Overclock.net banner

1 - 20 of 82 Posts

·
Registered
Joined
·
22 Posts
Discussion Starter #1
Hello all,

So I'm rebuilding my rig. Specs:
Asus Strix Z270G, Core i7-7700K 4.2 GhZ, 2 x GeForce GTX1080 TI (FE), Corsair Vengeance LPX 4x8 Gb

PROBLEM:
CPU temps are extremely high. In idle, 50-60 C and in gaming 100+. GPU's are fine, max 60 in 4k gaming on ultra settings.
And it's strange, because the rad stays ice cold, while the loop is freaking hot, specially at the monoblock with cpu.


Additional info:
All water cooling parts are from EKWB. The pump is an EK-XTOP Revo D5 PWM and the rad is a coolstream 240 with 4 x EK-Vardar F4-120ER (2200rpm) fans in push/pull configuration.

The loop goes like this: res -> pump -> gpu2 -> gpu1 -> ram -> cpu -> rad -> res
And while this may not be the optimal loop, I did it this way due the nature of this rig (see photos), and I did expect a temperature difference compared to my last loop with 2 res and 2 rads and cpu first in loop, but not that much.

So what seems to be the problem?


I have thought about this:
- Loop is wrong, cpu should always go first. But will that really make such a big difference?
- Rad is not working, and water is not flowing all the way in it before being sent back in the loop.
- Pump not strong enough (very unlikely). In pwm set to 100% always.
- Too much air trapped in rad, but how on earth do I find out and drain that air?
- Back connection (cpu -> rad) is too long (see photos).
- Cooling paste between cpu and block has gone too bad/ineffective.
- Res not tilted enough to provide enough pressure on pump (it's currently 5 degrees), but I tried having the pump directly under it, and no change.


Any suggestion is appreciated. Thanks everyone.
 

Attachments

·
Registered
Joined
·
92 Posts
Bad contact between CPU and water block.
 

·
Overclocker in training
Joined
·
12,954 Posts
Hi,
Your flow though the gpu is backwards you're going into the outlet
Yes it matters flow wise because the clash at the jet plate will slow flow down a lot.
Best to switch that around so flow goes into one point and splits into two at the jet plate.

Mono blocks are a pita so yeah likely also contact issues too.
 

Attachments

·
Shooting down fallacies
Joined
·
8,077 Posts
Yeah, definitely sounds like contact issues as said.
@ThrashZone - EK swears some of their blocks can be run either way, and some they changed the manuals to say they can't. Regardless of EK's manuals, I agree with you that the flow should be in the other direction. There is no way that block can perform as well flowing that way.
 

·
Overclocker in training
Joined
·
12,954 Posts
Hi,
Yeah that clash is x2 on sli.
 

·
Registered
Joined
·
21 Posts
You have seem to have a couple problems. The easiest one is most-likely what a couple of people have said, block not having good contact with CPU. Another mentioned the backward flow. The other issue is your radiator. You have both connections at the bottom. I am not sure your coolant is running completely through the radiator. I always have my inlets at the top when I stand a radiator vertically. It may, but I would prefer to let gravity help. Here is your posted pick that shows the connections to the radiator.
 

·
Registered
Joined
·
108 Posts
I would also think poor CPU contact.

I've not experienced any issues running a single GPU in reverse flow, with temperatures or flow loss - but other things changed at the same time, ymmv.

Grey beard makes a great point. It is going to be difficult to bleed the air from your radiator in that position. That could cause some cavitation in the radiator.
 

·
Registered
Joined
·
1,161 Posts
Ashcroft

The GPU loop order is fine if that's how you want to have it. There is no such thing as 'clashing' fluid flow. Restriction is the same in either direction. Extra restriction can't be created out of thin air.
It might result in slightly less than optimal heat removal from the GPU cores but it will be very minor (1 or 2C) compared to the massive temp drop VS air cooling them, as you have seen with the temps you have with two GPU's on a 240 rad.

Monoblocks can be notoriously hard to get good contact with the CPU, and the symptoms you describe are typical for that. Remounting it is the only real thing you can try in that case.

If you start the system after its been cold for a while are the cpu temps instantly high or does it take a few minutes to climb to those high temps?
If its instantly high with a cold loop its definitely a contact problem. If the CPU temp starts off low and climbs to unreasonable levels then there is something else going on with the loop. Opaque coolants can cause clogging problems.

The Radiator should be fine like it is. The pump should be producing enough flow to force air out even with the ports at the bottom. Its never caused me any problems. Its the fluid in the rad tubes that receives the cooling effect and they are so narrow that bubbles zoom through them far far faster than they can float.
The outside of the rad feels cold because its an aluminium casing that is insulated from the copper core section.
 

·
Registered
Joined
·
1,524 Posts
When did you notice the issue? Did it occur all of the sudden or after you changed something?

Can you take a good picture of the fins on your monoblock?
 

