• We should now be fully online following an overnight outage. Apologies for any inconvenience, we do not expect there to be any further issues.

i7 950 overheating on stock frequency :(

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Rifter

Lifer
Oct 9, 1999
11,522
751
126
2v8ivtl.jpg


ppl are running ther i7's at 35-38*c on stock cooler :(

intel C-state tech shud i enable it ?

I'll try this one final time, you are not overheating. And NO ONE is running their bloomfields at 35-38C under load on air with up to 35c ambients, thats not even physically possible.
 

crazy.wingman

Senior member
Jan 5, 2011
243
0
76
made an huge improvement in load temps, reduced voltage now my chip runs 12*c cooler at full loads but no change in idle temps its still 43-44*c

CPU 1.075V
CPU PLL 1.80V


6if8de.png


Cpu pll is at its least voltage.
any suggestions shud i lower cpu voltage more ?
 
Last edited:

Sephire

Golden Member
Feb 9, 2011
1,689
3
76
guys, This is real serious now.

i have changed my thermal grease now to Tunig TX 3 n added 2 more 120mm fans to my case but still my cpu idle temp is 42º-45ºc ? :(

1hr prime 95 stress gets it to 72ºc :( :(

overnight test will surely blowup my pc.

Like everyone has said to you nothing is out of the ordinary here. Its typical. Try one of those 120mm Zalman coolers if you want better cooling than what you have right now. And add more intake case fans in front of your case (I have two in mine).
 

Ben90

Platinum Member
Jun 14, 2009
2,866
3
0
WOA! 1.075V is a very dangerous Vcore. No wonder you are overheating. These i7s need much more than that, when the voltage is low it is forced to draw more amperage through the socket creating a lot of heat. Try setting it to 1.5v and if its still overheating, increase it slightly from there. Most people on this forum try to keep their i7s at 1.4v or higher.
 

crazy.wingman

Senior member
Jan 5, 2011
243
0
76
I'll try this one final time, you are not overheating. And NO ONE is running their bloomfields at 35-38C under load on air with up to 35c ambients, thats not even physically possible.

you are getting me wrong, the :mad:idle:mad: temps for i7's at stock shud be in 30's n ppl do have tht temps on idle with stock cooler with ambients of 28-30*c.
 

crazy.wingman

Senior member
Jan 5, 2011
243
0
76
WOA! 1.075V is a very dangerous Vcore. No wonder you are overheating. These i7s need much more than that, when the voltage is low it is forced to draw more amperage through the socket creating a lot of heat. Try setting it to 1.5v and if its still overheating, increase it slightly from there. Most people on this forum try to keep their i7s at 1.4v or higher.

1.075V is the CPU voltage shud i reduce it to 1.4 or 1.5 ? coz the HT,turbo mode an speed step is enabled.
 

crazy.wingman

Senior member
Jan 5, 2011
243
0
76
Like everyone has said to you nothing is out of the ordinary here. Its typical. Try one of those 120mm Zalman coolers if you want better cooling than what you have right now. And add more intake case fans in front of your case (I have two in mine).

i can't get zalman's in india, in india its just out of this world thing.
my case panels are open so no problem with the air flow, the least i can do is
try reducing the volttages more.
 

Sephire

Golden Member
Feb 9, 2011
1,689
3
76
Stop reducing the voltages! Put it back to stock settings!

Stability is what you need. Those people who has lower temperature than you may have better case cooling, better CPU cooling, lower ambient room temperature (airconditioned room), CPU temp taken during winter months.

You are fine. Or move to Aspen Colorado USA.
 

Drsignguy

Platinum Member
Mar 24, 2002
2,264
0
76
WOA! 1.075V is a very dangerous Vcore. No wonder you are overheating. These i7s need much more than that, when the voltage is low it is forced to draw more amperage through the socket creating a lot of heat. Try setting it to 1.5v and if its still overheating, increase it slightly from there. Most people on this forum try to keep their i7s at 1.4v or higher.



