Effect of Pump on Temps

R

RA3KW0N

n00b
Joined
Jul 22, 2008
Messages
3
Hi,

I've been watercooling for several months now, and when I compare my temperatures to high-end air-cooling, I don't feel fully satisfied (except when I look at the neatness of watercooling vs. having fans everywhere). The reason is that my temperatures seem to be quite high for my CPU.

This is my loop:
pump --> 360mm rad --> Q6600 @ 3.2ghz (Swiftech Storm Block) --> 240mm rad --> 8800gtx (Dangerden Block) --> 8800 gtx (Dangerden Block) -->

These are my specs:
Q6600
EVGA 780i
OCZ PC2-8500 4x 1GB w/ OCZ memory cooler
2x 8800GTX in SLI
Vista Ultimate 64-Bit

The addition of the 240mm rad was pretty new (I got it for cheap, thought I'd try it out) but didn't really add any benefits to the temps. And my GPUs are pretty chilly at 40C-45C (I like this).

Now, my CPU is running at low 60s - low 70s on full load, with only 1.34V being put into it. I've suspected that the contact between the heatsink and the CPU might've been the culprit... always trying to tighten the bolts and press down on the Storm block, but the Storm is nice and flat, and I lapped my Q6600 (the lap is nice and flat) as well.... applying thermal paste isn't an issue either, it's been placed pretty evenly, thinly, and well.

The only thing I can think of is that the pump may not be up-to-par. It's a 12V one but only about 600L/h. I just wanted to know if the pump may have any effect on the high temps of my CPU. Would getting a better pump reduce my temps? I can get 3.6ghz with <1.5V on the CPU but the temps just go too high to make it feasible. Please give me your input, thanks!
 
In your case, yes, the pump can have an effect on the temps.

Your loop is very restrictive with the loop having to go through so many blocks and two rads.

The 320 should be sufficient for your loop, adding the 240 to your loop did you no favors without a pump that can effectively circulate the water through the loop.

LPH doesn't matter as much as head/pressure in your case.


The other problem is: how are you measuring CPU temps? Are you only looking at the on board sensor reports? If so, those can be VERY inaccurate.
 
A higher pressure pump would help because of the restrictive cpu block design. I feel the gpu blocks because of their design (Low flow restriction low pressure drop ) are not part of the issue. The rads are restrictive by design and there is not a lot you can do about it.

Frankly I do not like the restrivctive jet impingment type blocks with cpus with heatspreaders.

From swiftech about your block.
While this design yields performance nominally equivalent to the MCW6000 series water-blocks in systems flowing 0.75GPM or less, it really starts shining in the 1 GPM and above range
Click to expand...

1 US gallon = 3.78541178 liters
1 gallon per minute = 60 gallons per hour
60 x 3.78541178 = 227 liters per hour.

So if you pump can produce even 1/2 its rated flow it should be sufficient but that brings us back to the question of if that pump can produce that kind of flow with the 2 big restrictive rads which depends on its ability to provide pressure.

Without knowing the specifics of your pump it seems more than able to provide sufficient flow unless (forgetting the gpu blocks because they are non-restrictive) the rads are too much for it to handle. Without some PQ curves for the pump the jury is still out. However lets assume you do not want to reduce or change the rads due to effort/expense and frankly a flat tube rad is a flat tube rad and it would be difficult to find rads with enough specs published to even start to decide if they are better, and by how much, over what you have.

So, this leaves - bigger pump (more pressure) or change the block to a low restriction type.

Me personally, I would change the block and go with a low restricion high flow system where the rads are the only restrictions.

While I was thinking about it, I would cut a spare length of tubing and take the 240 rad out of the loop and see what happened to temps without touching anything else.

Need to know exactly what pump you have. 600l/h iis not garbage.

What fans exactly ? Rads dont cool unless there is air flowing thru them. Perhaps you need to dump the low speed yate loons or whatever for some medium speed fans on a controller for some more cfm.

You are jumping on the pump as the culprit too fast IMO (but it worries me you didnt mention the model/brand).

low 60s with 4 cores loaded is not bad. low 70s is not so good (but no where near any danger levels).

You did not mention if that is with the GPU's being stressed as well .
 
Agree with Bill. Take a hard look at the fans you're using. Not too long ago CW dictated that mutiple rads wouldn't work as well as a single, large, well-blown rad. The idea was that having the warmest intake Delta would bring a good rad very close to its design potential, negating the need for multiple rads.

Also, while the Storm is still a damn fine block for single core, having it inline starves that room heating SLI rig of yours or what it needs, flow. ;) I've got a B3@3.4 (DD MC-TDX) and a 8800GTX (DD block) in an very HF loop, and it'll turn my room into a furnace with the door closed. With fresh air feeding my rad I get sterling temps though, and that's with 2 pumps.

Agree with Bill again. Consider replacing your CPU block and going with beefier fans. There are fans that have high static pressure that aren't very loud.
 
Thanks everyone, especially Bill (very detailed explanation), I greatly appreciate it. My pump is a thermochill pump, not exactly sure about the model number though.

I posted this on another forum and also had someone suggest that the CPU block was not sufficient for a quad-core and recommended a fuzion or apogee. I'm going to grab a fuzion and see what results I can get from there as I think my pump is decent (your expert opinions may prove me wrong though).

And sorry, I was mixing load temps (CPU) with idle temps (GPUs). Tired + trying to formulate sentences = :confused:. lol. The GPUs, even when stressed, stay frosty and dont go much higher than low 50s.

If the pump is the culprit then at least when I replace the pump as well I'll have a better system with the new cpu block. I'm too lazy to take out the 240mm rad atm lol. Also, I'm using the highest CFM(right units?) Scythe S-Flex fans on my rads.... quiet yet powerful, and sexy.

I was also using CoreTemp/Everest to measure my temps for CPU and Everest/GPU-Z to measure GPU temps... all accurate and within 1C of each other.

Thanks for your help guys. me --> happy camper.

Hopefully after the modifications I can push my Q6600 to 3.6 if temps stay frosty. Also, it helps now that I live in a ridiculous above-the-garage room where summer times makes my room a furnace, and when winter times are arctic temperatures.

But ugh... I am not looking forward to removing my mobo to install the backplate of the fuzion... :(:(:(
 
Your pump was definitely holding you back. You were likely getting around .7 GPM or less on your setup according to Martin's Flowrate estimator. You really want to keep it higher than 1 GPM for most setups, especially on the Storm. The other thing, as rightly pointed out, is that the Storm is not a good waterblock for quad cores. The cooling area on the block is not large enough for todays large chips, and therefore the Storm has been outpaced by the newer blocks on the market.

Just changing the CPU block should get you a lot closer to 1GPM. The other benefit here though is that the Fuzion is a much better performer at lower flowrates than the Storm is. You may still want to upgrade the pump though, because even with the change of waterblock your pump will be the limiting factor. It is up to you to decide if you care after seeing the new temperatures.
 
You must log in or register to reply here.
Back
Top