Custom waterloop- IvyBridge - Temperatures pretty high

[Denoxx]

Hey guys

 

I've ran my watercooling setup for a while now, and I've recently started overclocking it again (I ran it at 4.4Ghz for a while) and I just had to realize, that my CPU temps are pretty high (Unfortunately IvyBridge just runs really hot)

 

I run a 3770K together with 2x GTX780 in SLI.

 

Watercooling hardware:

CPU-block: Swiftech Apogee HD

Pump: Alphacool VPP655 

 

Im running the pump with 3700RPM (I read stuff about laminar flow, but I tested on all 5 levels that the pump can run at without seeing a difference bigger than 1-2 degrees)

 

The somehow sad thing is that I achieve the same OC like on air, which is not really what I expected from a watercooling setup like that

( 2x 480 rads ).

 

Now is there anything that could be the cause of my temperatures besides my CPU model which is just running hot by default?

I really hope some of you might have an idea

 

Thats how my loop looks right now:

 

 

I stresstest with Aida64 maximizing load on CPU and GPUs to achieve the highest possible temp that I could

The temps look a bit lower than they will be after I ran the stresstest for 30 minutes, they circle around 90 degrees then

[Majestic]

I see no GPU temperatures, and those CPU temps are really bad even for Ivy. Is your pump even running?

[ind3x]

A delid is probably needed, but make sure that the CPU block is firmly mounted.

[Denoxx]

I see no GPU temperatures, and those CPU temps are really bad even for Ivy. Is your pump even running?

Yeah sorry theyre around 48-50 degrees. Yes its running on lvl 3 of 5 possible speed-levels (3700RPM)

 

 

 

A delid is probably needed, but make sure that the CPU block is firmly mounted.

I dont think I'd want to take the risk of delidding it  :wacko:  I  took it off and mounted it again without  the temps changing really

[C0LL0SS0S]

are you using thermal paste between the cpu cooler ?

[Denoxx]

are you using thermal paste between the cpu cooler ?

 

IC-Diamond as learned by our godfather Linus 

[C0LL0SS0S]

IC-Diamond as learned by our godfather Linus 

thats strange, any way to check your water temperature ?

[Denoxx]

thats strange, any way to check your water temperature ?

 

Phu not really I guess, theres no temperature-reading hardware inside my loop

[C0LL0SS0S]

Phu not really I guess, theres no temperature-reading hardware inside my loop

do you have a thermometer to stick in?

[Denoxx]

do you have a thermometer to stick in?

Would that reading help anything? 

[Majestic]

thats strange, any way to check your water temperature ?

 

Since the GPU's are 50ish, the water can't be much hotter. Meaning there is a 30+ degree delta between the loop and the chip. 

Which can only be explained by;

 

-Pump not running, but you verified + the GPU's aren't overheating

-CPU Block not mounted properly, excessive use of thermal paste or not evenly tightened.

-Poor TIM application between LID and DIE

-excessive voltage.

 

Tbh, if reseating the CPU block doesn't fix it, you're looking at a delid. Try adding some o-rings like with AIO coolers.

[C0LL0SS0S]

Would that reading help anything? 

yes we would erase one potential error, have you maybe applied too much Thermal compound ?

[Denoxx]

Since the GPU's are 50ish, the water can't be much hotter. Meaning there is a 30+ degree delta between the loop and the chip. 

Which can only be explained by;

 

-Pump not running, but you verified + the GPU's aren't overheating

-CPU Block not mounted properly, excessive use of thermal paste or not evenly tightened.

-Poor TIM application between LID and DIE

-excessive voltage.

 

Tbh, if reseating the CPU block doesn't fix it, you're looking at a delid. Try adding some o-rings like with AIO coolers.

 

I tended to apply too much thats why I checked last time when I reseated the CPU block. I could swear its not too much, but I used the "dot in the middle and let the pressure spread it"-method could that be a problem? I think thats how "its done" currently

 

Excessive voltage not so much you can see on the CPU-Z screenshot that its running at 1.235V

[Majestic]

I tended to apply too much thats why I checked last time when I reseated the CPU block. I could swear its not too much, but I used the "dot in the middle and let the pressure spread it"-method could that be a problem? I think thats how "its done" currently

 

Excessive voltage not so much you can see on the CPU-Z screenshot that its running at 1.235V

 

Yeah, was just mentioning all the possibilities. Which means it leaves the CPU block not able to apply enough pressure (fixable with orings) or a Delid is required.

[Denoxx]

I just tried to tighten the screws on all 4 sides of the CPU and they are like "screwdriver-tightened" without trying to kill the CPU

[Denoxx]

Yeah, was just mentioning all the possibilities. Which means it leaves the CPU block not able to apply enough pressure (fixable with orings) or a Delid is required.

I wouldn't delid my CPU even though I know about the massive temp decrease.. It just looks so risky

[Denoxx]

Could it be anything else? Like I dont know, pump-speeds or fan-speeds or whatever? 

I dont think theres that much air left in my loop, but when i put the pump to max rpm, theres a little sound from air i guess

[Majestic]

If there was something wrong with the loop, your GPU's should also be affected. Moreso than the CPU because they use way more power.

[Denoxx]

Damn   Well the temps are within the acceptable margin so I just have to deal with them...

What a pity

[Majestic]

I wouldn't call those temperatures within acceptable margin...

You can still try some rubber o-rings on the backplate. It might help.

 

Otherwise, either delid or drop the voltage a tad. If that requires dropping frequency..so be it.

[Denoxx]

I wouldn't call those temperatures within acceptable margin...

You can still try some rubber o-rings on the backplate. It might help.

 

Otherwise, either delid or drop the voltage a tad. If that requires dropping frequency..so be it.

 

Whats this thing with the O-rings? will this just increase the pressure on the CPU?

[Majestic]

Whats this thing with the O-rings? will this just increase the pressure on the CPU?

 

You might think you're already applying allot of pressure, but it could just be on the mounting mechanism and not the CPU.

Those o-rings might help actually applying some pressure on the CPU.

 

Just can't really imagine a custom loop peforming this bad on an Ivy, even with it's hotter than usual temps. 

[Denoxx]

You might think you're already applying allot of pressure, but it could just be on the mounting mechanism and not the CPU.

Those o-rings might help actually applying some pressure on the CPU.

 

Just can't really imagine a custom loop peforming this bad on an Ivy, even with it's hotter than usual temps. 

Is there a risk of applying too much pressure? 

[Majestic]

Is there a risk of applying too much pressure? 

 

 

Well yes, but you have to force it regardless. o-rings won't change that. You'll know when you're applying too much force.

Also, are you cycling screws when tightening them? Don't completely tighten them one at a time, gradually cycle through the screws until they're secured.

[Sandy_Dandy]

From a thermal dynamic standpoint it sure sounds like bad conductance between the die and the water, and if the TIM is at fault, nothing will solve your problems, beside a delid. It's a risk yes. But it's also the only thing that make or break Ivy OC. 

 

Although, if delid is completely of the table, I would sell the Ivy and get a Sandy 2600 or 2700k instead. Those chips are proper maniacs when it comes to overclocking, especially under (real)water cooling. Slightly worse IPC, but if you can't get your Ivy above 4.4GHz it won't matter, a 5GHz+ Sandy will still be a much better choice.