water temp vs. cpu temp

spy2520

Weaksauce
Joined
May 11, 2006
Messages
115
generally speaking, how much difference between water temp in the loop and cpu temp is considered normal? like 10-15*C?

also, i would imagine with better setups the difference between the two is less and less. does this rely on the system as a whole? or is it more of what waterblock you use?
 
The answers to both questions rely on the CPU you're using, and what's in your loop.

Generally, water temperatures sould be around 10c above air temperature. And your load temperature should be about (watertemp X 2) for a quad core intel processor... Less for a dualy.

If you have an optimum waterblock mount, the best available waterblock, and more than enough radiator dissipation power... Then the delta between water and CPU should stay relatively constant... it's the delta between CPU and AIR that decreases.
 
how about water temp being 1-7*C over ambient, and cpu around 18-20*C over the water temp at idle? its a pentium d930. load temp after 5 hours at 100% is about double the water temp. loop is on just the cpu. ok i'll admit its an external kit.
 
What is the waterblock you are using, poor waterblocks will transfer the heat less efficiently from the CPU to the water, this would result in a higher water temp -> CPU temp delta.
 
its a zalman reserator xt system. the cpu waterblock is zalman zm-wb5. its just what came with it. nothing special.
 
ok i'm new to this. but i'm not seeing how a better radiator is called for at this time. i understand a prebuilt external kit isnt the best, but something just tells me that since the water temp isnt much different from ambient, but much different than the cpu temp, a waterblock would be a better bang-for-the-buck purchase at this time.

i dont really know though, what am i missing?
 
i also have updated temps, since the system has had more time to settle.

ambient = 29*C
water = 31*C
cpu idle = 42*C
cpu load = 58*C

this is at 4.05GHz @1.53v.
 
ok i'm new to this. but i'm not seeing how a better radiator is called for at this time. i understand a prebuilt external kit isnt the best, but something just tells me that since the water temp isnt much different from ambient, but much different than the cpu temp, a waterblock would be a better bang-for-the-buck purchase at this time.

i dont really know though, what am i missing?

It takes 81w of heat to raise 1L of water 1'c... I'm assuming you have a C2D... Which means you're dumping around 185w of heat into the loop... Which contains ABOUT 2L. Therefore, your water temp should rise around 2c.

But, who knows... I don't have one of these. Maybe it is the waterblock. If you want to update this, then look for a Dtek Fuzion... Arguably the best block on the market right now.
 
It takes 81w of heat to raise 1L of water 1'c... I'm assuming you have a C2D... Which means you're dumping around 185w of heat into the loop... Which contains ABOUT 2L. Therefore, your water temp should rise around 2c.

But, who knows... I don't have one of these. Maybe it is the waterblock. If you want to update this, then look for a Dtek Fuzion... Arguably the best block on the market right now.

how about water temp being 1-7*C over ambient, and cpu around 18-20*C over the water temp at idle? its a pentium d930.

D930 is dual core 3.0GHz presler (65nm)

i think the loop has around 1.25-1.5L of water. i'm just wondering what a different radiator would do. would it drop the water temp to ambient? i really wanna know how that would affect the system. obviously i wanna make it better so any help counts
 
i'm thinking the kit is working pretty good i guess. went from 250fsb(3.75 Ghz) on air stable to 270fsb (4.05GHz) with this kit. thats from starting at 200fsb(3.0 GHz). i was thinking i should get more, but after looking at stock temps and various temps at different speeds, it seems to be doing ok considering there is virtually no noise.
 
It takes 81w of heat to raise 1L of water 1'c... I'm assuming you have a C2D... Which means you're dumping around 185w of heat into the loop... Which contains ABOUT 2L. Therefore, your water temp should rise around 2c.

But, who knows... I don't have one of these. Maybe it is the waterblock. If you want to update this, then look for a Dtek Fuzion... Arguably the best block on the market right now.

Arcy you need to go look over the power consumption of the C2D again lol. 185w of heat is probably more than TWO C2D's put together.
 
Seeing as your coolant temps are good... your issue is likely the c/w rate of the CPU block. This determines how much the temperature of the item you're cooling rises per watt of heat that the block is asked to shift.

EG: You have a processor kicking 120w of heat. You have a waterblock with a c/w of 0.03

This means that for every watt of heat, the items temp will rise by 0.03 deg C, so 120x0.03 = 3.6, and therefore your CPU temp would be [coolant temp] + 3.6 deg C at load.

