Pentium G3220 + Hyper 212 = Bad performance

[Frick]

Hey hey.

So I recently got a decent deal on a CM Hyper 212 Evo and thought what the hey. Let's do it.

The thing is I can't make it not max out at above 60 C during prime95, which is worse than a Celeron stock cooler. The CPU idles at about 40 C.


The problem is poor contact; the heat pipes barely warms up (and removing the fan altogether doesn't really make a difference). I've reseated the cooler several times, experimenting with thermal paste (MX4). I can't tigthen it any harder, and playing around with the screws didn't help either. I have this case, so at the front is a honking big 200mm fan pushing air into the case, the CPU fan pushes it through the heatsink and the stock fan at the back pushes it out the case. They fans are all mounted the right way.

The plastic from the cooler base is gone, yes. I've checked.

The screenshot is an avarage case. I've tried various paste application methods and amounts and it still runs that hot.

In reality it doesn't matter much as it's still within the limits and the screenshot is from prime95+Furmark so unrealistic heat buildup, but still. It's annoying.

Any suggestions?

 

[Jetster]

Is there marks on the bottomm of the cooler where the CPU corners touch? Concave? Is the cooler flat?

 

[Beertintedgoggles]

Does the paste look properly / evenly spread over the CPU when the heatsink is removed? If not it could have a warped heatspreader on the CPU that only lapping would fix (or it could indicate the base of the heatsink is also warped which if it is a direct touch setup with the heatpipes I've had good and bad luck with lapping those). Otherwise, it sounds like it has bad contact between the cpu and the heatspreader which only delidding and repasting could potentially fix.

Best suggestion, uninstall CPUID.

 

[Ferrum Master]

Delid it!!

 

[NdMk2o1o]

This is stock temps? no OC?

 

[silentbogo]

Delid it!!

Just wanted to add: is it possible that this CPU was de-lidded before you got it?
Because heatpipes can't stay cold or barely warm with those temps, even if there is no full surface contact.

May need to crack it open.

 

[Frick]

This is stock temps? no OC?

OC not possible.

Just wanted to add: is it possible that this CPU was de-lidded before you got it?
Because heatpipes can't stay cold or barely warm with those temps, even if there is no full surface contact.

May need to crack it open.

Dunno, it used to sit in Rcoon's HTPC. The Celeron stock cooler I used before behaved ok, so I'm inclined to believe it's something with the cooler, unless I'missed something obvious (which isnt unlikely).

And they're not cold, they are just not as warm as they definitely should be. The VRM's on the motherboard provides most of the heat the back fan pumps out.

 

[silentbogo]

Did you check the TIM print? Is it OK.

 

[RCoon]

OC not possible.



Dunno, it used to sit in Rcoon's HTPC. The Celeron stock cooler I used before behaved ok, so I'm inclined to believe it's something with the cooler, unless I'missed something obvious (which isnt unlikely).

And they're not cold, they are just not as warm as they definitely should be. The VRM's on the motherboard provides most of the heat the back fan pumps out.

Yeah, sat in my htpc with the stock cooler on it for like, a year. It's had a comparatively easy life when put next to my other hardware.

 

[qubit]

@Frick Well, it looks like you've figured out the source of the poor cooling, the poor surface contact, certainly the symptoms you describe indicate it. There's really nothing you can do to improve the cooling performance unless this contact problem is fixed.

To me, it sounds like a manufacturing defect or a poor design, but I suspect the former much more. I'm curious what sort of deal this was. Could the seller have known this cooler had this problem and that's why it went for cheap? I mean really, there's no way that this tower cooler should be as bad as the stock cooler in cooling performance. I suggest returning it as faulty and asking for another one or preferably a refund.

 

[NdMk2o1o]

OC not possible.

My bad thought it was one of those pentiums you could OC that came out recently

@Frick Well, it looks like you've figured out the source of the poor cooling, the poor surface contact, certainly the symptoms you describe indicate it. There's really nothing you can do to improve the cooling performance unless this contact problem is fixed.

To me, it sounds like a manufacturing defect or a poor design, but I suspect the former much more. I'm curious what sort of deal this was. Could the seller have known this cooler had this problem and that's why it went for cheap? I mean really, there's no way that this tower cooler should be as bad as the stock cooler in cooling performance. I suggest returning it as faulty and asking for another one or preferably a refund.

I'd also be inclined to return it as faulty if you can confirm by putting it on another board? no need to get a refund and something else as they're well known coolers and perfectly adequate could be a case you just got a dud!

 

[qubit]

I'm saying preferably a refund as I suspect that the seller flogged a dodgy batch for cheap, so another one probably wouldn't help. We really need to know more about the deal to say with more certainty.

