M17x - 9400M G / MCP79 Temperature & Discussion Thread

Many posts about this issue exist in different threads so I thought it would be nice to focus on it in a separate one.
As many agreed the importance of NB cooling is paramount.
Someone tried copper modding it, others reapplying the thermal paste, with some positive results.
So bring it forth folks, your temps (both idle and load), possible slowdowns/fps drops associated with its overheating, your attempts to fix it, ideas etc.

Here's my part:

9400 enabled

Temps: idle-57-60c @max load-70-72c

A noticeable slowdowns observed whenever NB reaches 70c. At that time cpu temps may even be below 60c (P8700), and both 280s are around 70c.
So far my only thought is to find a way to lower the NB temps. The first thing I'll try is reapplying the thermal paste (as soon as my qx9300 arrives). If it won't help - copper mod.

 

yeah i have the same too - 57 - 70 C. imo this temp depends of ambient temp. when my window is open i usually get 60

 

The MCP always seems to run hotter than the other components. It should be cooler in the ATI systems since there is no Hybrid and PhysX support.

 

so is it mean that if i turn off my integrated video the temps will be lower?

 

It should. You can also turn off PhysX in the nVidia Control Panel and test the temps.

 

Even without an integrated GPU, Nvidia MCPs have always been known to run hot.
I wish this thing had an Intel chipset with a nice, cool northbridge.

 

Hmmm, I can't find PhysX in the Nvidia control panel. Can I just uninstall it from the win control panel?

 

If I still had it I'd give you a screen shot. Maybe someone else can help out.

 

i did a cooper mod before but didn't help a lot after that i replace the cooper shim for ic diamond 7 carat compound and my temps are 10c lower i have 3 extra ic diamonds let me know if you want to buy it

 

We were discussing this in Side Topics...

My 9400 hits 90-94c when gaming. Idle is normally about 75-82c That was with it enabled and with it disabled (You can check 9400/MCP temp in Everest when it is disabled)

Witcher had suggested the same to me there, some IC 7. Might give it shot.

I wish the temp would top out at 70c like some of you. I'd feel a whole lot better about it. =P

 

disabled 9400 in bios - the NB temps @ load stay the same (everest) , even slightly higher. But as a side effect got the SLI visual indicator vertical bar while in games. Can't remove it due to the stock drivers' nvidia control panel (sux imho, no option to disable Physx either). Is there a way to update nvidia control panel without updating the drivers?

 

If you have the 9400 disabled in the bios, along with hybrid disabled, you can install NB drivers from Nvidia and you can install laptopvideo2go drivers also without worrying about whether the hybrid will work. At which point it's just a matter of finding the latest drivers that work well.

Only things that wont work is that you can't lighten/darken the screen. =(

 

Removing the stock thermal pad from the northbrige and replacing it with Arctic Silver 5 lowered my Northbridge temperatures to 53°C under full load, instead of the very high temperatures similar to what others report it used to reach before.
You can use some better thermal paste, like MX-3 or even IC Diamond 7 carat, of course, but Arctic Silver 5 was the only top grade thermal paste I had at home or was available to me at the time I put the laptop apart.
While you've got your system open and are replacing the stock thermal compound with fresh high-quality one, I suggest you take the opportunity and do the same to the CPU and your discrete GPUs. It'll help to keep your system much cooler.

 

I must be missing something here - this is not the first post I have come across reporting this. I have the Integrated Graphics and Hybrid Graphics both disabled in the BIOS - SLI is active and I have installed a LV2Go (aka non-Dell driver). I am able to control the screen brightness.

What process did you use to install your driver and which driver are you running?

Just trying to understand this issue you are reporting. Thanks.

Glzmo, can you define 'under full load'? What is your initial temp reading for the 9400 @ power up and what is the idle temp?

Trying to figure out why my 9400 (when I do have it enabled) is sitting @ 50-53 @ idle when I have also applied aftermaket compound (Arctic Cooling MX-3) to the MCP79.

 

What is your 9400 temp while @ load, BatBoy?
IMO if I had 50-53 idle and 60@load I'd be more than happy.

So we got to the first stage - replacing thermal compounds.
I'd really appreciate your help in guiding the process. I'll get some AS5 or MX-3 in a few days. What tools/additional materials would be needed for this?
Any how-to videos?

Oh! And how much should I buy to be able to replace the cpu/NB/GPUs t. pads?

