M17x r4 PCH Temps

Hi All

I was just playing a bit of BF3 and the game just locked up... not to worried about that but what i am worried about is one of the temps i clocked on HWinfo

It says my Intel PCH Temp hit 87.2c that seems a bit high to me.
why is it so high? my CPU (66c) & GPU (68c) seem to be fine.

Any help would be welcomed

 

That's normal on a demanding game such as BF3, normal temps during gameplay should be anywhere between 80-90C however by performing some of these steps you can lower them a bit. Specially the cleaning and the cooling pad could be good for your system.

 

Hi, since couple days im another happy user of m17x r4 and im having same issue as lilsancho. My Intel PCH temp can reach 80 C playing diablo 3 where gpu and cpu are reaching maximum 65 C. The bottom base is very warm. Is this normal? Can anyone else confirm same situation with their own systems?

 

Did some research, asked couple guys from dell tech support and they said that as long as temps don't reach 95 C or above there is no need to worry about. Just monitor your temps, keep your machine as cools as possible and it should be fine. I think best option to solve this issue will be buying cooling pad which is something I'm consider right now.

 

Well, I think the PCH in my M17x R4 is causing Sata drop outs when the temp are over 75 C, with frequent hangs and stalls when over 80C.

Under normal everyday laptop usage, let's say web browsing, Office use, there are no problems. But as soon as I stress the Sata ports, for example during game play, in switching levels or scenes, or Digital Audio Workstation (DAW) usage with Cubase and Kontakt, then the Sata ports frequently stall. It's not related to CPU (or GPU) temperatures. Only recently I found out that Aida64 and HWmonitor show the PCH temperatures. Both programs show the same temperatures. More important is that the PCH temperature correlates perfectly with the Sata dropouts. Never a dropout below 70C. Around 75 C a dropout roughly every 15 minutes. Around 80 C a drop out every minute. It never reaches 85C, unless I stress both the CPU and GPU artificially at the same time with something like OCCT. Then the PCH collapses completely, all Sata ports are offline and all drives are gone, followed by a BSOD.

I have 4 SSDs in my M17x R4 though. A 256GB mSata. Two 512GB SSDs on the 6Gbps Sata-III ports in Raid-0 in Windows (not Intel Raid), and an older 512GB SSD in the optical drive bay. The Sata drop outs occur mostly on the Raid volume, but also frequently on the other 2 drives.

With the recommended iRST driver from Dell, the 11.0.0.1032 version, the Sata speed first goes down from 6 Gbps to 3 Gbps on the first stall, then to 1.5 Gbps on the second stall, and remains at 1.5 Gbps at subsequent stalls. Only with the beta 12.0.0.1013 iRST driver it stays at 6Gbps after a stall. I recently found out that with the 12.5 iRST version the speed also stays at 6Gbps after a stall / dropout.

It appears the PCH is located here on the motherboard: http://forum.notebookreview.com/alienware-m17x/696521-location-pch-chipset-r3.html

As can be seen, also from the system board pictures in the Owner's manual, the PCH is not cooled by any fan, and its heatsink is buried under a piece of concealing plastic that spans most of the motherboard.

Would newly applying thermal compound help? Would removing the overlay of plastic help? Or would only more drastic measures help, such as pasting a copper plate on the pch or its heatsink, that would span all the way to the top of the CPUs fan help?

Does anyone else have these Sata dropouts?

Edit: The morning after I wrote this, I found out that the 12.5 driver drops the Sata speed too. It just doesn't update the Sata speed in the iRST GUI, only after you restart the iRST GUI.

Edit2: I forget to tell that I know why the iRST driver recommended by Dell is the oldest v11 driver. I found out that it has the shortest re-activation time for a stalled sata port or a dropped out drive. This behaviour will let most people believe it's the drive or the OS taking a bit longer to fetch some data. The 11.0.0.1032 driver takes on average 3 seconds to recover. The 11.7 driver 90% of the times never recovers from a drop out and roughly 50% from a stall. That's probably the reason why it was only for one day the recommended driver on Dell's driver page. The never officially released 12.0.0.1013 driver does not drop speed and recovers on average in 2 seconds. The french site stationdrivers used to have it, now it can still be found on several sites, google it if you need it, or pm me. Or never mind, and apply some thermal pads as in my next post.

