i7 860 all stock 92C full load.

[slugman]

Hi all, 100% new here.

Just built a core i7 860 for a hackintosh. GA-P55A-UD4P board, running great except - I get what would appear to be HIGH temps during prime95 testing.

Depending on the outside temp, 80-92C under full load after a few minutes.

Idle temps are good; 30-40C depending on ambient temp. Sometimes below 30C at night.

I have no overclock, save 1600 Mhz RAM (temp problem occurs even with RAM clocked to 1333). Turbo enabled.

All testing done under Windows 7.

Stock cooler, but since I don't care about overclocking, that should be enough, yes?

I have checked 15 times that it's seated properly. All the pins are pushed in. Thermal paste seems fine. At high speed the stock fan spins at ~2250 rpm. Is that normal? Should it be going faster?

Temps measured with both coretemp and speedfan, they agree.

I have an Antec 300 case with a the rear 120mm and top 140mm fan, as well as a coolmaster 120mm fan in the front. Setting the case fans to high makes almost no difference. Same sort of temps with the case open.

I see people here worrying over 70C. Am I damaging my CPU? At 92C the CPU does not downclock, no thermal monitor seems to be activated. If I hadn't been reading the forums I'd assume 92C was normal.

Disabling hyperthreading and turbo mode gives me mid 70s max. But those are features I paid for and should be able to use? I plan to do a lot of video encoding and numerical simulations and will use the CPU at full load a lot.

I should add that as soon as I disable prime95, temps drop to 70C within a second, and creep down normally from there. Regular activity (3D games, even handbrake encoding) doesn't see temps climb above 75. Not that I'm going to be searching for mersenne primes on a daily basis, but... yeah. It's got me worried.

Any help appreciated!

Thanks so much.

 

[Markfw]

Well, that seems a little high for stock, but personally, even at stock I would get an aftermarket cooler, as they run cooler, and quieter.

 

[RussianSensation]

Stock cooler, but since I don't care about overclocking, that should be enough, yes?

Welcome to AnandTech!

No. Bottom line, don't waste any more time and get an aftermarket cooler. The stock cooler on the i7 860 is NOT sufficient (unless you are comfortable running 90+*C). My operating temperatures were in the 90-95*C at Prime95 and the fan was loud to boot. I upgraded to Prolimatech Megahalems. If you are not going to be overclocking, I suggest cheaper alternatives such as:

Cooler Master 212 Plus <$30
http://www.amazon.com/dp/B002G1YPH0/...SIN=B002G1YPH0

If you want to overclock, Scythe Mugen 2 is a good option <$40
http://www.heatsinkfactory.com/scythe-mugen-2-rev--b-cpu-cooler---scmg-2100.html

I am not sure what Ram you have, try lowering your DDR voltage to 1.55V or even to 1.50. Running ram at 1.65V or above significantly increases CPU temperatures of i7 processors.

 

[slugman]

Thanks for the tip about the RAM. Unfortunately, the spec calls for 1.65-1.84V (A-Data 1600Mhz). I could exchange it, as the system is still new. I'm not uncomfortable with 90-95C; silicon melts much higher. I am just wondering if this is designed behavior or something is wrong.

The 212 looks like a good choice, people report reasonable overclocks with that thing. Hey, if my temps were low enough I'd overclock a little.

But is the consensus that 95C is abnormal and something is wrong, or is this just stock behavior with the stock cooler?

 

[alyarb]

92 at full load is acceptable for stock. usually it's in the mid 80's but as long as you are below tjmax, the system will continue to run. you can get that down drastically with a 212, mugen 2, etc.

 

[slugman]

92 at full load is acceptable for stock. usually it's in the mid 80's but as long as you are below tjmax, the system will continue to run. you can get that down drastically with a 212, mugen 2, etc.

Thanks! Figured I'd report back in case other people have the same question - lowered RAM voltage and system seems to work fine but virtually NO temp difference.

If I spring for the 212, I'd probably try overclocking. What stock load temps should I shoot for before trying, say, a base freq of 160 (which should put my RAM at 1600 and non turbo CPU at 3.36Ghz, yes?)

Thanks again, this forum is great.

 

[SHAQ]

You should have load temps below 60C with the 212 at stock. A 20-30&#37; OC won't increase the temps very much. It's only when you go for a 50% OC that load temps will get over 70C.

 

[VirtualLarry]

Wow, the Intel stock cooler is really that bad for the i7s? Even the short mini-heatsinks that ship with the Core2Quads are mostly adequate for stock speeds. What is Intel thinking?

 

[n7]

wow, the intel stock cooler is really that bad for the i7s? Even the short mini-heatsinks that ship with the core2quads are mostly adequate for stock speeds. What is intel thinking?

$$$ i presume...

D:

That's one thing AMD usually does fine; their stock coolers are very decent.

