Sign-in / Sign-up
Your question

CM Hyper 212+ high temps - help!

Tags:
  • Heatsinks
  • Intel i7
  • Overclocking
Last response: in Overclocking
January 3, 2010 6:03:22 AM

Hi,

I just built an i7 860/MSI P55-GD65 based machine and I'm having trouble with temps that are lot higher than what reviews of the CM Hyper 212+ said I should expect.

At stock speed under 100% load with Prime95 I'm seeing temps go to almost 70C. I just pulled the system apart and reseated it again, making absolutely sure everything was very tight and I'm still getting really high temps like that. I applied AS5 per the benchmarkreviews.com method for HDTs.

Really baffled here, I can't even try OCing because of this, it'll go over the safe temp limit instantly.

Any help is appreciated!
Ryan

More about : hyper 212 high temps

a c 198 K Overclocking
January 3, 2010 11:37:53 AM

What kind of case? How is the air circulation through the case?
m
0
l
Related resources
January 3, 2010 5:05:19 PM

1. I've done this once already - I'm pretty positive it's on right.

2. It's a Coolermaster CM Storm Sniper. Cables are managed well and there's a lot of airflow. I tried pointing a table fan at the case with the side panel off and it made no difference at all.

3. That's basically exactly what I did - I didn't use a razor blade, but I did fill in the grooves and then wiped the excess away, then did 3 lines of AS5 down the pipes. (modifying what was in the benchmarkreviews instructions because it looked different from the cooler they'd used)

4. CoreTemp and RealTemp.


Check this out - this sounds almost exactly like what I'm experiencing:
http://forums.anandtech.com/showthread.php?t=2019595

These guys basically said the sensor was wrong and a BIOS update fixed it. I have the latest BIOS, so not sure how I can try that though.

My idle temps at stock are extremely low - I'm seeing like 19-22C or so. Almost instantly with load it shoots up to 70C.
m
0
l
a c 86 K Overclocking
January 3, 2010 8:52:26 PM

Check your stock voltages. Is the Mobo on Auto? Sometimes the Auto makes them higher than they need to be. Read about it in the CPU forums or Google it.
m
0
l
January 3, 2010 10:49:01 PM

I just installed the CM Hyper 212+ in the exact same case (CM Storm Sniper) and it's performing very close to my broken POS water cooler (Asetek LCLC w/240mm radiator).

On a core i7 920 D0 stock speeds I get idles in low 30's and a peak of 64C after an hour of Prime95. Measured with Realtemp, ambient 67F.

I just bought it to hold me over until I get something fancy or RMA my Asetek, but I'm pretty impressed for $30.

Check the seating (the block still wiggles around even screwed into the backplate) and try reapplying thermal paste. Last time I did it was years ago so I'm no master on technique, but I put a pea on the cooler surface, spread it evenly w/my finger covered in some saran wrap, then put a small pea on the CPU and slapped it on.
m
0
l
a b K Overclocking
January 3, 2010 11:54:14 PM

Quote:
spread it evenly w/my finger covered in some saran wrap, then put a small pea on the CPU and slapped it on. [/quuote]
That's exactly what I do too!
m
0
l
January 5, 2010 7:27:12 AM

Ok update - I reseated it again tonight using CM's paste instead of AS5 and I'm seeing about 10 degrees lower now. Better, but I still go over 80 almost instantly in Prime95 with even a 3.5GHz OC.

RealTemp sensor test seems fine to me, nothing is stuck.
m
0
l
a c 86 K Overclocking
January 5, 2010 12:59:40 PM

Are your volts on AUTO in the bios?
m
0
l
January 5, 2010 6:46:37 PM

Yeah, it's on auto, CPU-Z reports about .9-1.0v at idle and around 1.24 under load.
m
0
l
a c 86 K Overclocking
January 6, 2010 12:31:09 AM

So make it un-auto Number 1! You don't need that much for sure.
m
0
l
January 8, 2010 2:34:46 PM