 

here is a comparision of artic silver 5 vs ic diamond guys forget as-5 stay with the best http://xtreview.com/addcomment-id-2884-view-IC-diamond-7-vs-arctic-silver-5-review.html

 

It's sitting idle at 35°-40°C. The 53°C (now 54°, by the way), is the maximum temperature I got, even after hours of heavy gaming, stability tests (prime, OCCT, etc.) and a combination of both now. I did notice it takes quite a while for the Northbridge to cool down when idle after the system being under full load for a longer stretch of time.
Perhaps the layer of the paste you put on wasn't thick enough, I noticed the stock pad was rather thick and I therefore put a rather thick layer of Arctic Silver on the Northbridge chip. Maybe it's also because I didn't do the "put a pea sized dot in the middle of the chip and let pressure and jiggling of the heatsink spread it" technique. I found it more effective in the past to just spread the paste evenly on the chip and slap the heatsink on.
Also, it depends on how hot your CPU gets, as it's the same heatpipe and heat can carry over a bit from one to another. But from what I gather, your laptop is running rather cool otherwise.
Also keep in mind that each chip behaves differently, so that might factor in as well, although it shouldn't be such a big difference in temperatures.

Heh, yeah, if I could only get my hands on it where I live. Arctic Silver will have to do until the Diamond stuff is available here. But yes, go with the best.

 

Yeah - I'm going to look at this more tonight. I'll enable the 9400 for some gaming and post my temps.

With the MX-3, you actually can't spread it. Arctic Cooling's tech guys advised to just use the rice grain method, which I ended up following. I figured my temps were decent but also decided to just disable the 9400 since I haven't spent much time on battery. After seeing your report of under load temps I now have to re-examine this. If I can keep it sub-60 then by all means I'll leave it enabled.

I also need to re-install Everest as well. I only use HWMonitor. Have you compared HWMonitor and Everest to see if they are reporting the same temps?


-----------------

@Aikimox: In regards to a guide, be sure to watch Moo's Video as well as download the Service Manual. I basically used the service manual as a reference as I was taking it apart the first time. 1 tube of compound should be enough to do all of it. Be careful with the GPU's. I didn't apply MX-3 to my 280's because I did not want to mess with the memory chips. Next time I have it apart I will be applying it to the GPU cores. There are just too many memory chips on the card - lol. Rather not mess with it.

If this is your first time applying thermal compound be sure to do a few google searches on application methods as well as reference the compound manufacturer's site for application suggestions, etc.

 

Yes, HWMonitor and Everest show the same temps, but wow, I've got confused myself now... I thought that MCP is the NB, but now after some more research...it's the SouthBridge!!
So it means we don't really know if the NB is getting warmer when 9400 is enabled. All I can say is that SB is about 73c@ full load on my machine regardless of the 9400 state.


Thanx for the advice, BatBoy.
Going to do more digging around.

 

Going to run some games - here are the temps prior to any GPU intensive apps. I have re-enabled the 9400 (I typically have disabled in BIOS - both Hybrid & Integrated Graphics).

System has been powered on for about 90 minutes and I have been catching up on NBR items as well as some Excel work.

EDIT:
Temps after roughly 2 hours of game play tonight (note Max Column - far right). Left 4 Dead, Left 4 Dead 2, TF2. All in-game settings @ max and @ native resolution.

 

Sorry, I've been away from the conversation for the evening...


I read somewhere that controlling the Screen Brightness is a feature that is enabled by having Dell's drivers installed. Which has been true in my case. I couldn't control the brightness when I had 191.07 Drivers installed from laptopvideo2go with integrated and hybrid off. Re-enabling everything and reinstalling Dell's Drivers allow me to set the brightness now.

What version were you using BatBoy?

 

So it looks like MCP79 combines both NorthBridge and SouthBridge into one chip, and Everest sees it as MCP. 9400 is neither NB nor SB and therefore the temps will be different.
So basically - MCP =NB+SB
9400 is outside and has a different sensor.

PLease correct me if I'm wrong, but this confusion has to be solved.

 

the 9400m NB and SB are ALL on the same silicon die...they are the same chip and will therefore be the same temperature

 

We also need to ask, what is the ACPI (TZ00) reading? Where is it being read from? We know that the 9400M G/MCP79 is identified in HWMonitor as NVIDIA 0x0862.

@narkoleptik: I'm using the 187.66 driver currently. Prior to 187.66, the 186.82. Still haven't decided between the two which is the better choice for my needs. I have always been able to control screen brightness, but then again, I have not used (at least not on the M17x) the 190x series drivers since they are modded desktop drivers.

------------------

see my post above, added 'after game' temps.

