Am I the only one that rarely checks chipset heatsinks for proper contact?

SlowCobra

Limp Gawd
Joined
Mar 15, 2008
Messages
501
Let me preface this with some background. This isn't my first rodeo. I have been assembling my own PC's since the mid 90's so I have seen my fair share of motherboards and the corresponding chipsets. Ok now that I have that out of the way on to my question. :cool:

Often times when assembling a new PC I make the assumption that the heatsinks on the chipset have had the proper application of thermal compound or tape applied to them. I know I should never assume and usually I tell myself that I really need to check EVERYTHING when I get a new board. Then by the time I build a new one I go back to my assumption and general laziness. Perhaps it is just a desire to get my new toy up and running as quickly as possible.

Today I received my Gigabyte GA-MA785GM-US2H motherboard and something about the large heatsink just didn't sit right with me. It didn't feel like it was making the proper contact against the chip. I decided to take it off and inspect it before screwing the board into the case. Luckily I did as the thermal paste applied to it was hard as a rock and appeared uneven. While this is only an HTPC board and won't see any type of OCing or extreme duty use it is still a reminder that I (we?) should always take an extra few minutes and fully inspect everything.

Ok back to my build.
 
I've noticed with my Asus p55 mobo that the mobo's heatsink retention system isn't great and you can wiggle them around a bit, but it seems like all of the thermal tape is in place. Thanks for the reminder, I definitly didn't take a look when I put together my current rig.
 
If your temperatures are good, whats even the point of checking?
 
i remove every heatsink and apply ceramique, because 99 times out of 100 the supplied thermal paste is freaking petrified and unevenly applied. helps temps immensely when its done properly, and its a serious PITA to take care of after you mount the mobo and have built the rig...
 
Last edited:
If your temperatures are good, whats even the point of checking?

If you haven't even assembled it yet how can your temperatures be good? :confused::confused:

That is the whole point of this thread, it is often overlooked and yet it is an important part of the entire equation. Why not examine it before the problem pops up. Preventative maintenance if you will.
 
After frying one CPU (RIP, poor K7 :( ) due to a poorly installed HSF, I'm damn near paranoid about installing heatsinks these days :D
 
After frying one CPU (RIP, poor K7 :( ) due to a poorly installed HSF, I'm damn near paranoid about installing heatsinks these days :D

First thing I always do when I fire up a system for the first time is go straight to bios and check the temps. I'm pretty sure you won't ruin a system if you do that immediately. Also it helps to not get Nvidia Chipsets cause some of them run pretty insanely hot and sometimes require better/extra cooling to last a long time.
 
my old 680i board had about 1/16th of an inch of gunk between the chip(s) and the heatsink(s). i put arctic silver and real mounting hardware, dropped temps by 5-8 deg C.
 
Eh I ususally assume the same thing that it should be good.

If it doesn't look quite right though I'd be removing and putting on some AS5
 
I don't trust stock thermal paste/goop either, but uneven/hard paste on a new board is normal since it hasn't heat cycled yet to settle everything in.
 
First thing I always do when I fire up a system for the first time is go straight to bios and check the temps. I'm pretty sure you won't ruin a system if you do that immediately. Also it helps to not get Nvidia Chipsets cause some of them run pretty insanely hot and sometimes require better/extra cooling to last a long time.

Yeah, I have used an NF2-based mainboard once and wasn't too impressed with it. AMD/Intel chipsets FTW.

In the case of this K7 I wouldn't even had time to go to the BIOS as it fried itself before even beginning booting. There was this huge gap between the HSF and CPU on one side because I used screws which were too long. Still feeling retarded thinking back to it.

These days I check 25 times, shine a light source at the HSF from the other side than where I'm looking to see whether light is getting through... etc. etc. Anything I can think of to make sure there is good contact. After that is checking temps in the BIOS. Haven't fried anything since :)
 
will that void your warranty on anything? mobos, gpus, etc? im always afraid that making a quick check or trying to lower my temps 5C is gonna cost me $150 later when my mobo dies for an unrelated reason and they say its my fault cuz i tampered with it. :rolleyes:
 
In most cases, AIUI, it does/should void a warranty. Not that that stops the devoted--I actually had to do it on my now-previous motherboard due to a known issue with that (didn't help that much, but still).
 
will that void your warranty on anything? mobos, gpus, etc? im always afraid that making a quick check or trying to lower my temps 5C is gonna cost me $150 later when my mobo dies for an unrelated reason and they say its my fault cuz i tampered with it. :rolleyes:
Mobo chipset heatsinks aren't usually very complex, don't take much to get to. Sure technically the warranty may be voided, but that probably depends on the manufacturer. If it's not one of the fancier elaborate heatpipe ones, e.g. if it's mounted with the easy on/off finger clips, and you properly clean off the old and apply a quality compound, and remount the heatsink and there is no visible difference, nothing broken, I wouldn't have any ethical regrets about RMA'ing a board that dies months down the road. A manufacturer's RMA workbench jocky isn't going to notice.

