XPS 15 9570 Owners Thread

I owe you a bench as well. I will likely have time to do a 10min run tonight.

EDIT: Ran the test and got a crash, which is odd because I've been running my -150mv stable for a few days now. Will have to look into this further.

 

so probably the iunlocks method of transfering airflow to the ambient sensors would work great again.

 

Yep. But it takes some serious tinkering to make the additional "fin surface" above the VRMs (it is not just cooling the sensors, it is in fact cooling the VRMs).
Most folks just padded the VRMs to the backplate, for less fuss (not sure how well the recent idea of the thin pad "bridge" from the VRMs to the heatpipes not touching the backplate works).
Mind undervolting with MSI Afterburner wasn't practiced at the time so a less elaborate mod may work with it. But there was no i9 and 1050Ti and things may be worse with them.

 

Are you guys using Throttle Stop or Intel Extreme Tuning Utility, to undervolt your 9570's?

I'm new to undervolting but going to give it a try as it seems to be enough for the i7 and I don't fancy repasting a brand new laptop with these sausage fingers!

 

I think I am gonna go with padding to the backplate. I am not really convinced by the "bridge" method until somebody shows evidence.
iunlocks is for sure the best but it's quite elaborate.

Anyway, before doing anything I will do the "fan test" to be really sure that this PL throttling is indeed induced by VRM temps.

 

Did yours come in yet? I'm chomping at the bit, but Fedex is taking their time. I'm going to do a fresh install when I get it and then run some benches and get some temps.

 

@custom90gt It should arrive later today. I will just check for the main QC problems people are complaining about and then focus on understanding the thermals.

For my use case, the only important thing is that I manage to have CPU running at full potential under CPU-only stress. The GPU, I don't really care. I am a programmer. Anyway, I will try to make everything work regardless.

Probably repasting and undervolting immediately. Then making sure that the PL throttling is caused by accumulated heat in the VRM area. I will blast a fan over it, if it does not throttle, I'll start thinking about paddings and stuff...

 

The one thing I realized is that after applying grizzly paste, my temperatures didn't improve much, if any. Perhaps in solid load, I knocked off a degree or 2, but under heavy load it seems to be exactly the same. I am not sure if I put too little paste (I am used to pasting desktop CPUs) or it's a product of the system being able to push itself harder. I really have no way of testing and now I am debating whether or not I should re-pasted because I put too little. I went with the line technique, but I am wondering if spreading is better for the laptop cpu.

It could be that at -150mv, you get depreciating returns from anything beyond that.

 

It could be a lot of things. What I typically notice is similar temps but decreased fan speeds and increased clocks with a repaste. If you didn't take logs before and after, it would be hard to tell.

 

The only way to know if your repaste has a problem is to open up the system and look; make sure the die is covered evenly and the coating is thin. Compare to pictures of laptops on the internet. User @GoNz0 recommended trying to repaste-test-visually inspect-clean-repaste a few times that way you know "for sure" what the quality of your work is.

Also, when testing thermals, your cores should run within a few degrees of each other.

Low mounting pressure repastes of laptops are a bit dicey. There are not too many stats showing before-after repaste results for the 9570. The system is very similar to the 9560 so you could look at those stats and see before-after results.
​

Finally, your laptop could have come out of the factory with a good paste job so your repaste won't improve much there. The nearly identical 9550 and 9560 had universally poor factory paste jobs as well documented here.

​

Not my experience with higher-performance laptops and the XPS, which always can use more cooling to improve performance and/or reduce fan noise. . .

EDIT - Less paste is always better as you are trying to minimize the space between the die and heatsink; the paste is just to fill in microscopic defects. Thermal paste is a poor conductor of heat but it is much better than air​