Intel Core i7 920 massive heat issue

SmAcKtArD

Distinguished
Feb 10, 2009
9
0
18,510
Hey,


I have an i7 920 with antec 900 and akasa 967 heatsink. I also have extremely high temperatures with prime. It actually gets that hot the MOBO automatically compensates for the mass amount of heat and drops the 3.8GHz OC 920, to a x7.0 multiplier, (checked with CPU-Z) which finalises the core speed to 1350MHz and still runs at 91c after that! This happens instantly too, within seconds and the longer prime is ran for, the lower the multiplier becomes and the less load it can withstand.


On stock cpu speed, I also have a huge load temperature of 84c, the stock cpu cooler is way less than this, and this is one of the best heatsinks :/.


I have a couple of queries to the problem, however.


1. It is impossible to have the akasa-967 pointing towards the back in an antec 900 case WITH side fan, so instead, the fan is pointed upwards towards the 200mm fan above. The only problem I can see with this is the heatpipes won't be levelled up the same because its going across the chip sideways instead of up/down.


2. I used the thermal paste provided with the heatsink, spreaded onto the cpu with a plastic bag on my finger, so I am purchasing some artic silver 5 paste and re-sitting the cpu asap.


3. I haven't checked whether the cpu has any concavity yet, but I will and if it does, what is the solution to fixing it? (The heatsink 1366 pins are all 100% locked down).


4. I'm not sure with the position of the fan, I have tested the heatsink fan with some tissue paper and it seems to be acting as an intake fan, I'm not sure if this is a problem, as the big 200mm up top blows out and if the heatsink fan intakes, the big 200mm top fan won't be extracting the air from the heatsink fan directly. Although the fan HAS been put onto the heatsink correctly (with the fan sticker facing forwards) so I am not sure whether it is designed for intake?


Any help would be appreciated as I am extremely concerned that either I have a faulty CPU, or I have done something really wrong.
 

yomamafor1

Distinguished
Jun 17, 2007
2,462
1
19,790
Like Kari said, it might have more to do with the way the heatsink was installed. The 920 (2.6Ghz) I have with stock cooler never gets above 60C, even under SMP folding.
 

SmAcKtArD

Distinguished
Feb 10, 2009
9
0
18,510
Yea, I have it so the heatsink fan is blowing into the aluminium fins and blowing air out that way, but I was curious as to if it matters to which way the air is blown, up or down? Atm the air is being pushed down and not pushed up towards the top 200mm case fan.
 

someguy7

Distinguished
Dec 12, 2007
1,186
0
19,310
Why in the world would you even overclock a cpu at all if you are having temp problems. Let alone to 3.8ghz from 2.66ghz.

Also not that this is the cause but you dont need to and shouldnt move the past around with your finger or anything else for that matter. You do not "paint" the cpu or cooler with paste. If you follow the directions when you apply the cooler the pressure will distribute the paste.

You asnwered your own question. You said the stock cooler performs better. You def did things wrong. Not something. There are def more than one blunder on your part.
 

SmAcKtArD

Distinguished
Feb 10, 2009
9
0
18,510
I see... well tbh, I am still at a lose with how to apply the thermal paste...

There seems to be a fence where on one side people are applying the paste directly onto the cpu and self-spreading, people are applying the paste in a line diagonally across the cpu to interact with all 4 cores OR people are applying thin lines directly onto the heatpipes and pressuring it with the heatpipe itself.

Instruction that came with the Akasa-967 heatsink actually say to spread the thermal compound thinly across the board with a latex glove... So I haven't a clue what is best...

And I overclocked to test temperatures on load when overclocked to see if it would remain static, to answer your question.
 
If the stock heatsink is outperforming the aftermarket, there are 2 possibilities:

1) The aftermarket heatsink sucks
2) You have it installed wrong.

Also, I will repeat what has been posted above: if it is already running hot at stock, WHY ON EARTH would you overclock it?
 

SmAcKtArD

Distinguished
Feb 10, 2009
9
0
18,510
Lol, I hadn't done a proper temp test on stock before cranking it up, it was only after doing a proper stress test on OC that I realised how hot it was going, and I did only run the stress test twice at this speed, I had to make sure my eyes wern't deceiving me. N00b error I know...
 

sideshow

Distinguished
Jan 1, 2008
2
0
18,510
I have the 920 at 3.6ghz (180*20) with the Xigmatek S-1284, similaar heatpipe structure to the Akasa. I have found best results by applying a very thin line directly to the heatpipes and then seating the heatsink onto the cpu. A TIM like OCZ Freeze that will spread freely works better than a heavy one like AS5, IMO. With 1.3V going through it my 920 idles at 33-35c and never gets above 63c, even after a 3 hour prime test, in an older Lian-Li PC-70 tower.
 

mamw93

Distinguished
Jan 16, 2009
540
0
18,980
load up on fans. Cram asmany as you can into the case cause yer problem sounds like poor airflow. If that doesn't work buy a new heatsink... preferably Zalman.
 

