CPU cores are 10'c different using Cooler Master Hyper 212 Plus

[Andrew LB]

Ok. So I'm stumped. I've been building computers for 20 years now and have NEVER once had a CPU cooler which gave me a full 10'c difference in idle temperatures between two cores on the same CPU die.

I'm throwing together a budget PC for a friend and bought a e7500 C2D running at 3.5ghz, 780i SLI FTW EVGA motherboard, Antec 300 case, and a Cooler Master Hyper 212 Plus CPU cooler.

I've mounted, re-mounted, and even tried adding shims to the darn heatsink so it doesn't have uneven pressure and I'm still getting the same results. Both cores are at 3.5ghz, with one idling at 35'c (no speedstep) and the second at 45'c.

I figured since the cooler basically is held in its place by a single pin sticking out of the adjustable "X" type mount bracket, that perhaps its just not balancing correctly. So I added shims to ensure it was balanced, and no luck.

Oddly enough, I swapped back in the horrible stock Intel cooler and both cores stayed the same temperature, at around 50'c.

Any ideas to what I may be missing here? I've never had an issue like this.

[Captain_Harlock]

It sounds like the cooler has a not so perfect cooling surface, as in one of the heat pipes is sitting proud from the rest. It may also be that the retention system is warped, but I would assume the problem lies on the cooler and not the mount. You cant compensate for something like that. If it were the mount, then a washer (or shim) in the right spot would have shown some sort of result. So you can either return it and get another (the odds on another bad one are pretty slim), or you can lapp it. If you go the lapping route, then make sure to lapp the CPU as well. The CPU should be lapped to at least the same grit as the cooler. If you decide to lapp it then let us know and I'll be more than happy to give some instructions on how to lapp an exposed heatpipe cooler like the 212. But given the bother with lapping, I would recommened an exchange instead. Lapping isnt hard, but you'll spend $10 to $25 for the parts to do it unless you already have them and thats money you could have spent on a better cooler to begin with. Whatever you do, dont try to reseat the pipe if you can find out which one is causing the problem. They are soldered in to the aluminum retainer and it will simply crush or bend instead of seat further into the retainer.

[zalbard]

Possibly a heatpipe issue or
the CPU's heat spreader and the heatsink's base are uneven in a bad combination causing poor contact.
I don't think you want to bother lapping both so I'd just get a different heatsink...

[Jacka]

I'd say they're just inaccurate temperature sensors.

[Captain_Harlock]

It could be that one core is just that much hotter than the other. But even at 45c-50c, you are still getting acceptable temps. The C2D line can run safe at up to 100c. You're getting a decent OC out of it and still with plenty of headroom on temps. I wouldnt worry about it unless the fan is running at max speed and you are looking to drop the volume a bit. If you can get it up to 4GHz and keep the temps below 75c under load, then everything should be fine.

[zalbard]

I'd say they're just inaccurate temperature sensors.

But he has no such issue with stock cooler...

[Jacka]

The stock cooler can't keep it at such low temperatures. Think about the proximity of the cores on the die. Could they really have a 10C temperature difference?

[NEOAethyr]

Yeah they could have that much of a diff, seen it several times with older amd 64bit cpu's.
It's the cpu lid and the gunk they use for a tim that causes that.

In your instance, it would be the lid on your intel chip.
The only way you're gonna fix it is to remove that lid or work on improving the cooling in a diff way, like lapping the lid or heatsink.
Diff fans, better heatsink, etc.

If it doesn't give you any trouble then I wouldn't worry about it, I've only heard of one cpu failing because of this prob with the tim putty gunk between the lid and cpu.

Isn't your cpu supposed to have solder of some kind instead of the putty gunk that amd use though?
That would be kinda odd if they screwed that up...


Edit:
In your case it must be the heatsink though it seems...

[Buckeye]

I had something like this happen to me with a 920. I was using a Mega Shadow for Air Cooler and had 1 core with temps like you are talking about.

It was off by so much and started creating issues at higher OC's. So I ended up RMAing the CPU and the replacement worked just fine.