What? :confused:
 

Ben90

Platinum Member
Jun 14, 2009
2,866
3
0
Well since he doesn't listen to anything, I figured it didn't really matter what I posted. BTW for all you lurkers out there... DON'T run 1.4v+ to your cpu.
 

crazy.wingman

Senior member
Jan 5, 2011
243
0
76
Well since he doesn't listen to anything, I figured it didn't really matter what I posted. BTW for all you lurkers out there... DON'T run 1.4v+ to your cpu.

like you said i was trying to reduce the voltage, i tried as much as possible, right now i am running the chip at 1.03750v coz on turbo mode on full load cpu-z shows max 1.32v usage so i kept my cpu voltage at 1.0375v

my cpu is not overclocked its on stock, i am not getting wht you said before 1.5v for a overclocked cpu or for my stock cpu ?
 
Last edited:

ensign_lee

Senior member
Feb 9, 2011
401
0
0
You DO NOT want your voltage at 1.5V

Put your voltage settings back to "AUTO" on your motherboard and STOP FREAKING OUT about your temperatures. They are well within reason. Hell, on this older computer that I'm typing from right now, at 50% load, the comp hits 60-70 anyway. AT 50% LOAD.

So stop freaking out. Everything was well within spec before. You're only damaging your computer more by fiddling with it.
 

Seero

Golden Member
Nov 4, 2009
1,456
0
0
like you said i was trying to reduce the voltage, i tried as much as possible, right now i am running the chip at 1.03750v coz on turbo mode on full load cpu-z shows max 1.32v usage so i kept my cpu voltage at 1.0375v

my cpu is not overclocked its on stock, i am not getting wht you said before 1.5v for a overclocked cpu or for my stock cpu ?

Forget voltage if you can't get temperature right. Temperature is a function of many factors, and heatsink is only one of them. Reducing vcore without understanding its implications will lead to an unstable system. Note that parts will age and what works now doesn't mean it will work later and default vcore ensures that the system will still be stable after years of use.

If you really have a problem with the high temp, then set all fans in the case to max, but 70-80 under prime95 is okay. There are mechanism within the CPU that protects it from dying from temp as long as the heatsink is properly mounted, and you heatsink is properly mounted, else you won't be able to run prime95 at all. People who runs prime95 for days is not because of temp, but the stability of voltage and the system itself.

Half hour of FFS will force your CPU to reach its terminal temps. If it goes higher after that, then either hot air is circulating back into the airflow of the system, or cooling to the surface of the mobo is insufficient, generating heat towards the airflow. That is 1-2 degree you don't really care if you are running of shock speed and voltage. To overclockers, that 1-2 degree may lead to instabilities due to voltage, again, temp at that point isn't an issue.

Note that, by decreasing vcore meaninglessly, your system may hang, BSoD with reasons that you never thought of. Days of testing is needed whenever vcore is changed, and all changes must be written down somewhere so manual rollbacks can be done.

Do not get obsessed with temps, I can tell you from experience because I am obsessed with temps myself. Of course there are things that can be done to reduce the temp, but the question is, is it necessary? I will share them with you, but you really don't need to do it, it is perfectly fine as it is.

Those are from easy to hard, to insane.

Add a fan to the front of the case, meaning you have to remove the drive cover and find a way to mount a fan on.

Channel the airflow path so that there is exactly one intake from the front of the case towards the intake of the HS and no wire is blocking the way.

Learn and apply TIM so it archieve maximium efficiency.

Buy an expensive heatsink.

Buy an expensive case with lots of fans and siliencers.

Turn AC high, 24 hours a day, everyday.

Go liquid cooling.

Buy an expensive motherboard.

Get water block on everything.

Multi independent water loops.

bigger radiators.

More fans.

Turn AC to max.

Learn and Lap the surface of the heatsink.

Learn and Lap the surface of the IHS.

Replace all caps on the motherboard.

Vdroop mod.

ditch your current chip and pay premium price of the chip that is of highest quality of the batch.

sub-zero cooling.

2 phase cooling.

Nitrogen cooling.

Multiphase cooling.

If you seek for greater knowledge about cooling, post in the cooling forum and there are more masterminds who can satisfy you.
 
Last edited:

ThatsABigOne

Diamond Member
Nov 8, 2010
4,422
23
81
Wingman, read the effing posts lol.

YOU ARE NOT OVERHEATING. YOUR TEMPERATURUEs ARe GREAT!!!
 