You clumsily break something in the process, yes voided. Obviously.

Many GPU heatsinks are much more complex, have tightness/torque specs, etc. I usually shy away from doing these, current run of massive full side covered with heatsink graphics card. Shy away from messing with those, my 8800GTS has been running like a champ for over 3 years now with no service. But back in the day when they were two push pins, sure, no problem.
 
ceramique
Hmm, how well does this work compared to the latest Arctic Silver?

What I always do when getting new hardware, with the exception of a video card (riskier business with warranty), is changing the thermal paste with the latest Arctic Silver out there. I did this with my brother's computer on his northbridge and southbridge chipsets, and then I was able to keep a stable 400MHz overclock (before I unable).

I see video cards as "made to withstand stock manufacturing" -- which includes the thermal paste. I assume that how they applied the thermal paste and what they used specifically was chosen and done in a way that the video card won't have problems. Besides, if the card goes bust, they pay for it via warranty.
 
Yeah, I have used an NF2-based mainboard once and wasn't too impressed with it. AMD/Intel chipsets FTW.

In the case of this K7 I wouldn't even had time to go to the BIOS as it fried itself before even beginning booting. There was this huge gap between the HSF and CPU on one side because I used screws which were too long. Still feeling retarded thinking back to it.

These days I check 25 times, shine a light source at the HSF from the other side than where I'm looking to see whether light is getting through... etc. etc. Anything I can think of to make sure there is good contact. After that is checking temps in the BIOS. Haven't fried anything since :)

The NF2 northbridge chip was concave, had to lap the damn thing to get anywhere near decent contact. Mine's still trudging away after all these years.
 
Hmm, how well does this work compared to the latest Arctic Silver?

What I always do when getting new hardware, with the exception of a video card (riskier business with warranty), is changing the thermal paste with the latest Arctic Silver out there. I did this with my brother's computer on his northbridge and southbridge chipsets, and then I was able to keep a stable 400MHz overclock (before I unable).

I see video cards as "made to withstand stock manufacturing" -- which includes the thermal paste. I assume that how they applied the thermal paste and what they used specifically was chosen and done in a way that the video card won't have problems. Besides, if the card goes bust, they pay for it via warranty.

IIRC they are very similar in the hard thermal paste tests with the AS5 maybe edging it out a tad.
 
I rarely check with low power CPUs that usually use the stock heatsinks. They're generally pretty damn good with getting solid contact with the CPU, and haven't had an issue yet. As for chunky tower heatsinks, on high end OC chips, lapping, and checking for good contact is a must.

I learned my lesson back with my first Q6600. Installed the stock heatsink, ran some stress tests, and it would hit it's thermal cutoff only using two cores. I figured that the stock heatsink simply wasn't capable of handling the heat. Purchased a TRUE120, re-installed, and exactly the same results. Something was obviously up :p

The CPU itself was concave (approx .5mm deep in the center compared to the edges) and the TRUE120 was concave as well (almost 1mm deep in the center compared to the edges). God knows how long it took to lap both of those to something resembling flatness. Overclocked that beast to 3.8ghz, never getting anywhere near thermal cutoff after that :)
 
First thing I always do when I fire up a system for the first time is go straight to bios and check the temps. I'm pretty sure you won't ruin a system if you do that immediately. Also it helps to not get Nvidia Chipsets cause some of them run pretty insanely hot and sometimes require better/extra cooling to last a long time.

+1

I check temps and voltages before I even start installing the operating system. I have an Nvidia 650i Ultra that runs hotter then hell. I got a HSF for the southbridge and a fan for the northbridge and that helped a lot.
 