[Captain_Harlock]

Intel's stock coolers are not bad coolers at all. They just limit OC'ing and when at their max temps get very loud. But a stock cooler should be able to keep an e7500 under 65c provided the vcore is low enough for that chip. Of all my rigs liquid or air cooled, one of my quietest is my HTPC and its a stock intel cooler on a little e2180 OC'ed at 3.6GHz (at stock voltage). Keeps it at a nice 55c-60c but at medium speed. Drops to low 50's on max speed but the noise is really loud. So yes, that little round contact point on their stock cooler is able to get the job done and do it quite well. But when you have a chip that needs a higher vcore, then you turn to aftermarket coolers.

As for the cores being as much as 10c off from each other, yes. I've seen many a forum post (maybe a dozen or more), over the years on that exact problem. Seach over at OCF and you'll find a few. But in all, its pretty rare. No common reason is ever given other than the generic "every chip is different", but I'd hazard a guess at it being due to the way the IHS installed. When its placed on top of the cores, its done with a low melting point solder and under pressure. If the pressure was not equal when it was placed on, it could have left a thin game in the solder between part of the core and the IHS. This can case the part of the chip under that gap to be far less effecient in heat transfer to the IHS and you end up with crazy numbers like that. Again, just a theory I've had for a few years on this exact problem when ever I see it. No way to really tell as you would have to re-melt the solder to pull the IHS off and that would destory the gap by causing it to fill itself back in.

Now the OP could always pop the IHS off and go back to his stock cooler, but its simply not worth a the risk of a cracked core when he can get by with those temps just fine.

[Andrew LB]

It sounds like the cooler has a not so perfect cooling surface, as in one of the heat pipes is sitting proud from the rest. It may also be that the retention system is warped, but I would assume the problem lies on the cooler and not the mount. You cant compensate for something like that. If it were the mount, then a washer (or shim) in the right spot would have shown some sort of result. So you can either return it and get another (the odds on another bad one are pretty slim), or you can lapp it. If you go the lapping route, then make sure to lapp the CPU as well. The CPU should be lapped to at least the same grit as the cooler. If you decide to lapp it then let us know and I'll be more than happy to give some instructions on how to lapp an exposed heatpipe cooler like the 212. But given the bother with lapping, I would recommened an exchange instead. Lapping isnt hard, but you'll spend $10 to $25 for the parts to do it unless you already have them and thats money you could have spent on a better cooler to begin with. Whatever you do, dont try to reseat the pipe if you can find out which one is causing the problem. They are soldered in to the aluminum retainer and it will simply crush or bend instead of seat further into the retainer.

Oh... I forgot to mention that I did lap the heatsink after the first three attempted re-mountings. Only after that did I shim (with home made washers made from copper slugs (basically a quarter shaped disc of copper) which i drilled a hole in it to ensure the cooler would stay perfecty centered.

Lemme check one thing. I just remembered how much the backplate stuck out behind this cooler. My guess is, it's hitting something....

stand by..

[Andrew LB]

Ok.... here's the deal. I just popped off the back panel and behold!!! The stupid motherboards tray cutout is not large enough to fit the cheaply designed Coolermaster HSF retention bracket. So now i'm going to take the motherboard out, bring the case to the garage, and bust out an assortment of air tools. I've got a 3" high speed die grinder with steel cutoff wheels that will make very quick work of the offending section of metal.

I was also going to paint the case insides black but I already applied some Dynamat Xtreme to many surfaces inside the case panels. It actually made a huge difference in noise.

[killerxp100]

I had a similar issue with an old freezer 7 pro, turned out one of those cheep push-pins was broken & didnt latch down.

[NEOAethyr]

Glad you got it sorted .

Nice heatsink airflow arrangement btw, alot of people buy heatsinks that face towards the top and bottom of the case.
Yours blows the air out like it should.

I'm thinking about the corsair h70 just because of that reason...

[Andrew LB]

Actually.... it isn't sorted out. Last night I got out the power tools and cut a section out of the motherboard tray so the bracket wont hit and since I had it apart, decided to cut out the rear exhaust grill and top exhaust grill. This seems to have cut the noise from these fans by about 50%. Less restrictive metal mesh, less noise. I also painted the back of the case with some Rustoleum gloss black high temp enamel. Great paint!