Zap

Elite Member
Oct 13, 1999
22,377
7
81
you are getting me wrong, the :mad:idle:mad: temps for i7's at stock shud be in 30's n ppl do have tht temps on idle with stock cooler with ambients of 28-30*c.

No they don't.

You said your ambient temperatures get up to 35°C, and you're wanting your CPU to run lower than that? That is physically impossible with either air or water cooling.

Let me repeat what most others are saying in this thread:

Your CPU is not overheating.
 

crazy.wingman

Senior member
Jan 5, 2011
243
0
76
my ambient temps are 28-32*c like what its in my 2nd reply *check*

n i have said idles at 43-44*c and jumps to 55-58 to 65-69*c and goes on till 79*c within few seconds in turbo mode an thats not normal the chip is not overclocked an its still not summer here.
in summer the temps will be over 80*c for sure

maybe its an voltage issue, so i needed some help here.
 

Seero

Golden Member
Nov 4, 2009
1,456
0
0
my ambient temps are 28-32*c like what its in my 2nd reply *check*

n i have said idles at 43-44*c and jumps to 55-58 to 65-69*c and goes on till 79*c within few seconds in turbo mode an thats not normal the chip is not overclocked an its still not summer here.
in summer the temps will be over 80*c for sure

maybe its an voltage issue, so i needed some help here.
Hmm, the original post stated that it takes 4-5 minutes of prime95 to raise temp to 72c, now you are saying that it goes to 79c whenever turbo mode is on for a few second. I believe you are imagining things.

Turbo mode only kicks in when not all cores are used and will not exceed its TDP (Max temp before it hangs). Turbo will either burn your CPU as hard as Prime95, or even bring temps up as if all cores are in use. How was it possible that you get to 79c in seconds with turbo but take 4-5 minutes to 72c? You are thinking too much.

Heat generation is a function of voltage, and under default voltage your CPU won't fry due to heat as long as there is a Heatsink sitting on top of it. If you really believe your system is overheating, run prime95 until it overheats, and do not assume that it will overheat. Under normal voltage and clock speed, thermal control within the CPU will kill its power to CPU in about 10us when the center of the CPU(Tcase) reaches 67.9c, that is approximately 90c at core(TjMax), and it takes more than 90c to melt anything. Before the actual reading off the thermal diode(Tjunction) reaches 0, its reading is inaccurate. Besides, TjMax is a predefine number that users can change within any temperature monitoring software because the real TjMax of each CPU is different. Whatever temp you read is TjMax - Tjunction. One is a predefine number that is not accurate, the other is a number that is only accurate when it is near 0.

Really, if you really have a concern, run prime95 until it throttles. Chances are, it won't after days.
 
Last edited:

Ben90

Platinum Member
Jun 14, 2009
2,866
3
0
-Tcase is basically impossible to measure correctly and has no bearing on whats reported over PECI
-#Thermtrip activates within 500ms of an unknown value over Tjmax


I completely agree that you should try to make your processor throttle running stock settings. You won't hurt anything doing it and it will allow you to actually have insight into seeing if there is a problem other than "OMG GUYSSZ MY TEMPS ARE LIKE TOTALLY IN CONTROL FOR LIKE ALWAYS BUT MAYBE THEY COUDLNT' BE IF I RAN FOR LIKE 1000 HOURS AND MY CPU WILLL MELT"

I also recommend using Realtemp or something that tells you what the distance to Tjmax is directly. Although I believe most programs will read 5*C too hot for Bloomfield processors because I could swear the max is 95*C. I know we eventually got an answer strait from Intel, but I'm too lazy to dig it up.
 
Feb 19, 2001
20,155
23
81
I can hit 85 in Linx. 75 in gaming typically.

REalize that Prime and Linx aren't real world tests. They max out your CPU. 72 is not overheating, and going above 90 is a problem. Most OCers recognize 80s as high as they should go in Linx and 24/7 real world scenarios to really never peak above 80. Gaming should be the hottest it gets because the GPU heats up and the CPU heats up. Even then I never go past 80.

BTW I have an i7 930 @ 4.01ghz.

These chips are built to tolerate 100.

