have you opened the laptop to check the battery connection? What does ePSA have to say?

 

I think it updated bios settings or something like that! As my aliennfx lighghts gone. I also booted couple times with normal clocks befor the windows update got a firmware update or something then after restart i got warning messager that battery cant be detected and firmware update will be cancelled. Press F1 to continue)). Currently i don't have battery inside the laptop but even if i connect it it will change nothing. The epsa says no battery detected.How can i reset the CMOS on it?