Rad temps versus CPU temps

[mhw100]

When you read rad test results and find radX provides a 2C temperature reduction versus radY, is there a rule of thumb as to how much this will lower CPU temps by? Is it a one to one correlation or something else?

[alacheesu]

Consensus around here seems to be that a reduction in water temp gives an equal reduction in CPU temp.

[SoulsCollective]

In short, no. Too many factors - precision of DTS, how many other components in the loop, how hot those other components are, how powerful pump is and whether the rad with the greater heat dissipation has a higher restriction, fin density on "better" rad as compared to fans used, whether fans are getting air from inside or out...

There's a million and one possible points of difference between your setup and the reviewers setup, and a million and one possible points of difference between your loop setup and theirs. Even if you have the same hardware and the same components in your loops, you're still not able to make any certain predictions about what impact the new rad will have on reported temps other than that it will probably be lower (and if your DTS isn't calibrated precisely enough, you may not see any change in temp at all).

[mhw100]

It varies from different CPU type. 45nm Yorkie quads scale pretty well 1 for 1 on fluid temp reduction to core temp reductions. But 65nm don't., you'll see more than a 1 c average core temp reduction for every 1c fluid temp reduction. I've always said it has to do with the way the DTS probes are calibrated on a 65nm quad. Core temps on a 65nm just don't seem to be linear.

andyc

1:1 on 45nm duals as well?

[mhw100]

In short, no. Too many factors - precision of DTS, how many other components in the loop, how hot those other components are, how powerful pump is and whether the rad with the greater heat dissipation has a higher restriction, fin density on "better" rad as compared to fans used, whether fans are getting air from inside or out...

There's a million and one possible points of difference between your setup and the reviewers setup, and a million and one possible points of difference between your loop setup and theirs. Even if you have the same hardware and the same components in your loops, you're still not able to make any certain predictions about what impact the new rad will have on reported temps other than that it will probably be lower (and if your DTS isn't calibrated precisely enough, you may not see any change in temp at all).

I suppose with the myriad of variables at play I suppose one couldn't rule out a temperature increase?

[Trigger5521]

Sorry I don't want to hyjack but my question sort of applies.

What is a normal variance between air temp water temp and the CPU temp at idle.

My system works like this (idle on Vista desktop).
If room air temp is 22c my water temp is usually 31c and the CPU idle is 40c-42c.

Is that good, bad, normal? How good is a really good setup? I would have thought that the rad would have been able to cool water much closer to air. Granted I'm only using a thin dual rad to cool the whole MB and an OC 9650 quad core.

[Trigger5521]

Sorry, never owned one, so couldn't say.





There is no normal, all depends on the rads and fans. A good above average system will load with a 4-5c delta. Idle wouldn't be the thing to consider, load delta would be. That being said, you're showing a 9c delta under idle, which would be sub par cooling. But I run that chip, and your CPU idle temp doesn't match up with that delta, so I think your measuring your fluid temp wrong, just doesn't add up.

andyc

Thanks, pardon my noob but what does delta mean?

Fluid temp is coming from a sensor in the res. I assume it's accurate. It's always approx 10c above room temp(HW 120 Stealth GT RAD) it's all you can fit in a TJ10 without modding the case. Fans are S-flex 1600rpm. I figure it's sub par, looking at adding a second to get more cooling capacity.

[Trigger5521]

Not noob, just not familiar with the terminology:)

Delta (difference) T (temperature). Delta is a term used to denote the difference between two data points. In your case, air temp and fluid temp. We use that term Delta T to describe what your talking about.

I run that same chip on my rig OC to 4.25, 1.35v. I set my rig to idle to get a base point for you. 40-40-35-35 and my fluid temp is 25.56c. We won't compare the Delta between air and fluid temp, that's not the point. The point is, if my fluid temp was 31c right now, I'd be in the low to mid 50's at least, maybe even higher. So the temps your describing don't add up unless there some strange anomaly happening between our two chips and systems.

So I'd say either your reading your fluid temps to high, or your CPU core temps to low. Just doesn't add up. What your full load fluid and core temps?

andyc

Here's a quick screen grab. Room air temp is 23c right now. All my specs are in my sig if it matters.

So is this good or bad??? Intelburntest will put CPU to 80c+ and Prime95 about 60c -65c average. Full load fluid temps can get to 42c.
.

[Vapor]

Trigger, how is your airflow going into the rad arranged? 'Fresh' air going straight into the rad? Or using the radiator for exhaust duty as well?