I don't trust stock thermal paste/goop either, but uneven/hard paste on a new board is normal since it hasn't heat cycled yet to settle everything in.

if you're responding to my comment, then i believe i was misunderstood. one time i replaced a dead chipset fan on an old nf7s, that sure had time to settle although this was years ago...anyway it was practically a solid and was splattered onto about 70% of the chips surface. i admit, i shuddered...
Hmm, how well does this work compared to the latest Arctic Silver?

ceramique blows away stock pastes and performs very competitively compared to arctic silver 5. you could theoretically see anywhere from 0.5-2°C load temp difference, as5 coming out on top, but in my (limited) tests they've rendered identical results. plus ceramique is thicker and stays exactly where you put it.. takes a long time to break down, doesnt separate, and also isnt conductive or capacitive unlike as5 so its perfectly safe to use around sensitive traces or even on ramsinks. i like it a lot!
 
Don't let this crappy website fool you, this is a decent site

http://www.bestbyte.net/merchant/me...=BB&Product_Code=COTIASC22&Category_Code=COTI

$5 for a 22G tube of Ceramique... yes 22grams... your average arctic silver tube is 3.5 gram for $5-6 ? Same price for 6x as much paste... :D Also big syringe is much easier to dispense little amounts vs. the tiny arctic silver tube. Not saying it's worse, just this is a good deal.
 
I am now going to check the sinks on every motherboard i get. The one on my MSI SLI-FI is pretty loose and is extremely hot (not OCing now due to recent computer failure which i believe is due to NB being to hot).
 
If you are watercooling an i7, do you think it's necessary to lap the CPU?

Thanks.

x509

Well, depends on the CPU you get. Most CPUs I've gotten in the past year (several dozen) have been flat enough that lapping won't be of noticeable benefit. It's just those strange irregulars you get that are out of shape enough that lapping them will be of great benefit. Now, whether you're watercooling or aircooling, if you have such a chip, it's worth the bit of effort :)
 
I am now going to check the sinks on every motherboard i get. The one on my MSI SLI-FI is pretty loose and is extremely hot (not OCing now due to recent computer failure which i believe is due to NB being to hot).

if your ramsink is hot then your TIM isnt the problem. its your temperatures. the heat is transferring to the heat sink just fine, but theres just too much of it. try lowering the applicable voltage, or else get a bigger heat sink or a small fan to put on the existing heat sink. zip tying a 40mm fan to a nb heat sink does wonders for very little money
 
if your ramsink is hot then your TIM isnt the problem. its your temperatures. the heat is transferring to the heat sink just fine, but theres just too much of it. try lowering the applicable voltage, or else get a bigger heat sink or a small fan to put on the existing heat sink. zip tying a 40mm fan to a nb heat sink does wonders for very little money

Thats the thing i had a tiny little CPU cooler fan from an older socket on there and it over heated. I wish to know why its overheating now and not 2 years ago when i OCed it. Its at stock speed now :(.
 
Thats the thing i had a tiny little CPU cooler fan from an older socket on there and it over heated. I wish to know why its overheating now and not 2 years ago when i OCed it. Its at stock speed now :(.

you said the actual heatsink is hot right? like hot to the touch. that is kinda weird. are you sure some voltage didnt get cranked way up on accident?
 
you said the actual heatsink is hot right? like hot to the touch. that is kinda weird. are you sure some voltage didnt get cranked way up on accident?


I don't know how it could've been. I don't remember what i had it set to when i origionally OCed but it wasnt very much at all. Something like .0002 volts or something, maybe more but it ran fine. My computer is kinda just all of the sudden crapping out on me. And yes the sink is wiggly but it is very hot, well was, my ram is nice and cool now that im clocked down but the NB still seems kinda hot for stock.
 
As soon as I hook up the motherboard to make sure it will boot, I take everything back apart to pull the heatsink off of the northbridge and apply new thermal goop, currently MX-2. I've pulled off more northbridge heatsinks than I care to remember over the years and practically none of them had a good application of thermal goop. Therefore, as soon as I know the board is working, I replace the thermal goop especially since northbridges like to get a bit warm while overclocked nowadays.

Since some southbridges also get warm, I'll do the same for the southbridge heatsinks if it's easy to remove and put back on. It's generally not critical to keep the southbridge as cool as possible but it can help every once in a while.

I figure if I'm going to put the effort into making sure the CPU is as cool as possible for overclocking, I should do the same with the northbridge as it's just as important for overclocking as the CPU.

 
Back
Top