 

OK!!! SOLVED IT!!!

Not true: (see mikecacho's post below, the 'bolts' are nuts holding the cpu bracket in place on the other side of the system board)
Removing the heatsink of the PCH is a no-go. The heatsink is bolted to the system board and is definitely not meant to be removed, at least the bolts cannot be screwed, trying it, I thought, I might be screwed. You cannot see the bolts in the Owner's Manual clearly enough, but the 4 silvery things to the left of the heatsink are bolts:

So I cannot get to the PCH to apply new thermal compound. So I could get to the PCH, but did not know.

However, I noticed that the Palm Rest Assembly is made of heat conducting metal, perhaps alumin(i)um. More important, this metal rests partly on the PCH heatsink. There is however no heat conducting agent between the PCH heatsink and the metal standoffs of the palm rest assembly. I placed some thermal pads on the heatsink exactly where the metal of the palm rest assembly comes into contact with the heatsink.
Here:
That reduced the temp by 8 C.

I also noticed that the metal clips of the keyboard are screwed to the metal of the palm rest assembly. I also applied some thermal compound between the clips and the metal. That reduced the overall temperature of the PCH by 11 C.

I stressed all my SSDs and heated the CPU to 95C, but I cannot create a stall / dropout anymore! The maximum PCH temperature I was able to reach before was 85C, after this, 71C. Before the PCH idle temps were 67C, now 58C. I am so happy!

 

I will try it when modding the cpu heatsink, different location My M14x R2 PCH cooling mod

Under stress oc cpu/gpu 95C max, poor PCH cooling design for the third most important chip.

Second thumbnail is not visible.

 

The developer of HWiNFO also questions the reliability and accuracy of the readings of the PCH via the program.. Personally, I haven't found the heat an issue, it has been pretty warm according to the readings and my computer has performed just as well. I think you guys are worrying too much over nothing.

 

I am not worrying anymore. I was worried when my SSDs kept stalling when I tried to hammer them, and that I could not find the cause. Only when I found the PCH temperature readings of HWinfo and AIDA64, I discovered that these temperatures consistently and perfectly correlated with the Sata stalls and dropouts.

That the temperature readings are possibly not reliable works actually both ways. I for one, am quite sure that the temperatures reported are too low, meaning that the real temperature must be higher than reported. Why? Because chips normally don't start to produce errors at 75 C. In HPC clusters CPUs run close to 100C for cooling efficiency reasons, that's a perfectly fine approach. Before I applied my home-made fix with the laptop half open and the keyboard removed, Aida and HWinfo reported 62 C, the PCH heatsink was too hot to touch at that moment, and my IR thermometer read 60C. A difference of only 2C between the on-chip temperature measurement diode and the heatsink is highly unlikely. So the real PCH temperature must be much, much higher than reported by Aida and HWinfo.

Still, all of this does not really matter to me, in that I don't care that I don't know the exact temperature of the semiconductors on the chip at any given time. What did matter, is that I had some kind of measurement of this temperature, albeit skewed by an unknown amount, that still IS in perfect correlation with the sata stalls and dropouts.


The reasons why most people don't experience PCH problems, or just think they don't, are, non-exhaustively:

1. Their PCH is not really stressed. I have 4 SSDs connected, three of which have high random access loads, only during this usage the PCH exhibits this behaviour. Most people buy an Alienware laptop for gaming. Loading game (levels) from a hard drive does not stress out the PCH as much.

2. Their PCH has a better cooling system. My system board is v1.0 revision A00. Did I get the prototype? ;-)

3. Dell forgot to install the themal pads between the PCH heatsink and the palm rest assembly in my laptop, those that I installed myself.

4. The PCH in their laptops actually produce stalls, but they don't blame it on the PCH. They think it must be the game that needs 3 seconds extra to load, right? Or it's just f^&(*king Windows again, or the graphics driver, or whatever. To be fair, that's exactly what I thought when the stalls and dropouts first occurred when this happened in a combination of Cubase, Vienna Ensemble Pro, NI Kontakt and EW Play. I thought it must be: Play, VE Pro, Cubase, the Audio drivers, Windows, the Virus scanner, firewall, network drivers, all kinds of needless software installed, chipset, usb and iRST drivers, etc. etc. Mostly in the order mentioned.

