Questions regarding my new Clevo P775TM1-G

[MiRaGe]

Thanks, mate!

I believe with stock Thermalpads, rated at 4WmK the VRMs get overheating when both mem and core are overcklocked, strongly believe with the new heatsink from AliExpress and new 12WmK pads I would manage to squeeze a bit more from it…

[1610ftw]

On 2/6/2023 at 1:21 PM, MiRaGe said:

Thanks, mate!

I believe with stock Thermalpads, rated at 4WmK the VRMs get overheating when both mem and core are overcklocked, strongly believe with the new heatsink from AliExpress and new 12WmK pads I would manage to squeeze a bit more from it…

 

If you are curious if it is in any way temperature related you could keep the bottom open and blow some really cold air on it in a cold(er) room. If that doesn't help I would expect other limiting factors. 

 

In any case you did very well with your P775, what is the max power draw of the GPU during Time Spy?

[MiRaGe]

Thanks a lot!

yes, I have to try this with the opened bottom case👍🏻
well, it draws a good 206W on Time Spy

[srs2236]

Hi!

 

I have good news and bad news regarding 3333MHz and low timings (well the timings I have for example) and 1.45V on the Ballistix RAM.

 

Well the good news is that I believe the motherboard is in fact stable with 3333MHz and low timings.

 

The bad news is the unstability (in my opinion) does come from some serious overheating from the Ballistix sticks at 1.45V. Let me explain. Just like you found out, if I ever were to run some real stress test other than some benchmarks it would throw errors after 1-3 minutes of the test running. And once they start they just never stop and it keep spitting errors like crazy. I just kind of dismissed that and kept my original timings for like almost half a year and never really had a problem in the OS. But then again, I never did nothing else other than pretty much just play Rocket League or some other casual gaming. And once you made the post about the RAM errors I tried OCCT and sure enough it still threw errors after 1-3 minutes.

 

So I first started with creating a cooling solution for the RAM underneath the bottom cover. I used a TON of K5 PRO and on one side 0.5mm copper sheet and on other side 1.5 or 2mm copper sheet. Compressed both sides together as much as I could hoping K5 PRO will stick enough so they don't fall apart. As a safety measure I tried tightly wrapping kapton tape around the bare copper so that even if it came loose it wouldn't just fall off. Well after my first test I already confirmed its getting rather uncomfortably hot. I tried OCCT but it failed again in 3 to 4 minute mark.

 

I was thinking how to tackle the RAM stick underneath the keyboard but there are problems. Even a 0.5mm copper sheet doesn't fit under the ram and maybe a 0.5mm fits on top but might create clearance issue with keyboard and unwanted flex on the stick from the keyboard. So because I was bored, I began to fiddle with what I got. I took copper tape, which is like very very thin like aluminium foil just copper and sticky on one side. I began covering the whole keyboard bottom with the copper tape, then applied 1mm thermal pad on top of the ram, 1mm thermal pad (like a nice big patch) on left, middle (right where PCH is) and right side of the keyboard. I did this in hopes it would transfer some heat from the RAM to the metal casing below with the help of the copper tape.

 

I installed keyboard back in, put the screw for it and removed the back cover and litterally turned on the PC with it being on my lap screen down so I could test the thermals using my finger (very scientific I know xD). I started the OCCT benchmark and after the 3min mark (where it would normally start spitting errors like crazy) it was so hot on the RAM heatsink I swear I couldn't leave my finger on it it was burning so badly. I wish I had a thermal couple with me as that had to be very very hot. BUT the good news is, the keyboard thermal pad/copper tape mod had to have worked as for the first time I passed 5min test! I know its pretty much a joke 5mins but I am sure now it is just a thermal issue because of 1.45V on sodimm with all of them being put so tight together with 0 cooling capability on their own. It is not a good combination if you think about it. And once I figure out better ways to cool them I am sure they will run stable for longer than just 5 mins.

 

I will give picture to how the heatsink looks for bottom RAM, forgot to take pics of the keyboard because I was kind of messing about didn't think it would do something meaningful.

 

So, another thing, maybe if you leave timings alone OR loosen the timings and put 3333MHz on 1.35V it might be more stable as it shouldn't overheat as badly. Another theory is that the copper tape made the shielding better but I dont think so.

 

EDIT:

I actually just tried 10 mins and it was even stable on 10 mins! I have never seen this before ever! I am absolutely shocked that some copper tape and couple of chinese thermal pads actually did the trick! So I assume both of them need to be cooled somewhat otherwise they overheat on 1.45V. Anyways, I am off to try to push OC even more now to see if that has had some other impact aswell xD And before I run it longer than 10min I need to improve RAM cooling even more. I know it can be done, I have more ideas how to do it. Just need some time and tools. Also knowing how hot it got during the first 5mins I don't even want to know how hot it got at the end of 10 (have it screwed together already) and don't want to risk stressing it more until I improve cooling.

