CPU Heat

[Jodomcfrodo]

Hey LTT,

So I've been running benchmarks on my CPU for no particular reason and I've noticed something a little strange. When running intel burn test, my CPU temp is jumping all over the place and maxing at 75c. My cooling solution is a custom loop. 5x 120mm only cooling one GPU and CPU. When Intel burn test is running, the air coming out of the rad right after the CPU is not very hot. I just put my hand above the exhaust and I don't notice the temp difference between idle and IBT. Is this normal?

Thanks!

[KamiKatze]

I would take a wild guess and say that the torture test that you're executing is very harsh and your loop can just carry away 'that' much heat at a time. Try increasing the pump speed.

[harrynowl]

IBT is quite nasty to haswell chips, the temps also jump around like crazy normally.

[TheRunningOtaku]

Maybe your Thermal Compound isn't good? Try reapplying it

[runit3]

Hey LTT,

So I've been running benchmarks on my CPU for no particular reason and I've noticed something a little strange. When running intel burn test, my CPU temp is jumping all over the place and maxing at 75c. My cooling solution is a custom loop. 5x 120mm only cooling one GPU and CPU. When Intel burn test is running, the air coming out of the rad right after the CPU is not very hot. I just put my hand above the exhaust and I don't notice the temp difference between idle and IBT. Is this normal?

Thanks!

If you could physically feel the heat coming out of any rad in your system as "hot" or noticeably hotter than ambient something would be going very, very wrong within your loop. (For clarity if ambient was 22C and you took a temp of 38C on the outflow from a raditor=bad news)

 

It is perfectly normal for any air flow coming out of the rad to feel roughly the same as ambient, no matter how hot your card/CPU is. I've never had the liquid within a loop get higher than ~33C and that was after a 24 hour bench with the CPU pinned at 77C. When the liquid within your loop hits a radiator it is rapidly spread out over a very large surface area and cooled by the airflow, that is it's intended purpose. The volume of liquid that the pump is moving should not leave any one area of your loop hotter than another, for example the liquid right after your CPU should "feel" identical in temperature to what you would logically think is the coolest part of your loop (right after both radiators). All of the liquid in the loop reaches an equilibrium temperature very rapidly, that is the nature of heat transfer within water, and why it works so well in cooling hot components.

 

So long answer short, you are having no problems and your loop is functioning perfectly. B)

 

*edit* forgot the IBT portion. IBT wreaks absolute havoc on Haswell's in my experience, the fact that you are only getting a max temp of 77C is somewhat amazing. IBT is going to give you extreme temperatures and temperature fluctuations. It's called the Intel Burn Test for a reason. If you were ever going to OC your 4770K I would not use IBT to test stability, or really anything else, especially given the 4xxx's dodgy line of TIM on die. The only times I've ever used IBT was to test the thermal capabilities of my cooling loops, and that was with the CPU at stock clocks. AIDA64 is my go to synthetic test for Haswell's, and has proven itself as a good validator.

[Jodomcfrodo]

If you could physically feel the heat coming out of any rad in your system as "hot" or noticeably hotter than ambient something would be going very, very wrong within your loop. (For clarity if ambient was 22C and you took a temp of 38C on the outflow from a raditor=bad news)

 

It is perfectly normal for any air flow coming out of the rad to feel roughly the same as ambient, no matter how hot your card/CPU is. I've never had the liquid within a loop get higher than ~33C and that was after a 24 hour bench with the CPU pinned at 77C. When the liquid within your loop hits a radiator it is rapidly spread out over a very large surface area and cooled by the airflow, that is it's intended purpose. The volume of liquid that the pump is moving should not leave any one area of your loop hotter than another, for example the liquid right after your CPU should "feel" identical in temperature to what you would logically think is the coolest part of your loop (right after both radiators). All of the liquid in the loop reaches an equilibrium temperature very rapidly, that is the nature of heat transfer within water, and why it works so well in cooling hot components.

 

So long answer short, you are having no problems and your loop is functioning perfectly. B)

 

*edit* forgot the IBT portion. IBT wreaks absolute havoc on Haswell's in my experience, the fact that you are only getting a max temp of 77C is somewhat amazing. IBT is going to give you extreme temperatures and temperature fluctuations. It's called the Intel Burn Test for a reason. If you were ever going to OC your 4770K I would not use IBT to test stability, or really anything else, especially given the 4xxx's dodgy line of TIM on die. The only times I've ever used IBT was to test the thermal capabilities of my cooling loops, and that was with the CPU at stock clocks. AIDA64 is my go to synthetic test for Haswell's, and has proven itself as a good validator.

Thank you for your help but I believe you are very wrong. The heat needs to go somewhere. The fans are cooling the liquid going through the rad. The air, after passing through the rad, should be warm as that air has just gained heat from the water. The temperature of the liquid in the loop is almost equal but not the same as ambient. While playing bf4, the air is warm. This leads me to believe that the CPU block is not extracting enough heat from the CPU and transferring it to the liquid. The heat while playing BF4 is from the GPU.

[Jodomcfrodo]

Maybe your Thermal Compound isn't good? Try reapplying it

  

IBT is quite nasty to haswell chips, the temps also jump around like crazy normally.

The thermal compound is pretty good. Good contact and all. The only thing I can think of is how tight the block is on the CPU. I can still tighten them more with my hands. I just didn't want to really crank them down. Should it be as tight as possible? (No tools of course)

[runit3]

Thank you for your help but I believe you are very wrong. The heat needs to go somewhere. The fans are cooling the liquid going through the rad. The air, after passing through the rad, should be warm as that air has just gained heat from the water. The temperature of the liquid in the loop is almost equal but not the same as ambient. While playing bf4, the air is warm. This leads me to believe that the CPU block is not extracting enough heat from the CPU and transferring it to the liquid. The heat while playing BF4 is from the GPU.