OP, what thermal grease do you use? Try some IC Diamond 7 and the difference can be pretty big. Easily 5C difference. I played around with Arctic MX3, AS5, IC7. IC7 seems to be the best.
 
Last edited:

SHAQ

Senior member
Aug 5, 2002
738
0
76
I idle at 38C but that is at a 23C ambient temp. 35C ambient would force mine to 43C or higher I'm sure. Probably actually 50C since I am 15C above ambient temp. CPU temperature means nothing without knowing the ambient temperature. If someone tells you what their CPU temp is ask them about the ambient temperature.
 

tweakboy

Diamond Member
Jan 3, 2010
9,517
2
81
www.hammiestudios.com
Though 70's is hot, the chip is made to create heat . idle doesnt matter, on load

if your in 50's and 60's then that is good. 72c is pushing it Also 70's will still run and it will have a long life hopefully.
 

crazy.wingman

Senior member
Jan 5, 2011
243
0
76
Hmm, the original post stated that it takes 4-5 minutes of prime95 to raise temp to 72c, now you are saying that it goes to 79c whenever turbo mode is on for a few second. I believe you are imagining things.

Turbo mode only kicks in when not all cores are used and will not exceed its TDP (Max temp before it hangs). Turbo will either burn your CPU as hard as Prime95, or even bring temps up as if all cores are in use. How was it possible that you get to 79c in seconds with turbo but take 4-5 minutes to 72c? You are thinking too much.

Heat generation is a function of voltage, and under default voltage your CPU won't fry due to heat as long as there is a Heatsink sitting on top of it. If you really believe your system is overheating, run prime95 until it overheats, and do not assume that it will overheat. Under normal voltage and clock speed, thermal control within the CPU will kill its power to CPU in about 10us when the center of the CPU(Tcase) reaches 67.9c, that is approximately 90c at core(TjMax), and it takes more than 90c to melt anything. Before the actual reading off the thermal diode(Tjunction) reaches 0, its reading is inaccurate. Besides, TjMax is a predefine number that users can change within any temperature monitoring software because the real TjMax of each CPU is different. Whatever temp you read is TjMax - Tjunction. One is a predefine number that is not accurate, the other is a number that is only accurate when it is near 0.

Really, if you really have a concern, run prime95 until it throttles. Chances are, it won't after days.

agree but i have mentioned the Asus EPU turbo mode before and in prime 95 the temps jump to 44*c to 68-70*c within few sec's n after 5-10 mins it goes beyond 75. i have figuredout this all happens only in turbo mode n in asus EPU high performance mode the temps are 65-68*c.

for 1 hour prime 95 it was almost 80*c in real temp but we are taking abt the sudden jump of temp by 12-15*c, why is it so ?
Why the hell Asus EPU shows turbo mode on when i am running prime95 with all 4cores n 8threads ?
 
Last edited:

Ben90

Platinum Member
Jun 14, 2009
2,866
3
0
Honestly if your not going to overclock do this:

1. Reset bios to defaults.
2. Don't worry about overheating.
 

Seero

Golden Member
Nov 4, 2009
1,456
0
0
agree but i have mentioned the Asus EPU turbo mode before and in prime 95 the temps jump to 44*c to 68-70*c within few sec's n after 5-10 mins it goes beyond 75. i have figuredout this all happens only in turbo mode n in asus EPU high performance mode the temps are 65-68*c.

for 1 hour prime 95 it was almost 80*c in real temp but we are taking abt the sudden jump of temp by 12-15*c, why is it so ?
Why the hell Asus EPU shows turbo mode on when i am running prime95 with all 4cores n 8threads ?

First of all, the following isn't a guide to OC, but either things you should already know before reading guides about OC.

The temp jump is normal. What you don't know is there are several power saving features from bios which reduce the amount of electricity going into the CPU. Some people are confused, or never knew how it works to begin with, but you can't change amount of electricity to drive the CPU.

CPU itself doesn't use much electricity to work. What requires electricity is the signal going into the CPU. In layman terms, when there are electricity, than CPU understood as 1, when there are no electricity, then it is 0. From 1 to 0 and back to 1 you have 1 cycle, and out of this cycle, half of the time it generates heat as it receives electricity as a signal which it reads it as 1.

