Built my first custom loop. Should I have expected these temperatures??

[Selth Afrinon]

Hello everyone.

 

I just finished putting together my first custom loop and I wanted to know if the temps I'm experiencing are what I should have expected compared to when I was air cooling (see setups at bottom of post).

 

I when I was planning out the custom loop, I was hoping to decrease my CPU and GPU temps by some noticeable amount (5 - 10 deg C would have been nice), but instead I have a 5-10 deg C increase to CPU temps and a 10-20 deg C increase to GPU temps at the same loads. Here's what I've tried to get better temps so far:

 

1) Removed the old TIM and applied a new coat for both CPU and GPU. Got the same temps.

 

2) Removed the retention bracket from the CPU to see if it was blocking any contact. Got got the same temps again.

 

3) Plugged water pump into separate power supply, then turned off supply to see if it affected temps (this was to see if the pump is functioning). Temps with pump off became significantly higher (CPU broke 100 deg C).

 

4) Shaking and rotating PC to dislodge air bubbles. Got some air bubbles out of radiator, but temps remained the same.

 

I tried everything I could think of short of buying a new pump and a new radiator, and wanted to see if I had overestimated the cooling ability of my Ek-CoolStream CE 280 radiator or if I'm missing something. Would getting a second pump to increase the pressure head help? Do I need more radiators?

 

Please let me know. I really want to water cool my PC, but I'm considering tearing everything out and switching back to air-cooling if I can't get my temps any lower.

 

----------------Setup----------------------

 

Water-cooling setup

Hardware:

CPU: i7 - 8700k (5.0 Ghz, 1.452V, Delidded w/ Thermal Grizzly Conductonaut for CPU die - IHS and IHS - water block)

GPU: ASUS ROG Strix 1070 Ti (Stock)

 

Loop

Pump: Swiftech MCP655

CPU block: Ek-Supremacy EVO

GPU block: Ek-FC1080 GTX Ti Strix

Radiator: Ek-CoolStream CE 280 (45mm thickness) x1

 

Cooling:

140 Fans: Noctua NF-A14 x2

 

Temps:

Aida64 Stress Test CPU Temp: 75-80 deg C

Max Utilization BOINC CPU Temp: 85-90 deg C

Heaven 4.0 GPU Temp: 57-60 deg C

 

Air-cooling setup:

Hardware:

CPU: i7 - 8700k (5.0 Ghz, 1.452V, Delidded w/ Thermal Grizzly Conductonaut for CPU die - IHS and IHS - CPU cooler)

GPU: ASUS ROG Strix 1070 Ti (2117 MHz GPU, 8017 MHz)

 

Cooling:

CPU cooler: Cryorig R1 Ultimate (Dual tower, 140 x2 fans)

GPU cooler: Stock ASUS ROG Strix cooler (x3 fans)

 

Temps:

Aida64 Stress Test CPU Temp: 70 - 73 deg C

Max Utilization BOINC CPU Temp: 80 - 82 deg C

Heaven 4.0 GPU Temp: 42 - 47 deg C

 

 

 

[W-L]

5 minutes ago, Selth Afrinon said:

-SNIP-

Can you take a photo of your loop that can provide a good bit of information, as for why the temps are so high your rad space is a bit small for a heavily overclocked CPU and GPU. Usually for a single CPU and GPU setup it would be recommended to have at least a 360mm rad but it still shouldn't see such a jump in temps. I would check your pump and see what setting it's on for the D5 if it is the vario edition. 

[Snipergod87]

I think part of the issue is a single 280mm rad for the CPU and GPU as opposed to air cooled where they are isolated, a larger rad or more rads would make a difference.

[Selth Afrinon]

33 minutes ago, W-L said:

-snip-

The D5 is set to 5 for all the tests. For the picture, the res/pump combo is on the right. Intake and output run behind the mounting plate: the bottom-left run is pump output and runs into the GPU, CPU, then top-mounted radiator, then behind the mounting plate again.

 

I had originally planned to mount a 140 radiator on the back, but I didn't have the budget for it this month. Should i be expecting temps 3-5 deg C higher than my air cooled setup with my current water cooled setup?

 

6 minutes ago, Snipergod87 said:

I think part of the issue is a single 280mm rad for the CPU and GPU as opposed to air cooled where they are isolated, a larger rad or more rads would make a difference.

Do you think a single 140 would make a lot of difference? Or would I need more to get an appreciable temp drop?

[W-L]

2 minutes ago, Selth Afrinon said:

The D5 is set to 5 for all the tests. For the picture, the res/pump combo is on the right. Intake and output run behind the mounting plate: the bottom-left run is pump output and runs into the GPU, CPU, then top-mounted radiator, then behind the mounting plate again.

 

I had originally planned to mount a 140 radiator on the back, but I didn't have the budget for it this month. Should i be expecting temps 3-5 deg C higher than my air cooled setup with my current water cooled setup?

It's hard to say exactly but from my experience it shouldn't be that high especially on the GPU as you will usually see a substantial drop. One thing I would recommend to check is to see if any of your tubes are pinched if you running in behind the motherboard tray. The only other thing I can pin point would be a lack of intake due to the large number of hard drive and cages you have in the front, are you running everything with the side panel off? 

 

Also what case is that, Corsair 750D? 

[Selth Afrinon]

16 minutes ago, W-L said:

It's hard to say exactly but from my experience it shouldn't be that high especially on the GPU as you will usually see a substantial drop. One thing I would recommend to check is to see if any of your tubes are pinched if you running in behind the motherboard tray. The only other thing I can pin point would be a lack of intake due to the large number of hard drive and cages you have in the front, are you running everything with the side panel off? 