 

[Ben90]

TBH if your going to run stock, I wouldn't worry at all about a 92C load or getting an aftermarket cooler. There is a lot of hysteria on this forum about temps, some warranted, some not;but for stock voltage I see no reason to worry about it.

Anything below TJmax is fine if you don't mind your CPU's expected lifetime to be "on the other side" of the 3yr warranty period. (meaning 3.1yrs or some such)

The QA engineering that goes into the process development itself is all geared towards ensuring this much just from a minimization of extenuating liabilities standpoint to Intel's books.

There is no single temperature threshold above which your cpu's lifetime suddenly diminishes. It is a continuous function that is dependent on an exponential of the operating temperature thanks to the physics that underlie the Arrhenius equation.

As a rule of thumb, for every 10C higher your operating temps the expected lifetime of your CPU is reduced by 50&#37; (think of half-life). Whatever your expected cpu lifetime is if operating at 50C, call it X years, you can expect that lifetime to be cut in half if you operate your cpu at 60C, so X/2 years, and again cut in half once more if you operate it at 70C, so X/4 years, etc.

That may sound dire but understand the lifetime is engineered into the IC from the "top-down" in terms of the thermal specs. Meaning your thermal spec was set for your chip with the desire to minimize the number of in-field fails that would occur under warranty.

So making the assumption that your CPU has an expected lifespan equal to (really we should assume greater than as Intel would not be silly enough to make the mean of the distribution equal to their warranty period and then have to deal with the entire left-hand side of the distribution failing under warranty) the standard warranty period (3yrs) when operating at TJmax is a reasonable assumption. Then for every 10C below TJmax you operate the chip you should double the expected lifespan.

If TJmax is 90C and you operate at 80C then a very reasonable lower-estimate of your CPU's expected lifespan would be 6yrs (2 x 3yrs warranty period). If you operate at 70C then 2x2x3yrs = 12yrs expected lifespan.

What is the basis for my arguing this? At TI we required our process technology to be developed so as to enable the minimum lifetime requirement of 10yrs operating at max spec'ed operating voltage and max spec'ed operating temps in continuous 24/7 operation. It is SOP for the industry.

Now where you can really cook your goose (cpu) is over-volting and running hot. It doesn't take much to be operating your CPU in a voltage/temperature regime that in combination the two factors contribute to lowering the expected lifespan of your CPU to something <1yr.

Not too mention there is always a distribution to the lifespan and your particular chip could have some intrinsic weakness/flaw in it that puts its expected lifespan at a value below the mean of the distribution and by operating at elevated temps and volts it is destined to fail substantially sooner than the warranty period. (I killed my QX6700 in something like 18 months, never operated above TJmax or above Vccmax, but had lapped the IHS so no warranty replacement for me)

 

[slugman]

I dropped a whopping 20 bucks and bought the 212 at microcenter. Stock max temp at load is now 62-63C. I live in San Jose with no AC so my temps seem consistently about 5-10C higher than other people are reporting. Might be cause it gets hot in my room.

Anyway WOW. 20$ cooler and a 30 deg temp drop? Thermal paste was even included.

Probably going to try overclocking.

Any thoughts on upping the base freq to 160 with stock voltage?

1 response already said no big deal...

Thanks!

 

[alyarb]

you should have absolutely no problems at 160. 200+ is when you begin to put your system to the test.

 

[slugman]

Thanks, gonna try it right now.

Anyway, also of note is that idle temps, while lower, jump around.

they seem to go from 27-34C almost randomly and instantly. is this normal? Reason to worry?

Wheeeee feel so much better about leaving this running big jobs now.

 

[slugman]

EEK BSOD city unless I disable Turbo boost and set the memory "Performance Enhance" option (whatever THAT does on my gigabyte board) from turbo to standard.

Maybe I just have some bad silicon?

EDIT: It's just the turbo boost. With it on the cores can pass 4 Ghz at 160 base clock (26x160=4160) perhaps needing more voltage? Is this normal behavior? Should I have to turn off tboost at stock voltage?

EDIT2: Even with the new cooler and Tboost off, temps QUICKLY approach 90 at 3.36 Ghz. Oh well.

EDIT 3: should have made it clearer. Temps quickly approach 90 running prime95. Idle temps are 30-40

 

[alyarb]

what's the voltage at 3.36? 90 is too high. something is wrong either with your case airflow or the boundary between the base of the heatsink and heatspreader on the CPU.

 

[slugman]

Voltage is set to auto. It could be that I applied the thermal paste badly, or the thermal paste that came with it is bad.

But if either were the case, why would it work so well when not overclocked? Temps at load are right where everyone says they should be ~60.

Should note: that quickly approaching 90 is under prime95. Idle temps at 3.36 are 30-40.

 

[RussianSensation]

Voltage is set to auto.

Don't do that. I am able to get 3.9 ghz from 1.312 (so BIOS voltage is set to 1.2250 Normal + 0.10625 Dynamic Voltage option).