 

Ok - so I just re-installed Everest (removed since I use HWMonitor for temps) and looking at the two side by side, I find a few things which are of interest:

1. Everest lists the MCP @ a temp of 48c. There is nothing in HWMonitor which matches this.

2. HWMonitor lists the NVIDIA 0x0862 @ 44c and Everest has an NVIDIA GPU#3 entry @ 43c. I believe this is in fact the 9400.

So, either HWMonitor does not register the MCP79 thermal sensor and is only reading the 9400 portion of the MCP79 (quite possible) or, they are in fact seperate. I still believe they are both one chip and HWMonitor just doesn't support it, but why would the MCP79 incorporate 2 sensors?

3. I also recall, long ago, something relating to Intel CPU sockets where there was a seperate sensor from the CPU located underneath the CPU socket on the MB. This could be the ACPI temp reading in HWMonitor. In my screenshot you can see where HWMonitor is reporting the ACPI temp @ 34c and Everest shows CPU @ 34c as well as 4 additional readings for CPU which are for the 'cores'.

All very interesting and worthy of further research.

In case you are wondering, the reason for the difference between the CPU core temps (Everest vs HWMonitor) is due to my manual change of the TJMAX value in HWMonitor. I have it set to 105c which is what I believe is the max for the QX9300. Everest appears to be using 100c. This is something I will have to continue to explore - the QX9300 may not have a TJMAX of 105 after all.

Just want to say thanks to you Aikimox for your thread - now new questions have been raised and its going to bug me until I find the answer. Argh!

 

That's the overall CPU reading (the Sensor is somewhere between the CPUs - but that's not the exact description), AFAIK, the one Intel uses for thermal specification ("T-rating"). The other two/four are for the separate cores.

It looks like my CPU temperatures are a bit higher than yours and my GPU temperatures (both 9400 and 280M) lower. My CPU is overclocked, though.

 

Exactly my point. Why would there be 2 sensors showing 2 different temps for MCP and 9400, especially if they are on a single die.

Should I also order AS thermal compound remover&surface cleaner?

 

I used a Q-Tip and isopropyl (91%) to clean up. Be very careful if you go this route as a Q-Tip can easily leave behind fuzzies. Have a microfiber cleaning cloth handy. No need for thermal compound cleaner/conditioner IMO.

One thing I also noticed in Everest was (looking at the Motherboard ---> Chipset), notice how the Northbridge and Southbridge device links report different revisions (B1 and B2 respectively)? I installed the trial version of Everest, so I cant see the actual label Everest is reporting for the Southbridge device entry. I know its just a revision indicator - but why the different readings? This is really confusing. I wonder why Everest is picking this up. Are you also running the trial version?

NOTE: For late comers to the thread, nobody is saying they are not one in the same, I am just trying to identify the different info Everest is reporting vs. HWMonitor.

EDIT:
Remembered I snapped a pic of the MCP79. Attaching it - note the B2 reference.

 

Yep, I'm running the trial version as well.
That MCP reading is getting on my nerves Initially I thought it is just some artificial calculation like the one with the cpu average temp, but after some observations - there is no consistency between 9400 (gpu diode 3) and MCP readings, the first may go down while the last may well go up. As for the B1, B2 ... even more confusing. We need more tech info on MCP79.

 

That is pretty strange... I just now noticed gpu3 now also and mcp seperate and different temps.

Interesting indeed...

 

Ok, I must correct myself, I do see a consistency between those two temps though the MCP temp is 5c higher than that of the 9400. One die - 2 sensors (probably).

 

That would make sense, since there is only one chip under that heatpipe.

I mean, Aside from the CPU. =P

 

Well it is on a seperate heatpipe technically

 

After more than a week of testing:
MCP reading in Everest is ALWAYS 5c higher than 9400's.
If 9400 is bios-disabled the MCP temp drops 5-6c on average.
So unless your temps are perfectly low I'd recommend disabling 9400. In my case (@max load) the MCP temp dropped by 6c ( from 76c to 70c) with ambient temp hovering around 26c.
Still waiting for a new cpu and thermal paste...

 

Agreed. Disabling the 9400 is fine and if you are going to be on battery power for an extended period of time a simple reboot and bios enable is all it takes to bring the 9400 back to life.

 

Another update guys!
This one is very interesting. I wasn't sure until now, but it's too much for a coincidence.
If you disable your 9400 not only the MCP temps stay 5-6c lower but also the CPU temps are 5-10c lower as well. And this is something to think about.
It's probably due to the fact that both MCP and CPU are on the same heat sink.
Anyway, stay tuned. Got my AS5 kit. Waiting for a second HDD caddy + the replacement for the broken (by a tech guy) touch panel+ new CPU, and will update you with the results.