Hey Tabs im on the same boat. Just bought the same heatsink for my i7 860 after reading the great reviews. Im also upset that im not getting the expected results.! At stock im at idle 29-31c and 59-61c full load. I did a mild OC @ 3.36 and my full load temp went up to 67-69c??? Full FFT prime95. Supposedly at around this mild of an OC it should be in the mid to high 50's...?? My abient is 21c and i have a antec 902 case so airflo is decent enough that i didnt expect 69c @3.36. Voltage is .8-.9 idle and tops out at 1.24 at full load. nothing that i think should be too much to bring me up to 69c, if im compareing it to the reviews..... Did you solve it? Or was it all hype for this cooler?
m
0
l
January 18, 2010 3:14:00 AM

Haven't solved it... I don't see what I could possibly be doing wrong, I'm not new to building systems and installing heatsinks and stuff. Might just have to order a megahalems or whatever and see if that fixes this...
m
0
l
a b K Overclocking
January 18, 2010 12:23:55 PM

I installed a Hyper 212+ last week with good results. My max temps under IBT or Prim95 with 8 threads running dropped from 80-85 C with the stock fan to around 51 C with the Hyper 212+. I believe my idle is in the low to mid 20's. My ambient temp is around 19-20 C.

Fisherman, What method did you use to apply the paste? I followed most of the suggestions out there for applying the paste, filling in the channels between the pipes first, using a razorblade to make sure they were filled in and the base otherwise clean. Then I put two small, thin strips of paste evenly spaced on the channels just off of center. Basically as shown in the example at the bottom of this page: http://benchmarkreviews.com/index.php?option=com_conten... I assume you have the fan mounted in a Push configuration, pushing air through the cooler and out the otherside where you case's exhaust fan should be working to remove the hot air that comes through the cooler.

At stock I would expect temperatures somewhere around there. If you haven't set your Voltage manually yet, give that a go. The board is probably supplying too much V which is overheating your system. Also, make sure you have the latest BIOS update. For the Gigabyte, set your CPU Core Voltage to Normal, and then use the DVID offset to indicate what additional voltage you want supplied.
m
0
l
January 19, 2010 1:38:49 AM

Ok looks like the Auto voltage (1.24v) was the culprit.! I am now @ 3.4ghz- 1.17v stable with temps in the 57c-59c under full load P95.
I put a stripe of AS5 down each pipe. After taking it off and redoing it 3 times, i know its not the mount or the AS5. Each time i took it off I looked and it was nice and evenly spread. I wonder why in Auto it supplies 1.24volts? Isn't that just way to much??? I got 3.4ghz with only 1.17v. Whats with Auto giving you so much juice??? So then should i take everything off auto??
m
0
l
a b K Overclocking
January 19, 2010 12:24:40 PM

Fisherman619 said:
Ok looks like the Auto voltage (1.24v) was the culprit.! I am now @ 3.4ghz- 1.17v stable with temps in the 57c-59c under full load P95.

How are you measuring your V? CPU-Z? And just to be clear, that's Vcore right?
I'm also trying to figure out what BCLK you're using. Are you at 3.4GHz with Turbo on and multiplier = x21, which would put your BCLK at 3400/21= ~162? Or with Turbo off and multiplier at x22, putting your BCLK at 3400/22= ~155?

Those Vs look good. Auto tends to over do it. I've been slowly tweaking my voltage down, with a mild BCLK bump to 145 and Turbo on. Based on your findings I think I might be able to go lower. So far, testing with Prime95 I've hit:
Idle: x9 Multiplier, ~1.197 GHz, 0.848 V Core Voltage
8 threads: x22 Multiplier, ~3.190 GHz, 1.120 V Core Voltage
1 Thread: x26 Multiplier, ~3.768 GHz, 1.216 V Core Voltage

Quote:
Whats with Auto giving you so much juice??? So then should i take everything off auto??