 

 

Timings are the same as I always had.

[MiRaGe]

I do really recommend this light mod for P775, brought the temps down by a good 3-5C:

 

 

 

 

[zex4]

Noice, do you also use cooling pad ?🤔

[MiRaGe]

Nope

[MiRaGe]

OK, LM on GPU, CPU is delidded with LM applied between the die and the IHS. Max GPU temp at 200W - 74,5C, CPU dropped a good 6-7C from the delidding.

I am done with this P775TM1, it is as it is:

 (160\800 stable)

First photo of HWiNFO is before delidding, second is after

 

 

[1610ftw]

1 hour ago, MiRaGe said:

OK, LM on GPU, CPU is delidded with LM applied between the die and the IHS. Max GPU temp at 200W - 74,5C, CPU dropped a good 6-7C from the delidding.

I am done with this P775TM1, it is as it is:

 (160\800 stable)

First photo of HWiNFO is before delidding, second is after

 

 

 

Is this with the new heat sink?

 

The 775 already is a laptop that is quite heavy already so it would be interesting to know how much weight gets added by the new heat sink.

[MiRaGe]

1 hour ago, 1610ftw said:

 

Is this with the new heat sink?

 

The 775 already is a laptop that is quite heavy already so it would be interesting to know how much weight gets added by the new heat sink.

Yes, this is with the new HS. It added around 350g to the weight of the Laptop.

[MiRaGe]

After all I count myself lucky as the CPU came after delidding like this (caused by the transport company, but bad glueing is to be blamed either of course..):

 

 

 

still working….😳😊

 

 

[MiRaGe]

 

 

 

[MiRaGe]

Further cooling improvements:

 

 

 

 

GPU tops out at 69-70C.

[MiRaGe]

I can finally say I got the temps under control. This laptop is a bitch to be cooled down. 
After prolonged game session I have my GPU max 72C (+160/800@200W) and the CPU doesn’t hit and stays at 100C anymore when both GPU and CPU are under heavy load. 
95-8C for the CPU in this case.

87C max after 10 Min. in CineBench R23. Not bad. 

[MiRaGe]

final temps after all the mods and adding more LM between IHS and the die 10 days later:

 

 

before:

 

 

after:

 

 

At the end - delidding plus LM - 10C drop. 

[MiRaGe]

my best run so far:

 

 

[1610ftw]

6 hours ago, MiRaGe said:

my best run so far:

 

 

 

12K Graphics score in sight! 👍

[MiRaGe]

I don’t really know how to achieve more, tbh…

this is at +160/950, one mhz more and it crashes😂

My last shot would be shunt mod…

[1610ftw]

Hehe, been there, done that 🙂

 

Looking at 3Dmark there are probably only a handful of laptops in front of you so this is an excellent result!

So I would keep things as is and not risk a fried GPU for a few more benchmark points.

[MiRaGe]

Is it just my insanity or unstable UVed CPU works hotter than stable UVed???🙄

My 9900K works cooler at stable -100mV than at -125mV unstable where generates L1 cache errors in CineBench....

 

 

[FTW_260]

Yesterday I accidentally found this cool custom heat sink for the good old p775 on taobao . It has two integrated vapor chambers for the CPU and GPU, as well as larger grill areas where the fans are installed. (it seems that the rear decorative panel on the case needs to be removed to install this monster)

It looks massive, heavy, and powerful. 
There is also the option of separate gpu and cpu heatsink, also with vapor chambers.
The price is about 150 dollars.

Pretty sure it can handle 400W+ heat with custom fans.

[MiRaGe]

It looks great, but I just spend 200€ a month on the custom heatsink from AliExpress… it has 6 (3 shared) heatpipes for the CPU and 5 (3 shared) for the  GPU.

Do you think the one from TaoBao would be much better?

[FTW_260]

25 minutes ago, MiRaGe said:

Do you think the one from TaoBao would be much better?

So far, there is nothing better than a vapor chamber, and I am sure that the performance of a solution like on the picture should be amazing.

[MiRaGe]

Can you send me a link to see more photos of it?

[FTW_260]

28 minutes ago, MiRaGe said:

send me a link

Sure - https://item.taobao.com/item.htm?spm=2013.1.w4004-22757732840.16.1bba286d2HKCvC&id=626569350720