I work your CPU out to be around 171w of heatload. Your load temp - coolant temp is 27 deg C (ie: Delta from idle to load)

27/171 = 0.157

So at the moment, your current block has a c/w of 0.157 at your current flow rate.

You now need to know what your flowrate is (there are estimators out n' about to work this out for you based on the components you have, tubing and length thereof etc). Once you have your flowrate you can wander round the various manufacturers and find a CPU block that has a c/w of less than 0.157 at your flowrate... buy it, slap it on, and in theory you should see an improvement in temps.

Should be noted that you can lower the c/w rate of your current block by improving contact with the CPU / downforce (ie: you may just have a duff mount, which would explain high cpu temps and low coolant temps - the heat isn't getting dumped into the water)
 
Arcy you need to go look over the power consumption of the C2D again lol. 185w of heat is probably more than TWO C2D's put together.

Not really. I over-estimate the actual thermal output by 10%... Marci and I both got about 171w of real heat dump.
 
Seeing as your coolant temps are good... your issue is likely the c/w rate of the CPU block. This determines how much the temperature of the item you're cooling rises per watt of heat that the block is asked to shift.

EG: You have a processor kicking 120w of heat. You have a waterblock with a c/w of 0.03

This means that for every watt of heat, the items temp will rise by 0.03 deg C, so 120x0.03 = 3.6, and therefore your CPU temp would be [coolant temp] + 3.6 deg C at load.

I work your CPU out to be around 171w of heatload. Your load temp - coolant temp is 27 deg C (ie: Delta from idle to load)

27/171 = 0.157

So at the moment, your current block has a c/w of 0.157 at your current flow rate.

You now need to know what your flowrate is (there are estimators out n' about to work this out for you based on the components you have, tubing and length thereof etc). Once you have your flowrate you can wander round the various manufacturers and find a CPU block that has a c/w of less than 0.157 at your flowrate... buy it, slap it on, and in theory you should see an improvement in temps.

Should be noted that you can lower the c/w rate of your current block by improving contact with the CPU / downforce (ie: you may just have a duff mount, which would explain high cpu temps and low coolant temps - the heat isn't getting dumped into the water)

thanks for that post it is very helpful. my flowrate is adjustable from what looks like .5L/min to 2L/min. the idle temp only changes by around 2*C at the high setting but the fan turns up too so it seems not worth it to have it up, so all temps were at around .5L/min. i'll crack this box open and try to lower the fan voltage so i can run thep ump at full speed without turning the loud ass fan up. i'll look at the waterblocks though.

also i tried pressing on the waterblock to see if temps went down and they didnt change so i assume it is making good contact, i also reseated it once and the TIM showed good contact as well.
 
i
ambient = 29*C
water = 31*C
cpu idle = 42*C
cpu load = 58*C
this is at 4.05GHz @1.53v.
That doesn't make sence.
Ambient is 85*F? Do you live in a oven?
Water is 88*F with a hard OC and passive cooling?
How long was the CPU maxed to get the load temp?
How are you measuring temps?
 
That doesn't make sence.
Ambient is 85*F? Do you live in a oven?
Water is 88*F with a hard OC and passive cooling?
How long was the CPU maxed to get the load temp?
How are you measuring temps?

the ambient temp is probably read a little high because it is what the reserator is reporting. the water temp however, should be pretty accurate since the what the reserator reports and a temp sensor mounted on the waterblock reports are never more than 1*C apart, but usually the same. the ambient is closer to 75*F however, maybe 78-79*F after the computer has been dumping heat into the room all day. right now its been running orthos for about an hour, cpu at 62.5*C, water at 34-35*C. the cpu temp is what the motherboard sensors are telling systool, sandra, pc wizard, etc. sandra is also reporting a "cpu cooling system thermal resistance" of 0.12C*/W which isnt too far from the calculations in this thread.

i wouldnt call it passive, the pump and fan are spinning a little lol (.1GPM). Also, this pump is rated at 5L/min (1.32GPM) but i can only adjust it up to 2. i give it another week or so before i dismantle this thing and put the multimeter to use.
 
messed with the IHS now i'm down in thr 30s*C at idle. gonna check the load temps after i let it run a while.
 
ok, at 4.07GHz after 8 hours of orthos the load temp averaged about 55*C. water temp bounced around between 33-35*C. idle temps are close to motherboard temp (about 37-38*C). i can probably push the clock up a little higher now.
 
Back
Top