I've been trying to figure that out, too. Most people I ask just say they keep it at stock. And if I try to clarify does that means they keep it at Auto I often get a smart-ass answer or none at all. Which leads me to believe a lot of people use Auto. But I don't think that's necessarily the right thing to do. As you've seen, Auto can overdo it. Hopefully some others reading this can weigh in. Should all Vs be manually set? If only some are important to set, which ones?
m
0
l
January 19, 2010 10:28:47 PM

I stopped using auto for overclock voltages after my first study using my gigabyte mobo. After reading so much marketing from gigabyte, I did give their oc app a try....only to find that the voltages were far beyond what I wanted. Thanks to Andy's help, I first looked at vtt...it was set by gigabyte to 1.25v, even thou the intel spec max for i7 860 is 1.21v. After learning the role of vdroop in protecting against transients, and the significant impact of vdroop on the difference between voltage set in bios and voltage seen by the cpu, I wouldn't want "auto" to keep me from controlling the bios set voltage, especially since its possible to enable Load line calibration independently from the setting in bios for vcore. Having said that, if you do enough study of what auto does to voltages in each likely scenerio, it doesn't hurt, but that seems like it would be more work than just controlling voltages directly.
m
0
l
January 20, 2010 5:49:20 AM

Yup thats Vcore. I use CPUID HW Monitor and Real Temp to monitor. I use increments of 5 when upping my BCLK. 165x21= 3.46ghz Thats with turbo off. Hyper threads enabled. C1E disabled. EIST disabled. vcore set to 1.175. Timings and voltage of ram set manually to specks. all else is auto. Oh load line calibration enabled. Thats it. Prime95 small fft's 13hrs 0 errors. Large fft's 9hrs 0 errors. Blend 7hrs 0 errors. This is my 24/7 setting. I dont need much as all i do is game and basic PC use so im happy with this little bump in power.!

Quote:
I've been trying to figure that out, too. Most people I ask just say they keep it at stock. And if I try to clarify does that means they keep it at Auto I often get a smart-ass answer or none at all.

LMAO!!!! So True!!! I dont know why someone cant just type Stock=???v. Isn't that easier than the- Go search the forums for 167398 hours and find it!! This is a place for asking Q's and helping others right?? I still dont know what stock vcore is! :pt1cable: 
m
0
l
January 20, 2010 3:48:46 PM

ekoostik can you please post your temps. I want to compare them to mine. TIA.
m
0
l
a c 86 K Overclocking
January 20, 2010 4:24:05 PM

Stock voltages vary by mobo, the vdroop of the mobo and other issues.

You build the system, stock overclocks, and read the voltages using something like CPUZ. Thats your stock voltage at idle. Auto functions on many newer mobos vary wildly.

So it's not that simple. There is NO easy answer.
m
0
l
January 20, 2010 5:53:09 PM

Thanks for clarifying. I thought there was a base # like there is for max.
m
0
l
a b K Overclocking
January 20, 2010 6:53:46 PM

Fisherman619 said:
ekoostik can you please post your temps. I want to compare them to mine. TIA.

I'm not going to have access to that PC much in the next couple days so if I don't get to it right away and I forget, send me a PM and I should be able to check later in the weekend.
m
0
l
a b K Overclocking
January 20, 2010 7:34:02 PM

Fisherman619 said:
Thanks for clarifying. I thought there was a base # like there is for max.

I started a poll to find out in general what Voltage settings people change and to get some more discussion. Anyone who wants to participate can see it here: http://www.tomshardware.com/forum/256532-29-which-volta...

I've been reading the Intel Data Sheet for the P55. I started putting together a list of ranges for the Voltages. But I ran into trouble on a couple items. If anyone can fill in the details, or spot anything I got wrong let me know and I'll make the edits. I realize some people push past the abs max but I still thought it'd be helpful to put together. This is what I've come up with:

Setting - Minimum - Typical - Maximum - Absolute Maximum
DRAM - 1.425 - 1.500 - 1.575 - 1.650
QPI/Vtt (IMC) - 1.045 - 1.100 - 1.155 - 1.210
PCH Core - ? - 1.05? - ? - ?
CPU PLL - 1.71 - 1.80 - 1.89 - 1.98
Vcore - 0.65? - ? - 1.40? - 1.55

It was when I got to Vcore that I realized I took on more than I expected. I based the values above on feedback from other sites, reading it from the Data Sheet has me scratching my head. Maybe this whole exercise was a waste of time.

Intel Data Sheet reference: http://download.intel.com/design/processor/datashts/322...
See Tables 7-4 and 7-6.
Some voltage explanation: http://www.bit-tech.net/hardware/cpus/2009/09/21/overcl...