As for the CPU temps.... they are STILL 10'c apart. For kicks, i've got one of those Dynatron 1U copper micro-fin CPU coolers around here somewhere which i'll install to see if that gets the same results since it uses a back brace. The intel cooler had push pins.

Pics....

[Captain_Harlock]

So whats the status on the rig? Did the Dynatron give you even temps accross the cores? As for cutting out the mesh on the 300, that was the first thing I did with mine a couple of years ago. I use one for a workstation (the orginal version with the 3 1/2" HDD hotswap bay). Best case for cooling for $35 when they are on sale as they take mods well and allow plenty of targeted airflow. Its hard to find the 3 1/2" original version now. Most places only stock the newer version that has a 2 1/2" HDD hotswap bay and rarely drop the price below $55. I havent figured that one out yet, lol.

Anyway, your board had a bretty big bow to it from the looks of the pics. I wonder if that was causing some mounting issues?

[Andrew LB]

^^^ That bow was caused by the nut from the heatsinks retention setup hitting the motherboard tray, which is why I cut out that section of metal directly behind the CPU socket. Once that was done, everything was perfectly flat.

Unfortunately the damned CPU temps were still 10'c apart after all that work. I later tried removing the board from the case and attached a Zalman CNPS9900 NT cooler and STILL got the same damned result. I went down to Microcenter that afternoon and picked up a Q9300 quad core that was on sale for $90.00 and have never looked back.

The offending bizarre temp CPU is currently a paperweight on my desk. I'm going to sand the crap out of it so the heatspreader is paper thin, and if that wont work... off comes the heatspreader.

[rintamarotta]

The problem is your cpu's bad IHS (heat spreader), from inside its not equally distributing the heat to the heat spreader when under weight by H212Plus (weight point further distance away from the cpu than with stock intel cooler), intel stock cooler didnt have this 10'c difference because its so light compared to Copper/Aluminium that is in CM H212Plus, so it will not cause un-equal head spreader contact with cpu cores.

I had same problem and i tried with intel stock cooler in one of my machines, temps were equal but soon as i added bigger NH-D14 cooler it did show un-equal temperatures in cores.

Four ways to fix it.
1. You need remove cpu's heatspreader and re-mount it with non-electircally conductive thermal glue to edge of the heatspreader and core with better quality thermal paste like AS5 or Noctua thermal paste and then apply loads of weight to the cpe till that thermal glue has glued the ihs to the cpu, be shure that contact of the IHS is good with the cpu cores.
2. Support your cooler with something like zip ties or similar from top of the case so it spreads its weight equally across the cpu.
3. Buy low profile cooler with weightpoint closest to cpu possible.
4. Buy CPU that has better weight distribution by IHS under higer weight.

E7000 series cpus have allways had bad IHS due intel cutting from the IHS quality to save money in lower end cpu's, its enought for intel that IHS is good enought to stay under certain temperatures with intel stock cooler.

Retention mechanisms in newer cpu coolers are trying to fix is by distributing the weight of the cooler more higer in cpu.

[Andrew LB]

Ok. I'm going to give this chip thats been sitting on my desk another chance at redemption. I just got done lapping the heat spreader on the CPU, making it a point to remove as much material as I could in order to thin out the metal. I started with 150 grit Norton and worked my way up to 4,000 grit pink "polishing paper" made by 3m. Just to see how reflective I could get it, i buffed it out with two grades of polishing compound that I have around for finishing off automotive paint work.

Here is the result. The reflection is of my right monitor which had this photograph from another thread on lapping. Not too shabby. If this doesn't help, then i'll probably remove the heat spreader and do a direct core mounting of the heatsink. I'll be testing the chip in my little mini-itx media server since the board lays flat and the CPU cooler is a 1U solid block of server cooling copper and always mounts up perfectly.



Here is the little server PC. (currently a Q6600 @ 3.4ghz, blu-ray, 4gb DDR2, 2tb HDD, 9800gt 1gb overclocked, all packed into a tiny Thermaltake Element Q case)

[little_scrapper]

Nice heatsink airflow arrangement btw, alot of people buy heatsinks that face towards the top and bottom of the case.
Yours blows the air out like it should.

Alot of people have fans that blow out the top as well.