With your board, there should be an option to enable Dynamic Voltage. If you enable Dynamic voltage on the board, then you can run the CPU with SpeedStep at idle and it will lower the voltage for you (i.e., 0.96V in my case at idle). When the processor is loaded, it will compensate and increase the voltage instantly. If you don't enable Dynamic Voltage compensation, then you may find yourself running 1.7ghz in idle at 1.33V!!!

- Make sure to lock PCIexpress frequency to 100,
- lower CPU:Ram ratio so that your ram speed does not exceed DDR3-1600 (System Memory Multiplier)
- Enable Load Line Calibration (http://www.overclockers.com/wp-content/uploads/images/stories/articles/GigabyteGAP55/IMG_7219.jpg).
- Set lowest QPI Clock Ratio of 32x (http://images.bit-tech.net/content_images/2009/11/gigabyte-ga-p55-ud5-motherboard-review/bios3.jpg)
- Enable DVID (set CPU VCore to "Normal", then start dialing in Dynamic VCore control: http://www.xbitlabs.com/images/mainboards/ga-p55-ud3r/bios_09.png).

Keep in mind, if you are using a multiplier of 26x, you are running Turbo Mode. It would be much better to run 4 cores at 3.9ghz for example. So Disable turbo mode. Another thing to keep in mind, i7 processors overclock better with odd multipliers. Select multiplier of 21x and CPU voltage of 1.33-34V in the BIOS, Base Clock of 186. You should be at 3.9ghz.

Download EasyTune6 too to monitor your Vtt/QPI voltage. Generally, it is better to run it at 1.21V (although I am using 1.25V). Gigabyte boards can often set QPI voltage to 1.39/1.41V on Auto which is way too high for Lynnfield processors. You want to be especially careful with this voltage setting.

Download CPU-Z to monitor your DDR3 speed (that lower ratios are working properly) and your CPU voltage. Alternatively, HW Monitor (on the same website works too):
http://www.cpuid.com/

 

[alyarb]

oh ok, your motherboard's "auto" setting is causing it to basically set high volts when under load. for the sake of stability the voltage winds up being unnecessarily high.

On average, a system like yours should be able to do 3.7 GHz at 1.25v, and I'm being a bit generous there for the sake of minimizing the duty on the Hyper 212+. If you want to disable turbo and try that, feel free and see if it POSTs at 23x160 with 1.25 or 1.26v.

edit: better post above. definitely verify your vcore with CPU-Z. Feel free to overclock as high as you possibly can without compromising stability, but personally I wouldn't go past 1.35 in your case.

 

[Pneumothorax]

Agree with posters above as I have the same mobo. I hate how the manufacturers have turned "auto" into an over volt option. Auto should mean stock voltage!

 

[slugman]

Amazing. Manually set voltage to 1.185, load temps at 3.36 now seem to settle in at 61-66C (that's the spread between the cores- is it too big? I'm worried it comes from uneven thermal paste or something). Maybe this was my problem in the first place.


But "Dynamic Vcore" is grayed out, and I notice in CPU z that even at a 9x multiplier my CPU voltage does not drop. x.x. Any suggestions?

You guys are amazing, thank you so much. I hope someone else reads this and saves some time.

EDIT: During nonload situations (web surfing etc) I am noticing BIG core temp spreads between what I guess are active and inactive cores- just saw core 0 at 43, core 2 at 28C. Can a thermal gradient like that be bad?

It evens out to about a 5C spread after all cores become active, and when idle they're all the same temp. Seeing 64-64-59-63 in prime95 now.

Also: Turbo now works but seems to ignore my voltage settings. With turbo on, I'm seeing 1.36V in cpuz at full load even with 1.185 and +.106 specified in the bios, causing the same heat issue. And setting the multiplier to 22 automatically enables all turbo multipliers. Any way to enable 22x but not 25-26x multipliers?

 

[lopri]

That's one thing AMD usually does fine; their stock coolers are very decent.

Even AMD has been cheapening on HSF. Back in the days, dual-core Opterons came with this:



Now BE CPUs come with this:



In my limited testing the older one performed better and quieter.

 

[RussianSensation]

On Gigabyte boards, there is no way to run 22x multiplier without enabling Turbo as far as I am aware. However, even up to 180 Base Clock, you should not have to raise the QPI voltage to more than 1.25V. Did you put enough thermal paste/spread it evenly?

 

[slugman]

Pretty sure I did a good job with the thermal paste. If I hadn't, it probably wouldn't be right where it should temp-wise at stock settings.

I think I just got some less hardy silicon. Microcenter here (where I got mine) is selling the i7 860 CPUs new for $199. Maybe they have some deal where the worse of the batches go to them in quantity.

Or it could just be a fluke. Or the thermal paste could really be bad.

I'm not obsessing about this anymore. My wife is going to kill me.

It runs cool at stock and that's what I got it to do.

Thanks for all the help, everybody. Really appreciate it.