There's also the distinct possibility your rad temp sensor just isn't behaving/performing properly of course :p:

[Sadasius]

With those volts, not really bad, but for the life of me I can't see how your fluid temp can actually be that high. The funny thing is, those core temps aren't bad at that voltage, using the rad you have listed.

So I don't know to tell you except don't worry about it, your rig at that OC must be running fine regardless.

andyc

It could be the placement of the water temp sensor. If it's near a heat source then it is not exactly going to be accurate.

[fox3]

I do not mean to hijack the thread my apologies to the OP.

I have been searching the forum for acceptable temps with pretty much the same setup. Cuplex di and full BP chipset for the asus RE, Q9650 OC to 4.32, vCore 1.38. My temps at idle are 35,35,33, 29 (stuck sensor @ 29°C). Ambient is 22°C,at idle water into RAD 24.3 °C, water into RES 23.9°C. After 6 hours Prime CPU cores is 60°C and 20 runs IBT CPU cores is 72°C. I was going to split the loop but now it seems my temps are acceptable.

Should I consider putting the chipset with the GPU loop or can I add the GPU to this loop I have another feser 480. :shrug:

*edited CPU cores

[Waterlogged]

mcoffey, Vapor...look at his sig, it'll become painfully obvious. ;)

Trigger, what fans and pump are you using?

[SoulsCollective]

I suppose with the myriad of variables at play I suppose one couldn't rule out a temperature increase?Only in that a larger or more restrictive rad could reduce flow-rates.

[Trigger5521]

With those volts, not really bad, but for the life of me I can't see how your fluid temp can actually be that high. The funny thing is, those core temps aren't bad at that voltage, using the rad you have listed.

So I don't know to tell you except don't worry about it, your rig at that OC must be running fine regardless.

andyc

I think read somewhere here that CPU temps are inaccurate at anything less than 40c maybe that's why your water at 25c doesn't make your idle temps report less???.

The whole point of my question was to figure out if water 10c above room temp is ok or should I be looking at another rad to get it closer to room temp.

Vapor...rad and fans are inside the case blowing out, so fans are exposed to case air temp. Humm I should put my air thermometer in the case to see how warm the air going through the RAD is. Case air temp has to be higher than room temp.

Sadasius...water temp sensor is in the bottom of the res(pics in my sig), the res is directly after the rad so in theory it should be the coolest water.

[Trigger5521]

mcoffey, Vapor...look at his sig, it'll become painfully obvious. ;)

Trigger, what fans and pump are you using?

Scythe SFF21F S-Flex 1600 rpm. Pump is Swiftech MCP655.

I know the rad I'm using is not favorable but there are only a few rads that'll actually fit in this case, stealth being one I could acutally order in canada.

[Jhors2]

You might want to turn those fans around. It's better to have the funs sucking ambient air in as its fresher and not victim to the heat inside of the case.

[Trigger5521]

No...I would consider that very poor and good luck to you

andyc

Thanks again, I think I'll try fans reversed to see if sucking room temp air will help.

Cheers!:up:

[Waterlogged]

Scythe SFF21F S-Flex 1600 rpm. Pump is Swiftech MCP655.

I know the rad I'm using is not favorable but there are only a few rads that'll actually fit in this case, stealth being one I could acutally order in canada.

Actually, in addition to the rad being a sub par model, I was suggesting that you have an awful lot of stuff in your loop. So much so that I personally think you don't have enough rad.

[Trigger5521]

Actually, in addition to the rad being a sub par model, I was suggesting that you have an awful lot of stuff in your loop. So much so that I personally think you don't have enough rad.

Right...whole MB is on the loop AND my NB is at 1.60volts to handle 8GB of RAM over clocked. This is not helping things here.

I'm going to check my inside case air temp tonight and measure that as the distance of difference. If room is 22c but case is say 25-26c then water at 32c is not completely retarded.

Then I'll reverse the fans and try sucking cooler room temps through rad... see if water goes lower.

Thanks for all the info guys...love this stuff...soooo much to learn. I want to build another rig just to try all the new idea's I have :) addictive no doubt.

[Trigger5521]

You might want to turn those fans around. It's better to have the funs sucking ambient air in as its fresher and not victim to the heat inside of the case.

WOW!! I just reversed my fans to suck room air through the RAD instead of pushing case air out. Looks like a 2-3c drop in water temp.

Thanks Vapor & Jhors2!!!