Using words like "warm" to quantify a finite temperature aren't really appropriate. The outflow air from your radiator will NEVER be hotter than the temperature of the water contained within your loop (unless you were heating your radiator with an outside source).

The human hand is not what I would qualify as an excellent temperature measuring device. When you said it's not "hot" I would assume you were quantifying that as holding a hot cup of coffee, which is around 40 C. If the liquid in your loop at equilibrium was 40C I would call that an extremely bad loop. Having an outflow temperature higher than ambient would indicate your block is extracting heat efficiently and the radiators are doing their job of offloading that heat.

You do not want outflow to feel "hot", in a perfectly ideal loop that outflow air would be exactly ambient, which is unrealistic for most volumes in any given loop. Having an outflow temp slightly higher than ambient or "warm" is what you should be feeling, in an amazingly efficient loop with multiple large radiators it will feel very much like, if not exactly like, ambient using your hand as a temperature device.

Your temperature of 77C under IBT is perfectly normal, as is the rapid fluctuation of temperatures. That is what IBT was designed to do. If you want a more realistic worst case scenario high load test use a different program.

[Jodomcfrodo]

Using words like "warm" to quantify a finite temperature aren't really appropriate. The outflow air from your radiator will NEVER be hotter than the temperature of the water contained within your loop (unless you were heating your radiator with an outside source).

The human hand is not what I would qualify as an excellent temperature measuring device. When you said it's not "hot" I would assume you were quantifying that as holding a hot cup of coffee, which is around 40 C. If the liquid in your loop at equilibrium was 40C I would call that an extremely bad loop. Having an outflow temperature higher than ambient would indicate your block is extracting heat efficiently and the radiators are doing their job of offloading that heat.

You do not want outflow to feel "hot", in a perfectly ideal loop that outflow air would be exactly ambient, which is unrealistic for most volumes in any given loop. Having an outflow temp slightly higher than ambient or "warm" is what you should be feeling, in an amazingly efficient loop with multiple large radiators it will feel very much like, if not exactly like, ambient using your hand as a temperature device.

Your temperature of 77C under IBT is perfectly normal, as is the rapid fluctuation of temperatures. That is what IBT was designed to do. If you want a more realistic worst case scenario high load test use a different program.

Ahh ok I thought you meant The air shouldn't be warmer than ambient.

But I still disagree that a perfect loop would have the exhaust as ambient. That would be an extremely inefficient loop because no heat is being moved from the liquid. If the exhaust was ambient, then the liquid would also have to be ambient meaning that the liquid is not taking heat from the components. An ideal loop would have the exhaust being as hot as possible. That would mean that the liquid is taking a lot of heat from the components and all that heat is being moved out of the liquid and into the air passing through the rad.

At least I think so.

[runit3]

Ahh ok I thought you meant The air shouldn't be warmer than ambient.

But I still disagree that a perfect loop would have the exhaust as ambient. That would be an extremely inefficient loop because no heat is being moved from the liquid. If the exhaust was ambient, then the liquid would also have to be ambient meaning that the liquid is not taking heat from the components. An ideal loop would have the exhaust being as hot as possible. That would mean that the liquid is taking a lot of heat from the components and all that heat is being moved out of the liquid and into the air passing through the rad.

At least I think so.

That would be the case if you had a limited radiator size and there was a 100% heat transfer between the CPU and the water. I'm talking about a situation in which you have an extremely large radiator, or a group of large radiators. The more you spread that heat out over an extremely large surface area, the more cooling potential you have (When I say extremely large I'm talking a hypothetical radiator the size of a table with a massive pump). In either case we're essentially saying the same thing.

 

In your particular case you have 2 very large and efficient radiators, they are more than enough to cool a stock CPU and a fairly low draw GPU (the 780). If you felt HOT (38-40C+) outflow it would indicate that you need more/larger radiators in your loop, the closer you can get the liquid within the loop to ambient temperature the better. If your CPU was at 85-95C and the outflow was very near ambient I would say there is in fact something wrong with your CPU block/mount and there is not enough heat being transferred. What you are experiencing is exactly what a well put together loop should be like, outflow slightly higher than ambient, and incredibly low temps for IBT (I've never had a chip undergoing IBT spike to any less than 90C with a slight OC, part of the reason I almost never use it).

[Faceman]

Set your voltage to manual before stress testing.  If you left your voltage on adaptive, that is probably causing the massive jumps in temperature.

[TheRunningOtaku]

 

  
The thermal compound is pretty good. Good contact and all. The only thing I can think of is how tight the block is on the CPU. I can still tighten them more with my hands. I just didn't want to really crank them down. Should it be as tight as possible? (No tools of course)

 

you shouldn't be able to tighten them with your hands, you should tighten them with a screwdriver until it feels firm, but don't use force

The retention plates are usually metal, it'll be fine

[Jodomcfrodo]

 

 

 

you shouldn't be able to tighten them with your hands, you should tighten them with a screwdriver until it feels firm, but don't use force

The retention plates are usually metal, it'll be fine

 

Well considering they are thumbscrews and don't even have threads for a screwdriver, I kind of assumed they wouldn't have to be too tight. The manual did say how hard to tighten the screws. I'll tighten them a bit. Thanks!