Siggy19

Distinguished
Mar 5, 2009
144
1
18,690


Heat rises... if you are blowing the hot air DOWNWARDS, guess what is going to happen to it.

The ideal airflow for a computer case is from the bottom up. That way, the hot air works with the fans rather than against them.

As well as switching the fan on the cooler, you may want to check the flow of any other fans to try to ensure that they are working together.

Oh, and only use a tiny BB sized blob of paste. It really doesn't matter how it gets spread, pretty well any method will work as long as you do not use too much... the paste is not as good as metal to metal contact. It is intended only to fill in the microscopic imperfections in the metal surface... NOT to seperate the metal itself.
 

Lozza81

Distinguished
Mar 13, 2009
19
0
18,510
Another thing I'll add for you to check is I have read on other forums some i7 920 overclockers with aftermarket coolers forgot to peel off the clear plastic seal off of their heatsink before applying their thermal paste.

They were then getting crazy 90C temps through a simple overclock and as they had applied their thermal paste so quickly they could not now see the clear plastic protective stip which manufactures put on the their heatsinks contact plates with for protection. I know its a noobie thing to do, but we can all make mistakes and miss things sometimes.

When you unseat your heatsink and CPU double check you definately have taken off that thin plastic sticker. Also check to see how much contact has been made on thermal paste to doublely make sure the heatsink is definately contacting the CPU correctly.

My 920 (non-overclocked) idles at 30-32C and never gets above 40C with my Thermalright 120 with just 1 x 120mm blowing air in across the fins. What were you temps before you ramped her up to 3.8ghz? Also I suggest you put the stock cooler on also and write down some idle and load temps. You need to start basically ticking off different possabilities until eventually you will solve it.
 

celestus

Distinguished
Mar 20, 2009
6
0
18,510
My core i7 920 at stock speeds reached 95c and underclocked using arctic silver after removing the original paste on my stock cooler - Never used original paste. A thin layer was applied to the entire chip using a credit card ($10 ncix plastic certificate ^_^). This was in a coolermaster cosmos S case - completely perforated with a massive 200mm sidefan blowing on the board in a cold room.

Switched to a water system with 2 ~180mm oil radiators, eihem 1250 l/hr pump, 1/2 in tubing to cool the cpu and 2 4850 cores. The core i7 overclocked at 4.2ghz 1.3675v with 4850x2 stock in the same cooling loop can still reaches 60-70c running furmark and prime95. The radiators are under the top of the case with fans blowing outward and the radiators are putting off a ridiculous amount of heat.

I don't know if I have problems, a bad chip, or i'm doing something wrong but there is no way stock cooling should have been reaching the thermal max. Used realtemp to monitor temperatures, motherboard fan monitor showed all fans working properly.
 

Oahn Macleod

Distinguished
Mar 27, 2009
1
0
18,510
Guys,

I've been building PC's since my first custom setup back in 1990... adding a 1mb Trident card to my 386sx/20 and using Stacker to compress my 110mb harddrive. This with 4 mb of ram at 70ns onboard.

Fast forward to the present and my current rig is a Gigabyte EX58-UD5, Intel I7 920, 6GB G.Skill PC10666 and BFG GTX295 (the rest is irrelevant for this post).

I am currently running the CPU at 200 x 20 in Turbo mode and using a Cooler Master V8 which runs it at 4.2Ghz in Vista 64 Ultimate. The voltage is at a mere 1.4v and the temp is a cool 52 Celcius under full load (Crysis, Far Cry 2, Left 4 Dead, Fallout 3) never dumping or had a problem. The ram is running at 9-9-9-21 @ 1.5v.

What I can't understand is... How is it some of you can't get your rigs past 3.6 or more? The methods some are talking with water cooling and such... Am I just lucky or something here? Just seems like people are trying to get an extra point or two out of some benchmark and never getting a chance to use their new rig?

And when you add all of the time you spend in trial and error and benchmarking and divide it by nanoseconds and frame rates and add that to your CPU/GPU/RAM benchmarks... how fast is your system then?

From a guy who is getting older but still pwn'ing you on the virtual battlefields!

Oahn Macleod