 

Also what case is that, Corsair 750D? 

The case is a Corsair 780T. I looked over the loop, but I didn't see any pinching. All tests were run with the side panels off at the time; I reran Aida and max BOINC with the side panels on just now and the temps remain around the same. I'm considering either adding a 140 radiator on the rear exhaust or a second D5 pump in serial (or both). Do you think these changes would make a significant difference?

 

[W-L]

5 minutes ago, Selth Afrinon said:

The case is a Corsair 780T. I looked over the loop, but I didn't see any pinching. All tests were run with the side panels off at the time; I reran Aida and max BOINC with the side panels on just now and the temps remain around the same. I'm considering either adding a 140 radiator on the rear exhaust or a second D5 pump in serial (or both). Do you think these changes would make a significant difference?

 

Try running everything at stock speeds first to see if it's just the overclock that is pushing the cooling capacity of the loop. I believe it might be the radiator space to a certain degree coupled with the high overclock. Also top up your res if it's a pump combo unit from XSPC you need to make sure it's pretty full since it cannot draw water if it's below ~40% or so due to the inlet port. 

[Selth Afrinon]

23 minutes ago, W-L said:

Try running everything at stock speeds first to see if it's just the overclock that is pushing the cooling capacity of the loop. I believe it might be the radiator space to a certain degree coupled with the high overclock. Also top up your res if it's a pump combo unit from XSPC you need to make sure it's pretty full since it cannot draw water if it's below ~40% or so due to the inlet port. 

Ok, made sure res is topped off and reran tests at stock CPU and GPU settings (GPU was at stock settings when performing the water cooling tests from original post):

 

Stock:
CPU Temp Idle: 33 - 40 deg C (4.3 GHz stock turbo, 1.296 V)
CPU Temp AIDA64: 53 - 56 deg C
CPU Temp Max BOINC: 60 - 63 deg C
GPU Temp Heaven 4.0: 58 - 59 deg C

[W-L]

4 minutes ago, Selth Afrinon said:

Ok, made sure res is topped off and reran tests at stock CPU and GPU settings (GPU was at stock settings when performing the water cooling tests from original post):

 

Stock:
CPU Temp Idle: 33 - 40 deg C (4.3 GHz stock turbo, 1.296 V)
CPU Temp AIDA64: 53 - 56 deg C
CPU Temp Max BOINC: 60 - 63 deg C
GPU Temp Heaven 4.0: 58 - 59 deg C

Yeah those looks a lot more reasonable, your dumping a lot of voltage through that 8700K which would explain the high CPU temps. Usually the recommendation is to not go past 1.4V as a good cutoff point. The GPU is a tad warm but that might be due to the raditator size, try increasing your fan speed to see if it makes a difference at all, if there is a big drop in temps I'm pretty certain you need more radiator space more than anything else. 

[Daymin]

If you're pushing that much voltage at the cpu a 280mm Rad isn't enough to cool cpu and gpu.

Adding a 140mm radiator will help but how about another 280/360 in the front?

You don't need to another D5 pump in the loop that won't make a lot of difference,as you radiator size is the limiting factor nor your flow rate or head pressure 

[Selth Afrinon]

9 hours ago, W-L said:

Yeah those looks a lot more reasonable, your dumping a lot of voltage through that 8700K which would explain the high CPU temps. Usually the recommendation is to not go past 1.4V as a good cutoff point. The GPU is a tad warm but that might be due to the raditator size, try increasing your fan speed to see if it makes a difference at all, if there is a big drop in temps I'm pretty certain you need more radiator space more than anything else. 

So with the same GPU clock speeds, Heaven 4.0 temps look like this:

 

Lowest fan setting: 55 - 59 deg C

Highest fan setting: 50 - 53 deg C

 

So there's a good difference there. I guess I'll just run everything at stock until I can add a new radiator to the loop.

5 hours ago, Daymin said:

If you're pushing that much voltage at the cpu a 280mm Rad isn't enough to cool cpu and gpu.

Adding a 140mm radiator will help but how about another 280/360 in the front?

You don't need to another D5 pump in the loop that won't make a lot of difference,as you radiator size is the limiting factor nor your flow rate or head pressure 

Cool, I'm glad the issue isn't with the pump. I originally considered front mounting a 360 radiator, but I need the 3.5 cages in the front to act as my primary media storage. The most radiator space I think I can get into the case is 280 or 360 top, 140 back, and 120 bottom. With a few changes to layout, I could fit a  240 on the bottom.

 

I think I'm going to get a 140 for the back first and see if causes a significant enough drop; I'll go straight to a 240 bottom if it doesn't. Do you think I'll need another pump with the 240 bottom, though?

[W-L]

2 hours ago, Selth Afrinon said:

-SNIP-

Yeah with such little difference in temps it's most likely due to the lack of radiator space, if you can squeeze in a 240mm rad that would be a decent amount of added cooling potential. As for the D5 it can easily handle all those components and more. 

[Selth Afrinon]

46 minutes ago, W-L said:

Yeah with such little difference in temps it's most likely due to the lack of radiator space, if you can squeeze in a 240mm rad that would be a decent amount of added cooling potential. As for the D5 it can easily handle all those components and more. 

Sounds good. That'll go into the budget for next month. I'm just glad that it isn't a problem with defective components.

 

Thanks for hearing me out.

[QBtech]

Have to back everyone up here: the issue is definitely not enough radiator space. The rule of thumb is one 120 per component + 1 extra for good measure for reasonable cooling at stock. For over clocking that should be doubled if not tripled. Honestly the rule of thumb usually provides similar cooling to middle of the road aircoolers.