Sign in with
Sign up | Sign in
Your question
Solved

High temps with i7 920 not OC

Last response: in Overclocking
Share
December 19, 2009 6:58:38 AM

My Build:

i7 920 (2.66 ghz stock)
asus p6t
6 gb ocz ddr3 1600
evga gtx 260
2 WD 640 Black Edition
samsung 22x sata burner
antec 500 watt psu
cooler master HAF 922 case
cooler master V8 cpu cooler

So I put my build together and got windows 7 running. I haven't tried to OC yet but I was reading the forums etc.. I decided to get some data first to make sure my build was doing good. I got real temp 3.0 and my idle temp was 37-40c which is to be expected from what i read in the forums. When I downloaded prime95 64 my temps went to mid 90C, i was totally shocked, people with my simlar build should expect 50s or 60s. Did I do something wrong? I used the thermal paste that came with the V8.

any suggestions would be most appreciated.

More about : high temps 920

Best solution

December 19, 2009 10:34:52 AM

The heatsink is probably not getting good contact with the CPU, I'd suggest taking it off and putting it back on to see if that helps
Share
December 20, 2009 6:50:21 AM

I decided the problem could be either the thermal paste of the V8 installation. I got new thermal paste (Artic MX-3) and some high quality isopropyl alcohol for cleaning. Reinstalled the CPU cooler after cleaning and adding new paste.

The results were dramatic, the system idles now at 33-34C (39-40C before) and average max temp is like 66C now (mid 90C before) after an hour of prime95, which is acceptable. I was hoping for mid 50's C but this is okay for now.

What temps do people get with the V8 and i7 920 at stock?


Screenshot

m
0
l
Related resources
a c 202 à CPUs
a c 148 K Overclocking
December 20, 2009 1:36:18 PM

I'm overclocked t 3.7 Ghz under 100% load on the everyday OC profile and getting temps in mid 50's w/ Megahalems. You could try lowering CPU voltage for better temps. I'm at 1.125 at the 3.7
m
0
l
December 21, 2009 8:26:34 AM

I'm not sure if the thermal paste needed some time to break in but I tried to OC now and I was reasonable for me.

Specs:

3.6 ghz
180 BCLK
1.66 Ram Voltage
1.88 CPU PLL
1.20 CPU
1.20 QPI

Max temp is around 68-69 after 1 hour of prime 95, avg temp is around 66. With my normal usage for the past few day the highest temp has been like 48 maybe, I guess i don't really need all that power. I'm still working on tweaking the settings, in the future I hope to sell my V8 and get a better CPU cooler, also my PSU is 500 watts I may upgrade to around 750 and get a 2nd vid card. For now though I will focus on keeping this build safe. I should add i OC'ed my gtx 260 SC as well.
m
0
l
December 21, 2009 9:42:08 AM

I have Intel i5, and I had many problems heat too... I'm using the intel stock cooler...(for now)
The temps that I got were 51C-52C ideal and 80C and up (Till I got sudden shutdown), I read your post and decided to replace the paste, so I did it... I cleaned it and tested two pastes... one gave BAD results and the other gave "ok" results... and now I'm standing at 47c on ideal and 80c and up on load...

This was all done with the case open, closing the case caused a raise of 10c...

I can't play games, and can't stress the CPU... what to do??!?!? anyone?
m
0
l
December 21, 2009 10:54:59 AM

Well, since you're using the intel stock cooler with the terrible push pins, the problem could be that the heatsink just isn't seated properly. Are all the pins securely in place? Since opening the case made quite a bit of a difference, it could also be the cooling in your case. What kind of fans do you have in it, and in what configuration? Also, how much thermal paste did you use?
m
0
l
December 21, 2009 6:21:48 PM

the stock cooler is not great, and I've noticed that it has bad push pins, I'm sure that it's seated as good as it can get, it's not moving a bit...
I read a lot about the thermal paste before I used it, so I cleaned it with pure alcohol, and used as much as needed to cover the middle with a very thin layer of paste... it did made a difference, but not good enough to be able to load the cpu.

I've noticed that the cpu cooler fan is not speeding up at all... it's always is in the same speed, not matter if it's loaded or not...
someone said that it's before it thinks that it's loaded from the beginning so it's on its max all the time, and that's why it's seems not to change speed...
well, I know how fast these things can go, and I feel that it's not doing its best...

I've checked the bios, and I didn't find anything that I can change to make it go faster, only to let it decide.. but it didn't work...

any ideas guys?
m
0
l
December 23, 2009 11:13:26 AM

OK, I've found the problem...
I have a 1600MHz memory from GSkill and the Motherboard from GigaByte that doesn't supprt 1600MHz by default. so the store had to tweek the BIOS settings so they had to speed it up, speeding the CPU caused it to worm so much...