This OC article cracked me up, I think someone redacted the numbers! "Eventually I found that the CPU voltage had to go back up to 1.500 volts. I adjusted CPU PLL to 1.xxxx with PCH voltage at its maximum value of x.xx. IMC voltage had to be set to 1.xxxx volts. " http://www.hardocp.com/article/2009/09/09/asus_p7p55d_d...
m
0
l
January 21, 2010 11:20:45 PM

Correct me if I'm wrong but I've read that one should keep vdimm within .5v from vtt. So, you can see that if you raise one significantly, you must raise the other also. How do you differentiate between "maximum" and "absolute maximum"? I haven't read anywhere that I'm at risk running vtt at 1.21. I can add my settings if you wish: vdimm: 1.64; vtt: 1.21; vpch: I'll have to check this one, I raised it one notch, but haven't really paid attention to it since; vpll: 1.86; vcore: 1.2875 (llc disabled).
m
0
l
a b K Overclocking
January 22, 2010 1:32:04 AM

Thanks for adding your voltages. I pulled the maximum and absolute max voltages from the intel spec sheet that I linked to in the post. They're in Tables 7-4 and 7-6. I have heard the same thing about keeping vdimm within .5v of vtt. Which has made me wonder if a lot of problems I see people encounter with RAM isn't because their RAM is 1.65 and their vtt is more around 1.05.
m
0
l
January 22, 2010 11:23:19 AM

I ordered a Tuniq Tower 120 Extreme this morning, gonna give that a try and see how it compares to what I was seeing with the Hyper 212+. If it's better, the Hyper is definitely going back to Fry's next week. If it's not any better I'm gonna start suspecting problems with the temp sensor.

EDIT - rethinking now, should I just get the new Thermalright Venomous-X instead of the Tuniq?
m
0
l
January 22, 2010 6:25:56 PM

Quote:
EDIT - rethinking now, should I just get the new Thermalright Venomous-X instead of the Tuniq?
Recently I decided to upgrade my hsf. After gobs of reading I concluded that you can find any number of sources tell you any of the top products is best. It seems that the important thing is to buy out of the top class of products, which one from within that class is less important. Seems both your choices are top class, so I doubt its worth a great deal of time choosing which one. B
m
0
l
a b K Overclocking
January 22, 2010 6:56:43 PM

Tabs, have you checked your Voltage settings? I found that "at stock" mine were too high and have been able to lower them and remain stable. Also, Fisherman619 reported his temps lowered and he was able to overclock and remain stable once he set his voltages to be lower than what auto was setting it to.
m
0
l
January 22, 2010 7:43:41 PM

I agree with Bob. Either choice is a good one. Probably just consider size and weight. Please update us with your temps as i would like to compare them. I am considering returning my 212 for a Megahalems.

Bob5568 would you mind posting your temps? I would like to see the difference between our coolers.
m
0
l
January 23, 2010 3:37:38 AM

Hi Fisherman: Here are temps from 2 cores of both my 3.78 stable oc and my 3.6 oc By stable, I mean surviving occt test for 1 hr with no errors.

hot core @ 3.6:


cooler core @ 3.6


hotter core @ 3.78:


cooler core @ 3.78:

m
0
l
a b K Overclocking
January 24, 2010 12:07:49 AM

Fisherman619 said:
ekoostik can you please post your temps. I want to compare them to mine. TIA.

I checked my temps. Last week at Stock using Prime 95 with ambient temps of 18.8 C:
Idle: 18 C
1 Thread: 41 C
4 Threads: 47 C
8 threads: 51 C

Tonight with BCLK = 145, Dynamic Vcore at -0.01250 (yes, that's a negative) with ambient temps of 19.3 C:
Idle: 19 C
1 Thread: 43 C
4 Threads: 49 C
8 Threads: 53 C

Both sets of tests were done with Turbo, EIST, C1E, C3/C6/C7 Support all Enabled. Temps were measured with Real Temp 3.50 (I think Beta 4). Core Temp 0.99.5 confirmed temps.