So thanks to the PCH temperature readings I could finally solve this issue. Therefore, if anyone experiences sata stalls on high SSD loads, then this easy fix might help you out.

 

Interesting, those copper plates. That was what I wanted to do but couldn't because of the bolted heatsink. Just make sure it doesn't come loose and comes into contact with the electronics.

 

Today I've managed to work on my most complex musical arrangements without the circumventions I had to apply to keep my laptop from stalling. I realized that the laptop has more than twice the musical processing power than I thought it could support. The cpu is now the bottleneck, no longer the SSDs and the Sata ports of the PCH. I am even happier than yesterday when I discovered this fix works, because it works even better than expected.

 

Have you extracted the +400mhz your CPU has to offer?

 

Changing thermal pad would not cause it to read 15c different either, but it does, so if it is not spot on with the exact readings i'm still confident that it was reduced afterwards... Changing the thermal pad with a thicker higher quality grade pad helps greatly, I have done it and seen a 15c decrease (85 down to 65). If you are about better perform, heat reduction should be a big deal at any stage.

 

It is not bolted, you need to remove the CPU bracket on the otherside, once you do that then you will have to fight the TIM that is holding the plate down, it seems to be a liquid metal TIM directly underneath the CPU, honestly just bend the plate up enough to remove the old thermal pad, that is what I did and it was fairly straight forward, you still want to remove all set screws and repaste CPU since you have to loosen them anyways.

 

Do you mean overclocking the CPU? Yes I am using the BIOS overclocking features, plus Intel Extreme Tuning Utility. I didn't even know it was possible before buying the M17x. It's quite a bit of difference, also in temperatures though. It never crashed on CPU intensive tasks however, not even when the CPU temps are close to 100C. Perhaps I do need to re-apply thermal compound to the CPU.

 

Thank you so much! So in fact it can be removed!

At the moment I don't need further temperature improvements, it seems, although that hot summer day still has to come ;-)

A liquid metal pad. I thought those aren't too bad. They are difficult to remove I heard, almost as difficult as 2-component thermal adhesives. Some guy pulled the heat spreader off his desktop CPU when removing the cooler, after having erroneously applied a 2-component adhesive.

Thanks again though, I will have a try if the PCH plays up again during that hot summer day. That, and a copper plate all the way to the cpu fan, possibly even on the cool air inlet side on the bottom.

 

I just took my R4 apart this morning, there were NO TIM (idk what that means) underneath the CPU it was 3 foam pieces with a very sticky double sided tape, to take the heat sink off all you need to do is loosen 3 black screws and remove the CPU heat sink then carefully(but forcefully) pull.

And when I took off my heat sink, guess what I found! One crappy piece of pad! I replaced it with some high quality pad.(Fujipoly Extreme System Builder Thermal Pad) My PCH used idle at 76C~, now it idles at under 60C

 

That does look like your usual cheap thermal pad :/ The PCH can operate at higher temperatures simply from being a much larger process compared to the other chips (intel use their older nodes to make the chipsets).

 

Glad to see you were able to remove it fully, not all machines are made equal (i had a greenish pad originally). After a month of gaming with the new thermal pad replacement and max stress test at 1 hour intervals, my PCH temp never exceeds 78C under heavy load.

 

A small bumb because i was busy today with this certain problem.

I had the same issues, i downloaded hwinfo and when i use it it will hit 90 - 95 degrees and people can say, that's normal, really, now well, it isn't normal!
I think it's not normal because 95 degrees is just crazy, especially a almost lazy chip , it doesn't make sense, but OK.

What i did, i had some left over thermal pad and some paste.
I removed my keyboard and that's it, then you can see the plate of the PCH.
First thing i see was the little bumb of the cover trying to touch the PCH, or the iron plate anyway, i thought, let's try fill this gap, i used my thermal past and forced it in there, and on left of that bumb i did 2 small thermal pads of 1,5mm stacked. ( why I did this I don't know, it just felt good xD)

I reinstalled it all and all I can say is wow, I don't what changed, or what i did lol xD.
Just a little bit of thermal paste and pads and now the temps max on 80 degrees after 1hour!!! (instead of 90 - 95!!).

I almost cannot believe this, took me 10min to fix this....

Senshin