They changed the BIOS setting to work with the memory in 1333MHz speed, and the CPU speed was also increased (again) but didn't caused it to worm...

I don't understand overclocking so much, but it seems that the CPU speed is the same now (About 2770MHz) and the memory on 1333MHz, and it doesn't make it worm, something with multipliers, BUS Speed etc.
m
0
l
December 23, 2009 6:46:47 PM

JackNaylorPE said:
I'm overclocked t 3.7 Ghz under 100% load on the everyday OC profile and getting temps in mid 50's w/ Megahalems. You could try lowering CPU voltage for better temps. I'm at 1.125 at the 3.7


So I've sent some time in the last few days reading about "undervolting" and I got a few questions. If a i7 920 is not OC what voltage does it run at? I know that when you increase the voltage you increase the current or amperage that goes through your processor, and the more amperage you have the more "power" your cpu has. So my other question is: If I can reach a stable OC running 1.25V at 3.6 ghz or 1.1185 at 4.0 ghz which would give me better performance (assuming that the cpu temp is the same in bothcases)? (This is a hypothetical case, I have yet to test this out) In other words which setup would benchmark faster.
m
0
l
December 23, 2009 9:48:22 PM

Less volts = less heat = better. Other than that, it does not directly influence performance.
You can use CPU-Z to see the current voltage to your CPU, and Core Temp or Real Temp (among other programs, core temp might not work with i7) to see the CPU's VID, which IIRC is the 'default' voltage for that particular chip, and can vary between different CPUs of the same type.
For example my Q6600 has a VID of 1.21v, but its undervolted, running at 1.17v
m
0
l
December 24, 2009 8:09:45 AM

irkjab said:
Less volts = less heat = better. Other than that, it does not directly influence performance.
You can use CPU-Z to see the current voltage to your CPU, and Core Temp or Real Temp (among other programs, core temp might not work with i7) to see the CPU's VID, which IIRC is the 'default' voltage for that particular chip, and can vary between different CPUs of the same type.
For example my Q6600 has a VID of 1.21v, but its undervolted, running at 1.17v


I'm not claiming to be an expert, but if lower voltage was better and you did not sacrifice anything by lowering the voltage then no one would ever increase it. From what I understand people buy expensive heat sinks and water cooling systems to dissipate as much heat as possible so they can keep their voltages high and thus the amperage their processor receives. I think there must be some benefit to higher amperage, I just don't know what specifically. Anyone else have any input?

PS: 1.20V cpu and 200 bclk is unstable for me, but 1.2V cpu and 180 bclk is very stable for me.





================================
Edit: Another questions.

If I OC my system from 3.6 ghz to 3.8 ghz and increase the voltage and my bench mark score drops (few percent) does that mean my PSU is not powerful enough?
m
0
l
December 24, 2009 9:07:16 PM

The only thing that affects your processor's performance is the GHz. Your processor will be faster at 4.0 GHz than at 3.6GHz no matter the voltage to the CPU. The only reason to raise the voltage is to allow you to raise the CPU clock speed (more GHz).

As an example, my Q6600 is running at 3.2 GHz, with 1.17 volts. If I raise the clock speed to 3.4 GHz, I'll find that although my CPU is faster, the overclock is no longer stable, and the CPU will start to give me errors. To make my overclock stable, I'd need to increase the volts to the CPU. Increasing the volts to 1.25 would make my overclock stable again.

In short, read this i7 overclocking guide before going any further:
http://www.tomshardware.co.uk/forum/page-253365_11_0.ht...
m
0
l
December 24, 2009 10:37:53 PM

If I understand you correctly, you should only have enough voltage to keep your CPU stable. Any more voltage than that will not increase performance, and could risk overheating your CPU. At high clock speeds you need increased voltage to maintain that stability, but it comes at the cost of heat, which is why OC'ers need great cooling. So if all that is right then I understand the undervolting aspect.

What about my second question. I got a stable OC at 3.8 ghz (while keeping all the other settings the same) with good temps. Then I ran 3Dmark Vantage and got not only a lower total score, but also a lower CPU score. Does that mean I don't have enough voltage? or that my PSU is not powerful enough?

Thanks for all your help guys!!
m
0
l
December 25, 2009 12:11:50 AM

Yup, that's pretty much correct :) 

I wouldn't blame the lower total score on either insufficient voltage or a PSU that's not powerful enough.
Try running vantage a couple of times with the OC and without it, and see if the results with the overclocking average out to higher than without the overclock.
Make sure you close programs running in the background, too.

If it still looks like you get a lower score with the OC, I don't really have an answer. Though, while 3Dmark vantage is a good benchmarking tool, keep in mind that it isn't perfect or 100% accurate
m
0
l
!