·
Overclocker in training
Joined
·
12,954 Posts
Hi,
If you remove the ek stainless steel jet plate it would be less of a problem
Leave it in and two directions of fluid have to fight to go through that small slot x2 being sli

Mono block is last in line for fluid so it will be hot no matter what.

Wall on the rad is also on the exit end so air flow is corked too looks like
 

·
Registered
Joined
·
1,161 Posts
Even if clashing flow was a real phenomenon, splitting the flow to two cards could not increase the effect, only reduce it. Restriction is quartered when flow rate is halved. So splitting the same flow rate between two blocks results in half the total restriction of a single block.

But its not real anyway.
https://www.ekwb.com/blog/can-i-use-the-inlet-port-as-the-outlet-and-the-outlet-port-as-inlet/

Don't remove the jet plate, it will only reduce the performance you get.
 

·
Overclocker in training
Joined
·
12,954 Posts
Hi,
Yeah okay go by ek's words maybe refer to page one post 4 :thumb:
At some point physics take over whether there is a little or a lot of difference there is a difference the jet plate actually degrades nickle plating
Image sucks but this is a ek nickle block and clearly the jet plate degraded the nickle surface where it sits.

 

·
Registered
Joined
·
1,161 Posts
'Its Physics', with no actual explanation is the cry of people who don't actually understand physics.

So EK would deliberately mislead their customers and knowingly falsify test results just to tell people they can use their blocks in reverse for worse results ... for some reason.

Its not like anyone could just test it for themselves I guess, its not like heaps of people have flow meters and the ability to do a simple test.

Why not take that risk, right, makes total sense. After all, they gain a lot by having people use their blocks like that.

Also, discolouration or dulling does not equal degradation.
Stainless jet plates have been used for years by all the major manufacturers. If it was a problem it would have been noticed long ago.
 

Attachments

·
Registered
Joined
·
22 Posts
Discussion Starter #14
Hey guys,

My apologies for the late follow up, I've been swamped with work.

So here's what happened: I did like some of you said, made new tubes and changed the loop to: res-pump-cpu-ram-gpu1-gpu2-rad-res-pump
... and MAAAN, what a difference. No doubt that the flow got much faster (or bigger? what's the word?), just by eye observation, and the system was able to take twice the amount of coolant as I used in the first configuration, so I'm convinced that rad is 100% full now.

As for temps:
GPU temps dropped down to around 25-30 C in idle, and maximum 45 C in 4k gaming. And yes, that's with a 240 rad, so to you people who thinks that's not enough, I'm just like "huh?". So clearly a big difference.

AS FOR CPU (THE ONLY REMAINING PROBLEM):
So I did replace the paste before re-doing the loop. I use the Thermal Grizzly Hydronaut. I made an "X" pattern and patched it all up again. Difference: 35-40 C in idle and maximum 75 C in gaming. Clearly a difference, but still extremely bad. I used to run 30 C in idle and max 45 in gaming in the previous build. Monoblock feels cold when I see 75 C, so clearly it's a contact issue. Temps skyrocket quickly with the smallest tasks, such as opening a program or a large file. And I don't dare gaming.

What should I do now?
Re-apply the paste again? It's a lot of work to do :mad:
Replace the monoblock?
Or just leave it the way it is and ignore the high cpu temps?
 

·
Overclocker in training
Joined
·
12,954 Posts
Hi,
Post another image or two where you're at now.
 

·
Rig Advisor
Joined
·
1,965 Posts
The general consensus on this forum is that Monoblocks usually aren't the best route as they do not cool as well as CPU only blocks. That said, I think your issue sounds more like a contact pressure issue rather than a thermal paste application failure, are you sure there is enough pressure from your block to your IHS?. Glad your temps on the GPU's normalized, but the "Huh" on a 240 being able to cool your setup made me say "Huh". 🙂

Edit. Side note. That block your rad is mounted to has blocked airflow. I would cut a huge hole in the middle so air can flow freely.
 

·
Registered
Joined
·
22 Posts
Discussion Starter #18
The general consensus on this forum is that Monoblocks usually aren't the best route as they do not cool as well as CPU only blocks. That said, I think your issue sounds more like a contact pressure issue rather than a thermal paste application failure, are you sure there is enough pressure from your block to your IHS?. Glad your temps on the GPU's normalized, but the "Huh" on a 240 being able to cool your setup made me say "Huh". 🙂

Edit. Side note. That block your rad is mounted to has blocked airflow. I would cut a huge hole in the middle so air can flow freely.
What do you mean by "are you sure there is enough pressure from your block to your IHS?"? You mean if I screwed the monoblock very well on the mobo? If yes, well, yeah, I screwed it very tightly.

And you're right about the hole to the rad block, working on it.
 

·
Rig Advisor
Joined
·
1,965 Posts
Don't know anything about the standoffs your monoblock came with. Could be you used some from the wrong platform if 2 sets were included?
 

·
Overclocker in training
Joined
·
12,954 Posts
Hi,
Looks cool about what mono blocks are all about :)
How many screws does that mono block have 4 or 6 ?
 
1 - 20 of 82 Posts
Top