Update: I ran prime blend test for about 4 hours last night. All four cores ran at 54-55 C. One core did hit 56, but settled in with the others at 54-55. This is with ambient temps of 19.3 C.
m
0
l
January 25, 2010 12:04:07 AM

My results with my Hyper 212+ and i5-750 have been decent. Daily OC of 1.28v @ 3.8Ghz the max temp reached is 67c using coretemp and running Prime95. OCCT returns pretty much the same results. If I OC to 4Ghz at 1.33v the max temp reached after 2 hours of stress testing jumps to 75c using coretemp. Pretty decent cooling for a $28 cooler. Adding a second CM BladeMaster dropped temps another 2-3 degrees, not much for the added noise.
m
0
l
January 25, 2010 5:29:58 AM

Thanks for the temps guys!! Looks like the $30 212+ is a Great cooler for 3.8ghz and below. At 3.8 i also get temps in 72c-75c so i stay @3.6ghz 67c-69c.
m
0
l
January 26, 2010 11:42:09 AM

Holy crap guys, I think I just figured this out here... I really hope I'm right but I'm gonna feel like a dumbass if this was it all along.

Is hyperthreading supposed to be on or off when OCing the 860? I just tried with it off, I have the CPU OCed 3.5Ghz @ 1.3 Vcore and I'm seeing nothing above around 64-67C now running Prime small FFTs.

Pleeease tell me this is it. I had hyperthreading on previously and that feature is obviously responsible for the huge temp difference I'm seeing here.

I ended up canceling the Tuniq 120 Extreme order and I decided to wait to do a bit more research on a replacement HSF, but if I'm right here then there's nothing wrong with the Hyper 212+ in the first place.
m
0
l
a b K Overclocking
January 26, 2010 12:55:12 PM

HT definitely causes the heat to got up. I haven't turned it off, but then I haven't pushed my OC nearly that high either. I have seen some OCers that recommend turning it off. Let's see what the others chime in, and you may want to snoop around some of the other 860 OCing threads in the forum.
m
0
l
a c 86 K Overclocking
January 26, 2010 1:26:36 PM

If your a gamer your better off with HT OFF and a high overclock. If you code DVDs or crunch for SETI etc it's best to have it on. Turning HT off in simple terms shuts down some of the cores.
m
0
l
January 26, 2010 3:46:58 PM

I leave hyperthreading on, as it makes a difference in my heavy graphics rendering work. I'm the one posting OCCT temperature plots above at 3.78ghz and 3.6ghz oc. So, no need to turn ht off to acheive decent temps, although it will affect how your overclocking project ends up.
m
0
l
January 26, 2010 6:00:35 PM

Yes HT off will result in lower temps and be easier to OC higher. Looks like most of our posted temps have been with HT on. Mine are. Like said above, if you run programs that can utilize HT then leave them on. If not turn them off and go for the 4 cores@ high clocks. 4ghz!! Me with HT on i get 175x21=3.67ghz 72-75c FL P95. HT off 190x20=3.8ghz 70-72c FL P95. close to same temps but higher clock! My 212+ cant do 4ghz =( 80-83c.
m
0
l
January 26, 2010 7:11:52 PM

Hmm... well with HT on I still can't even do a moderate OC (3.2 etc) without it going over 80C. So maybe something still is wrong...

I'm at 3.8 @ 1.3v Vcore now though, idling around 38 and getting just over 70 in Prime. System ran all night and seems stable. Guess next thing is to try turning down the voltage until it becomes unstable.
m
0
l
January 26, 2010 7:53:20 PM

On the 3.2 do you have turbo on?? If so thats why. Me with turbo on,HT on i can do 3.0ghz=BCLK 145 auto settings 67-69c.
Your on target with the 3.8 tho. I have the same results. @3.8 i do 1.3v 70-72c. HT off. You may be able to get better results so yes keep lowering to find out!
m
0
l
January 26, 2010 9:03:29 PM

I didn't change any of the other voltages besides vcore and vdimm (which is at 1.65v) - do I need to or no?
m
0
l
January 26, 2010 9:19:51 PM

I didnt have to. 9hrs P95 stable. If your not getting stability try upping your VTT a little. That may help.
m
0
l