 

Thanks for the confirmation. Figured it would lower temps a little bit without the added 'work'. Since I have always defaulted to HWMonitor I never saw the actual 9400/MCP79 temp with it disabled. Everest is a nice one since it will report the temps even with it disabled.

Will watch for your update.

 

My NB is sitting at 42*C idle, and goes to around 65*C after a few hours of Crysis/Wolfenstein.

That's with the 9400 enabled, and on a base model M17x.

 

Meanwhile, I encourage you to post screenshots with your HWMonitor/Everest temps (+the room temp) and details on what thermal paste was used (if any).
It would be nice to have some statistics on this issue and make it helpful to all AW owners.
Since temps vary from game to game (also OS and drivers), it is best if you provide 2 screenshots, say 1 is for idle (after 1-2hrs of web-browsing), 2 @load (1-2hrs, specify the game), - pretty much what BatBoy did.
So I'm gonna be next.
Idle (few hrs of browsing) BEFORE applying thermal paste. Room temp is 26c

@Load BEFORE changing thermal paste (Kings Bounty The Legend) Room temp is 26c
It's not the most demanding game, so temps are lower than in Crysis.

 

Temps for me idle for 2 1/2 hours is bouncing from 49 to 51c.

 

Can you provide a screenshot and a room temp?

 

Room temp is about 74*F. Closed out 3 browser sessions and Windows media player to take the screenshot. System has been on for about an hour. No cooling pad, just have it sitting beside my recliner on one of those wooden dinner tray tables. I have applied no thermal paste, farthest I've opened it up is to see if my 2nd HD hardware was included. Current was 52*C....max got up to 55*C.
The temps that I had the other day 49*C to 51*C was in our bedroom with the Roomtemp/AC on 65*F.

 

Thanks for the update. We will also need the load/gaming temps.
Now keep in mind that the actual temp on MCP is 5c higher than 9400'. So idle (ambient 23c) your MCP is ~57-58c.
It's similar to mine.
I'll probably do a first round of thermal paste (AS-5) in a couple of days, then upon getting a new CPU will try something else (MX-3/OCZ-freeze) and see what works better.

 

Update.
Finally got time+tools to try Arctic Silver 5.
Removed the thermal pad from the MCP heat sink (will keep it, just in case...), cleaned all the previous paste from CPU+Heat sink using Arctic Clean Thermal Material Remover and used their Surface Purifier before applying the AS-5.
Indeed, the gap between the MCP and its HS doesn't allow for a perfect "grain" approach (I still used it on the CPU though with no problem), so here I applied more paste to ensure the contact between the surfaces (as Gizmo advised).

Didn't wait 200hr of advised curing time before posting screenshots (if anything changes in a few days from now, will keep you posted), so here is the fresh screenshot 2 hrs after booting up (same room temp of 26c, same apps running, browsing etc):
As you can see, already there is an improvement of 3-5c on both CPU and MCP.
Next step (in a couple of weeks) is the MX-3/OCZ-freeze. Will find which works the best.
Now for the gaming test...

 

Nice job Aikimox... AW @ NBR's very own AS5 vs. OCZ Freeze vs. MX-3 shootout. Look forward to your findings!

 

Ok. Same game (Kings Bounty The Legend), same room temp - better temps!
AS-5 is definitely better than the stock paste+pad+method that Dellienware used.

 

you play a dangerous game with that artic silver on the chipset because you have to apply to much to fill the gap​

and with the heat that think can melt and mess you chipset because artic silver is electric conductive​

the best thing you can use beside the thermal pad is IC. diamond 7 carat thermal compound​

the texture is thick and it wont melt also is more heat conductive than silver believe me when i first got my m17x i try AS5 alumina, cooper mod , and the best result was IC.7 thanks to viciousxml for the tip ​

 

that looks like a poem because it was all centered haha

 

thanks man i guess i am falling in love

 

I'm aware of the risk level, but thanks for the warning. There was no other thermal compound available in the entire city, so I got what was within my reach. Wanted to test AS-5 anyway before going to more solid and less conductive ones. Already ordered MX-3 and OCZ Freeze. Meanwhile I'm not gonna push the system to the highest temps (62C was the max for today). And just in case, gonna check the condition of the paste tomorrow. Anyway, I'm sure it's not worse than the stock paste they used, as it was smeared all over the CPU itself, no wonder I had higher temps.
IC7 is overkill for me, no offense, as I don't wanna risk damaging the chips and heat sinks if I ever need to replace it. And I do intend to switch CPUs quite often.