Water temp lower = lower cpu temps???

[Shadowman]

Was just curious if a cpu was overclocked and say the water temp was hitting like 40c on full cpu load and then say everything stayed the same ( cpu overclock, voltage) and the ONLY thing added was a extra radiator.

 

With the extra radiator the water temp know was hitting  34c on full cpu load would the cpu temperature also drop and when I mean drop like 5c+ ?

[Jurrunio]

should be, heat dissipation depends on temperature delta after all

[Mira Yurizaki]

If you're increasing the radiator size, all other things equal, then the CPU temperature should drop.

[Shadowman]

Thanks, but realistically what could you expect? I mean if the difference between 40c and 34c is like 6c should that be what should be expected from the cpu temperature also a drop of 6c?

[NotSoEpicMods]

Something of note there's also a point at which you'll get diminishing returns so yes the temp should go down but it might not go down as much as you might want. You did say that the water temp went down 6c so you should see a cpu temp drop.

 

Edit: I wouldn't expect 6c drop maybe like 4c is more realistic because of thermal transfer losses through IHS and thermal paste etc.

[Jurrunio]

Just now, Shadowman said:

Thanks, but realistically what could you expect? I mean if the difference between 40c and 34c is like 6c should that be what should be expected from the cpu temperature also a drop of 6c?

same as liquid temperature? As I said, heat transfer depends on temperature delta. Same heat output = same temperature delta. Transistors also work a bit more efficiently at lower temperatures, but it can be ignored for this small temperature change

[Enderman]

No, it also depends on flow rate and turbulence and plenty of other factors, not just water temp.

[Jurrunio]

7 minutes ago, Enderman said:

No, it also depends on flow rate and turbulence and plenty of other factors, not just water temp.

Those are kept the same

[Enderman]

2 minutes ago, Jurrunio said:

Those are kept the same

You would think, but temperature changes viscosity and heat capacity and therefore reynolds number and nusselt number.

 

This means the heat transfer is not linearly proportional to temperature delta which means the CPU temp is not going to change exactly at the same rate as the water temp.

It will be very close, but not exact.

[Midnitewolf]

I am not sure but I think your asking if a 6 C reduction in load temps is realistic when adding an additional radiator.   

 

Well the first thing you should consider is that cooling can't go lower than room temperature no matter how many rads you have mounted.  If your 40 C at load and you have 22 C room temp, you only have a maximum of 18 C to work with.  If you are seeing your load temps drop to 34 C, that 6 C difference amounts to a 33% reduction in the overall heat.  That is a pretty significant decrease when you think about it that way.  The problem is that the lower you go, the more the returns diminish.  For example adding a 3rd radiator might result in another 33% reduction but since the maximum you could reduce the temps is now 12 degrees, that 33% results in only a 4 degree reduction or all.  Add a 4th radiator and you only get a 2.5% reduction.  This of course also depends on flow rates and such as well but you get the picture.

 

 

[Jurrunio]

11 minutes ago, Enderman said:

You would think, but temperature changes viscosity and heat capacity and therefore reynolds number and nusselt number.

 

This means the heat transfer is not linearly proportional to temperature delta which means the CPU temp is not going to change exactly at the same rate as the water temp.

It will be very close, but not exact.

would this even be measurable when read through thermistor inside the CPU though since it only reads to the nearest degree? I know Ryzen reads better but he is on a 7900X from his previous post

[Enderman]

Just now, Jurrunio said:

would this even be measurable when read through thermistor inside the CPU though since it only reads to the nearest degree?

Well I don't know exactly how significant the difference will be, but even if it's an extremely small difference it's definitely possible for a single digit difference.

If one temp is .49 and another is .5 then even if it reads to a single degree it will round down on the first one and up on the second.