If you have ever played with light switches by flickering it really fast, you probably know that the state of the lightbulb is neither on or off as the lightbulb requires time to charge up and discharge completely. In short, this event happens within CPU too, except that it happens really fast. 1 hz = 1 to 0 back to 1 onces a second, and your CPU flickers the power switch much faster than you can imagine. The speed of this act is also known as clock speed.

Imagine I am playing at the switch and you are looking at the light, will you be able to determine the speed I flicker the switch? You may say yes assuming whatever I do impact the light, but what you may not have consider is there are hundreds of people like me playing this in the same house. While it will be fun to play, you probably won't be able to determine that because every action anyone does with electricity have an impact to the lightbulb you are trying to use to determine my flickering speed. To make it more complicated, I will try to send a morse code through flickering the switch and you are to read the morse code send by me.

Whether you will be able to successfully receive the morse code I sent depends on the speed which I sent it with, the amount of interference you get, the charge/discharge time of the lightbulb, and the brightness of the lightbulb. The brightness of the lightbulb is important as you can determine the difference between 0 and 1 more easily in 2 ways. 1) the charge/discharge time decreases, and 2) the signal is more clear. While kids are busy inside sending messages with lightbulbs, heat is being generated. Each wire that connects the switch and lightbulbs itself generates heat.

If CPU is a room, then it looks pretty much like the room described above with millions of parts trying to do the send messages. To ensures that each and all of these messages are sent correctly, electricity going into this house must be very stable, and increasing the voltage means signals are more clear inside.

But what happens when no one is sending messages inside, well then CPU consumes less electricity, but there is always one kid who is playing, that kid is called the clock generator. However, when the activity inside the house it low, it means the interference is low, and therefore doesn't require the voltage needed as if millions of activities are going on, and therefore there are features in bios to reduce the voltage going into the house. Even if the voltage remind unchanged, the fact that there are no much activity going on, the amount of heat generated within the house decreases.

One 9volt battery can generate lots of heat, but that is when circuit is closed. If the circuit is open, then no heat is generated. This also happens within the CPU.

By default, Bios supplies about 1.2 volt into the CPU. At idle, since there are not much activity going on, and therefore not much heat is produced. Generally speaking, the heat from the CPU at idle is from the electricity needed to keep CPU running + the heat from the clock generator. At load, heat is produced based on the types of activities occured. The other way to predict heat generation is by measuring the amount of electricity consumed by CPU. At load, CPU consumed far more electricity then when it is idle.

Like tweakboy said, you can't stop heat from generating or reduce the amount it generates. You can however radiates the heat generated by off the CPU. When switching from idle to load, temperature will rise no matter what method you use to radiates the heat. The rate of heat trasfer is by itself a huge topic, but you don't need to know how does it work as long as it works. If contact between the heatsink and the integrated heatsink on the surface of the CPU is a problem, then the temperature will rocket even before your windows starts. That means BSoD. If it starts, then you run prime95 to ensure the signal is clear and the msgs sent within the CPU is indeed correct, else prime95 will show errors or BSoD.

The purpose of using Prime95 to test radate capability is to see if heat actually gets out of the case or circulating within the case. If you can run prime95 for without crashing for 1 minute, then the contact is fine, meaning the TIM is fine. If you can run it for 1 hour, then the Heatsink is up for the task of radiating heat off the CPU and that speed and voltage. If you want to know how much headroom you have for OC, the number you what to see is the "distance to TjMax", not temperature. If it is under 10, then you don't have alot of headroom. If it is under 20, then you can push the clock higher until voltage becomes a problem. If it is above 30, then you can play with the voltage gently.

In 1 minute of prime95, heat probably got from CPU to heatsink, you don't know if heat is being radiated of the heatsink. At around 30 minutes, temps should be raising very slowly. If temp increases after that, then it means the ambient temperature around the heatsink increased. After 1 hour, if temp increases, then the ambient temperature within the case probably increased. After several hours, the room temp increases.

If torturing your PC isn't the point of all these tests, 1-2 hour of prime95 is enough to ensure stability of the PC. In your case, everything work as intended.
 
Last edited: