Why is my i7 950 so hot? Should I lap it?

Hey guys.

About 3 months ago I finished my first PC build and last night I overclocked it for the first time. I've got my i7 950 running at a reasonable 3.72GHz, but it seems quite hot. It idles at around 52C (running Windows 7 64Bit) but when I run OCCT the temperature stabilizes around 83-85C.

Before my first attempt at OCing I remounted the CPU cooler and got great temp decreases (almost 20C) as a result, so I'm pretty sure my cooler is mounted properly. All fans are maxed. But I did notice that it looks (from the thermal paste) like there is a slight depression in the middle of the CPU IHS. I'm wondering, should I try lapping my CPU and heatsink, or are these temperatures normal?

Below is an outline of my system and an extract from the BIOS setup.

System:

CPU: Intel i7 950

CPU Cooler: Zalman CNPS9700 LED

Motherboard: Gigabyte GA-X58A-UD7

RAM: Corsair Dominator TR3X6G1866C9DF

GPU: 2 x Sapphire Radeon HD5850 in CrossFireX

PSU: XFX Black Edition 850W

Installs Disk: WD Velociraptor 150GB

Data Disk: WD Caviar Black 1TB

Case: Antec Twelve Hundred

Display: 2 x 21" Samsung SyncMaster P2250


BIOS:

CPU Clock Ratio: 24x

CPU Frequency: 3.72GHz (155x24)

BCLK Frequency: 155MHz

System Memory Multiplier: 12.0

Memory Frequency: 1860MHz

CPU vCore: 1.21875V

QPI/Vtt Voltage:1.440V

IOH Core:1.340V

DRAM Voltage: 1.500V


Everything else is pretty stock standard.

Thanks

that cooler is not the best to be quite honest

you should think about investing into a prolimatech mega, noctua d14, true venemous-x as they are some of the best air cooling solutions around

lapping is risky, but can yield a 5+ degree drop..for such an expensive cpu.. i wouldnt do that unless you know what you are doing

also make sure the case airflow is adequate.. make sure you tucked away every wire that wasn't necessary to be out in the open as it restricts flow

and to the forums!

Hi

Thanks!

Yeah I don't think I will lap the CPU any time soon. I might try lapping the heat sink though - not much to be lost there. And I will be careful. I already got a 4C temp drop by re-cabling the whole rig and cleaning the intake filters.

Thanks again.

Lapping is bad! Everyone I know who has done it has either had the same temps after or actually higher temps.
The only way it would be better would be if the heatsink was actually touching the chip. Since it doesn't the temps are likely to rise.

The max temp for the 950 I think is 85 degrees so it's not that bad. Idle temps are about 45-55 with similar coolers.
While it's running does the heatsink feel hot? If not when push the heatsink down (not to hard but enough so it makes a good strong contact) and see if the temps drop. If they do then the heatsink isn't mounted properly of you haven't applied the TIM correctly.

Don't worry, I'm not going to lap my CPU.

The heat sink is mounted pretty well, hey. I spread the TIM (Arctic Silver) all over the CPU, then scraped it down with a credit card, then put another thin line down the center and screwed the heat sink down good an hard. The heat sink feels pretty warm while running and the air being pushed out the back of the case is considerably warmer right behind the CPU cooler.

The max temp for the 950 is 100C. OCCT doesn't get it above 85C so I'm not worried. I just hoped for higher clocks. But I think Psionic is right - I just need a better cooler and possibly an additional intake fan. Two HD5850s and two very high speed HDDs mean that temps inside the case are pretty high and airflow is tricky.

tinymouse2 said:

Lapping is bad!

Click to expand...

Sorry man, but every chip I've lapped has had better temps than before.

ReneeJade said:

Don't worry, I'm not going to lap my CPU.

The heat sink is mounted pretty well, hey. I spread the TIM (Arctic Silver) all over the CPU, then scraped it down with a credit card, then put another thin line down the center and screwed the heat sink down good an hard. The heat sink feels pretty warm while running and the air being pushed out the back of the case is considerably warmer right behind the CPU cooler.

The max temp for the 950 is 100C. OCCT doesn't get it above 85C so I'm not worried. I just hoped for higher clocks. But I think Psionic is right - I just need a better cooler and possibly an additional intake fan. Two HD5850s and two very high speed HDDs mean that temps inside the case are pretty high and airflow is tricky.

Click to expand...

Too much TIM if you did that, and too much TIM acts like an insulator. Clean it all up, and just put a small blob in the center. If you were to pick up the HS and look and the squish pattern it should extend out to almost cover the entire IHS, but not quite, and none should go over the edges of the IHS. But once you squish it down, don't pick it back up unless you plan on cleaning it up, and starting over. You can introduced air pockets if you re-use.

Although lapping can produce better results, it can be tricky, and if not done correctly can make things worse. Try to add more pressure before going down that road b/c even if you do lap you'd want to add more pressure to make up for the extra gap produced by removing material. Adding extra pressure to my mount and switching to MX-2 dropped my temps about 8*C.

ReneeJade said:

Don't worry, I'm not going to lap my CPU.

The heat sink is mounted pretty well, hey. I spread the TIM (Arctic Silver) all over the CPU, then scraped it down with a credit card, then put another thin line down the center and screwed the heat sink down good an hard. The heat sink feels pretty warm while running and the air being pushed out the back of the case is considerably warmer right behind the CPU cooler.

The max temp for the 950 is 100C. OCCT doesn't get it above 85C so I'm not worried. I just hoped for higher clocks. But I think Psionic is right - I just need a better cooler and possibly an additional intake fan. Two HD5850s and two very high speed HDDs mean that temps inside the case are pretty high and airflow is tricky.

Click to expand...

Yeh WAAAY to much TIM.
You want to put a blob smaller than a grain of rice in the middle and just push the heatsink down, it'll spread it for you there. Less is more

Re-apply it with this knowledge and your temps should fall.


EDIT:

ghost_recon88 said:

Sorry man, but every chip I've lapped has had better temps than before.

Click to expand...

Well there's no reason for lapping to work though...
I remember reading how thermal transfer works best when the surface is slightly rough so the TIM can actually make a strong and good contact with the chip.
Reducing this should mean higher temps.

I'm just going by what an article said and what friends have told me so my sources might not be trustworthy

tinymouse2 said:

Well there's no reason for lapping to work though...
I remember reading how thermal transfer works best when the surface is slightly rough so the TIM can actually make a strong and good contact with the chip.
Reducing this should mean higher temps.

I'm just going by what an article said and what friends have told me so my sources might not be trustworthy

Click to expand...

You don't lap a chip, just to make it nice and shiny... lapping is used to flatten the surface. Most IHS are convex/concave, the extent to which this is the case, varies from chip to chip and socket to socket. Lapping is used to produce two perfectly flat mating surfaces. Whether these are lapped to a shine, rough, or just slightly dull; and the subsequent effectiveness, is debatable, but the core principle of lapping (to flatten the surface) is not debatable. You cannot get rid of an IHS "pocket" with tension, the only thing that you can do is, fill it with TIM or lap it smooth. Not all IHS's need to be lapped, I've seen some that are quite flat, but I've also seen some that are COMPLETELY unacceptable (as in contact in one area, but required a lake of TIM in another)

nzaneb said:

You don't lap a chip, just to make it nice and shiny... lapping is used to flatten the surface. Most IHS are convex/concave, the extent to which this is the case, varies from chip to chip and socket to socket. Lapping is used to produce two perfectly flat mating surfaces. Whether these are lapped to a shine, rough, or just slightly dull; and the subsequent effectiveness, is debatable, but the core principle of lapping (to flatten the surface) is not debatable. You cannot get rid of an IHS "pocket" with tension, the only thing that you can do is, fill it with TIM or lap it smooth. Not all IHS's need to be lapped, I've seen some that are quite flat, but I've also seen some that are COMPLETELY unacceptable (as in contact in one area, but required a lake of TIM in another)

Click to expand...

Like I said, second hand info. I've already said all I know so I can't comment passed here

nzaneb said:

You don't lap a chip, just to make it nice and shiny... lapping is used to flatten the surface. Most IHS are convex/concave, the extent to which this is the case, varies from chip to chip and socket to socket. Lapping is used to produce two perfectly flat mating surfaces. Whether these are lapped to a shine, rough, or just slightly dull; and the subsequent effectiveness, is debatable, but the core principle of lapping (to flatten the surface) is not debatable. You cannot get rid of an IHS "pocket" with tension, the only thing that you can do is, fill it with TIM or lap it smooth. Not all IHS's need to be lapped, I've seen some that are quite flat, but I've also seen some that are COMPLETELY unacceptable (as in contact in one area, but required a lake of TIM in another)

Click to expand...

That is a good clarification.

To cover the only part not mentioned - direct contact between the heatsink and IHS interface is preferable, TIM is only there to fill the gaps which otherwise would be tiny insulating air pockets. This is the motivation for lapping, flatter surfaces mean increased direct contact between heatsink and IHS.

One could still debate if the lapping is effective, few people have the tools with great enough accuracy to read ambient, let alone surface "flatness" which is difficult to observe with high accuracy. Most people don't have temperature probes accurate enough to take good ambient readings to know if their internal CPU diode readings are actually improved, or if room temperature has fluctuated (don't trust any cooler review which tells you ambient was consistent - it WASN'T, the reviewer just didn't have the tools to observe the fluctuations accurately).

tinymouse2 said:

Yeh WAAAY to much TIM.
You want to put a blob smaller than a grain of rice in the middle and just push the heatsink down, it'll spread it for you there. Less is more

Re-apply it with this knowledge and your temps should fall.

Click to expand...

Yeah I know that *usually* that would be way too much. But something is not flat. Either the IHS or the heatsink. Because this was the third time I've remounted it and the first couple times using a tiny amount just did not work - it didn't spread. So I have to use the goo to compensate for the wonkiness.

Also, I'd never remount without cleaning first.

Also, I get the concept of lapping (am a physics student). It makes sense if the chip is not flat, provided there is no fancy in-built cooling technology in the IHS.

I.M.O.G. said:

That is a good clarification.

To cover the only part not mentioned - direct contact between the heatsink and IHS interface is preferable, TIM is only there to fill the gaps which otherwise would be tiny insulating air pockets.

Click to expand...

TIM is there to fill microscopic air pockets, since on the molecular level, metals are quite lumpy. The TIM should not have to fill up actual noticeable gaps, but sadly, in my case, it does.

ReneeJade said:

TIM is there to fill microscopic air pockets

Click to expand...

That could be debated but its more or less accurate. However, the flaws between the mating surfaces of the IHS and Heatsink base are often visible with the naked eye and that's how TIM's are typically used - on off the shelf components. Then there's lapping... lapping hasn't been decisively shown to offer improvement with a scientific level of verifiability. Ya, it should help, but it hasn't been shown that it makes a reliable, observable, reproduce-able improvement to CPU temperatures.

So ya, it'd be ideal if the TIM just fills microscopic air pockets, but in reality its often doing more work than that with the components we use, and improving the flatness of the mating surfaces by lapping hasn't been shown to be a decisive victory/failure.

IMO, you really need to upgrade your cooling on that i7 to a better heatsink. I am on the second day of playing with my first i7 system (980X with a busted P6X58D mobo I will have to RMA ) and I'm finding this 980X not hard at all to cool on high end air. So far all the overclocking I have done with it is by multi only but I have it at the following speeds and the temps you see in Real Temp are from running a sensor test, which runs Prime95 for a little while. I did an hour run or so last night at this same clock speed with P95 and temps didn't get too much higher than this. I presently have a Venomous X installed and am running a Yate Loon High in push and a Sanyo Denki 1011 in pull (long story, due to troubleshooting this broken *** Asus board that has 4 dead mem slots). So in my rather limited experience with i7 and my processor, temps aren't as bad to deal with as I thought they might be.

^^

Thats just gross

muddocktor said:

IMO, you really need to upgrade your cooling on that i7 to a better heatsink. I am on the second day of playing with my first i7 system (980X with a busted P6X58D mobo I will have to RMA ) and I'm finding this 980X not hard at all to cool on high end air. So far all the overclocking I have done with it is by multi only but I have it at the following speeds and the temps you see in Real Temp are from running a sensor test, which runs Prime95 for a little while. I did an hour run or so last night at this same clock speed with P95 and temps didn't get too much higher than this. I presently have a Venomous X installed and am running a Yate Loon High in push and a Sanyo Denki 1011 in pull (long story, due to troubleshooting this broken *** Asus board that has 4 dead mem slots). So in my rather limited experience with i7 and my processor, temps aren't as bad to deal with as I thought they might be.

Click to expand...

Nice work! And thanks for the advice - I'll upgrade my heat sink next time I've got a bit of spare cash (god knows when that'll be).

Bummer about the main board mate. I had that too. I spent months researching my build (coz I knew nothing about hardware about 6 months ago), then spent my summer working my arse off at McDonald's to pay for it. I finally got all my parts and I was so excited to build it, only to discover (after much painful and suffering) that my top-of-the-wozzer motherboard had a totally dead PCIe setup. Very frustrating. Good luck with the RMA.

Besides a better cooling solution the 950 just runs hot, from all the 900 series i7's I've had the 950 was the hottest running CPU.

ReneeJade said:

...So I have to use the goo to compensate for the wonkiness...

Click to expand...

lol, too technical for me.