 

[xvi]

As a homebrew fix, perhaps try placing a bit of a shim between the backplate and motherboard? Should pull the mount back and allow you to tighten it more.

 

[Frick]

@Frick
I'm curious what sort of deal this was. Could the seller have known this cooler had this problem and that's why it went for cheap? I mean really, there's no way that this tower cooler should be as bad as the stock cooler in cooling performance. I suggest returning it as faulty and asking for another one or preferably a refund.

It was a sale from a reputable store, so no shenanigans involved.

 

[qubit]

It was a sale from a reputable store, so no shenanigans involved.

In that case try for another one.

 

[SnakeDoctor]

How long are you running benchmark to get to 60c?
I don't think the max temp is too bad depends how long you been priming for , +-5c would be better.It should'nt get to those temp on day to day activities.

I using some cheap cooler master paste ,max at the moment 55-59c@4ghz priming
I don't bother too much on brands of cpu termal paste as most do , as long as not fongkong , got a tub of entry level cooler master for servicing clients pcs.

 

[Norton]

The problem is poor contact; the heat pipes barely warms up (and removing the fan altogether doesn't really make a difference). I've reseated the cooler several times, experimenting with thermal paste (MX4). I can't tigthen it any harder, and playing around with the screws didn't help either.

How much wiggle does the cooler have when you lock it down and did you check or adjust the center screw?

 

[Kursah]

I'm sure you removed the plastic protective sticker off the base of the cooler right?

I've used a lot of 3220s and even on stock cooling they don't get that hot... in Prime or OCCT maybe....depending on airflow and ambient temps. A 212 shouldn't be anywhere near that bad performing. I'd almost wonder if you have a bum chip. You got another system to test the 212 in?

Only things I can think of is maybe the CPU IHS has a bubble or missing section in the TIM between CPU and IHS, plastic cover on the 212 is still there (you would've noticed that after reseating several times tho...so this one is likely out), or maybe somehow the 212's bracket isn't allowing a good seat. The 212 should have some wiggle room..it doesn't clamp down to the point you cannot twist it slightly, so a little wiggle ability is normal for these in my experience.

If different seatings and orientations make no change...try a different CPU or different build with that cooler to verify. IMHO that'll tell you quickly if its the CPU or cooler.

 

[ThE_MaD_ShOt]

Also make sure the center pin on the x bar is seated properly in the cooler and the pin on the cooler is seated in the notch of the x bar. Also double check you are in fact using the right set of stand off screws. There should be 2 sets in the pack, one for 2011 and another for everything else. Also is your cpu and 1150? If so double check compatibility and when I look at the manual I don't see 1150 listed in the install instructions.

 

[alucasa]

Also, for the kick of it, threaten the hell out of the cooler.

"You are worse than Intel stock cooler, you metal piece of s...."
"I swear I am going to make you cool an AMD FX-9590. Do you really want me to do that to you? Huh???"

Something like that.

 

[fritoking]

I read that you can get a nylon or plastic washer and place it between the crossmount and the cooler base OVER the center pin to give you a bit more snug fit ...maybe try that ?

 

[AsRock]

@Frick Well, it looks like you've figured out the source of the poor cooling, the poor surface contact, certainly the symptoms you describe indicate it. There's really nothing you can do to improve the cooling performance unless this contact problem is fixed.

To me, it sounds like a manufacturing defect or a poor design, but I suspect the former much more. I'm curious what sort of deal this was. Could the seller have known this cooler had this problem and that's why it went for cheap? I mean really, there's no way that this tower cooler should be as bad as the stock cooler in cooling performance. I suggest returning it as faulty and asking for another one or preferably a refund.

This^

And would not go and delid it as if it's the coolers fault ( which it sounds like it, and the stock cooler is works ok ) putting some thing on a uneven on the cpu core could have tragic consequences.

 

[rtwjunkie]

@Frick, I know what your OP said, and I didn't want to insult your intelligence, but I simply musk ask, are you sure it's an Evo and not a Plus?

The reason I ask is because Plus has HDT, with the actual heatpipes flattened and forming the base, with strips of cooler base in between the pipes. In there you have gaps, which if not filled with TIM you have air in between, instead of one, smooth heat-transferring surface.

I noticed behavior like this on the Plus in similar heat situations, which were fixed by filling the gaps.

If it is an Evo, then the surface is not flat, as others have suggested.

 

[Jetster]

I'm still saying something is not straight. Lap that CPU and HS

 

[ne6togadno]

make sure you've setup bracket porperly (follow printed instructions very carefuly) and tighten mid screw a bit.