[SOLVED] Water Cooling - Lowering Temps - Extra Rad?

Liudom

Distinguished
Dec 9, 2012
10
1
18,510
Hey, I'm a professional lurker at this point, and thought I'd start getting more involved :)

I have:
EK-Velocity Block for 3700x
EK's MSI Trio Block for a 2080
1x 360 rad (30mm wide, Corsair XR5 + 3 Noctua 1700rpm Fans) with my CPU and GPU in one loop.
Pump/Res is a EK-Quantum Kinetic TBE 200 D5 PWM
I have went for Hard Tubing using Barrows fittings throughout.

My setup is Pump/Res > GPU > CPU > RAD > Pump/Res

I have made a few changes in the bios like offsetting cpu volts by 0.1 & disabled PBO. The changes from guides I'd seen from loads of places to prevent spikes. Not seen any performance dip and is still packing a punch when I do PS and AE stuff.
I do plan to reverse all of this and OC.

CPU idle temps is: 32 - 36
CPU Stress test is: 44-50

GPU idle temps is: 29 - 32
GPU Stress test is: 46-50

Ambient temps is only like 18.
Pump is on a curve, for noise purposes, I try to keep it under 70% speed. Fans are set to be quiet and never goes over 1000rpm per fan.

I know these are decent temps, but I always had the impression it can be better. I mean the money you pay, you'd expect to get those results else I would have just stayed with AIO on both CPU and GPU, which gave better results tbh.

My question is, would another rad be helpful at all? Is it worth putting in a extra inbetween CPU and GPU?

I'm using a Thermaltake View 51 case, the air flow is.
In from the Top and Front
Out from the Back and Side.
I have a feeling I should just switch the front to out and back as in, so its all flowing in one direction. Would this even help the temps?
 
Last edited:
Solution
Here's the thing about temps. If they are under throttle limits, they don't matter. A cpu at 30° runs the exact same as a cpu at 40° or 50° or even 60°. To the cpu its all the same.

Liquid loops are not so much about lowering temps as they are about controlling temps. And yours have plenty of control. You'd be looking at @ 100w for the cpu, 250w for gpu at absolute maximum loads simultaneously (@350w), on a rad that's somewhere around the 400w point or better. Gaming you'd be closer to @ 70% of that 350w.

Adding a rad will mean 2 things. First, you'll not slow down the pump. The added resistance and added volume adding weight to the flow will require a little more 'umph' than that single rad.
But, you would be able to turn the fans...

Phaaze88

Titan
Ambassador
I'm using a Thermaltake View 51 case, the air flow is.
In from the Top and Front
Out from the Back and Side.
I have a feeling I should just switch the front to out and back as in, so its all flowing in one direction. Would this even help the temps?
According to that^, no.
It says that the View 51 has so many open spaces that it isn't far off from being an open chassis, so you can throw the positive/negative pressure argument out the window(pun).
Changing the fan orientation won't change much.

Ambient temps is only like 18.
I assume that's room ambient, and not chassis ambient.
The chassis ambient temp is what you should measure. It's always warmer than the room ambient.
That's going to be your baseline cpu and gpu temperature - the liquid temp will also follow this rule.
So the baseline looks to be 29C? Does the liquid temp match this?

My question is, would another rad be helpful at all? Is it worth putting in a extra in between CPU and GPU?
I don't believe it would for the cost to do that, but I'm no liquid cooling expert...
A)For the best cpu and gpu thermals, they really should be on separate loops.
There's a limit to how low to expect thermals with a 285w gpu with a 140w(?) cpu on the same loop.

B)Nothing beats direct die cooling. NOTHING.
The gpu? That's direct die cooling.
The cpu? Nope. The IHS is the greatest obstacle to cpu cooling. Too bad cpu DDC is so niche; it's available/compatible for very few models.
This alone makes custom gpu cooling far more worth it; the gains from cpu cooling are less impressive while the IHS is on.


All my opinions said, those are darn good thermals.(y)
To get even lower than that, though? Uhh... I think that's going to require some expensive changes that may not be worth it when it's all said and done...
 
  • Like
Reactions: Liudom

rubix_1011

Contributing Writer
Moderator
Honestly, your temps are pretty good at load. What CPU? (may have missed it)

But yes, agree - GPU is direct-die and CPU uses IHS which means another 2 materials to exchange heat through - core/IHS paste and then the IHS itself ....then the usual thermal paste and cooler/block.

D5 is OK to run at 100% if you wanted, the extra flow rate is beneficial for delta-T of your coolant. Flow rate is always good unless you feel there is an issue with noise, which if you do with nearly any pump, there is something wrong....pumps should be very quiet compared to most PC components, like fans.
 
  • Like
Reactions: Liudom

Karadjgne

Titan
Ambassador
Here's the thing about temps. If they are under throttle limits, they don't matter. A cpu at 30° runs the exact same as a cpu at 40° or 50° or even 60°. To the cpu its all the same.

Liquid loops are not so much about lowering temps as they are about controlling temps. And yours have plenty of control. You'd be looking at @ 100w for the cpu, 250w for gpu at absolute maximum loads simultaneously (@350w), on a rad that's somewhere around the 400w point or better. Gaming you'd be closer to @ 70% of that 350w.

Adding a rad will mean 2 things. First, you'll not slow down the pump. The added resistance and added volume adding weight to the flow will require a little more 'umph' than that single rad.
But, you would be able to turn the fans down as the excessive surface area would be closer to being a passive system, not requiring much airflow per rad to dissipate that small amount of wattage.

Whether or not you'll really move the temp gauge is another story. You already do decent amount of absorption as is, so much of what you see from the cpu is actual cpu temp created by the workload, there will always be a raise in temp.

You could add another 360mm and not really see any change at all, or add a 120mm and drop 5°, there's no real telling and it'd make no difference to anything but your perception.

You are looking at max idle of 36° and max cpu of 50°. That's a measly 14° difference between 5% load and 100% load. Honestly, that's ****ing incredible. You'll see many coolers go from 30ish to 70's to 80's easily, 40° gains.

Me personally, I'd not even mess with it. It ain't broke, it works better than ok. Only thing broke is your perception of what should/could be.
 
  • Like
Reactions: Liudom
Solution

Liudom

Distinguished
Dec 9, 2012
10
1
18,510
Love these responses, thank you so much. This is an eye opener. Its defo just me being picky.

Its just bad practice in a way to watch loads of vids and see peoples temps being so far down that made me think twice about my own loop.
The 3 responses made 100% sense to me and have so much cooledpun? my urge to try reach even lower temps.
 

Karadjgne

Titan
Ambassador
Ryzen aren't Intel. Don't work the same or behave the same. With Intel, at idle all cores are dropped to minimal voltages and loads, but all cores remain active. This means background tasks are split up amongst all the cores, each seeing only a few services. Temps are read from the hottest core at any given time, so you'll see small bumps in temps as one core starts a service, another stops a service etc.

With Ryzen, it's similar, but different. All cores are shut down, inactive, except 1. That core takes the full load of every background process, concentrated, so you'll see by default higher idle temps. With that as the only active core, it also sees every service and process start, so max idle spikes are comparatively higher.

Freaks some people out, because they compare a single, low load Intel core vs the much higher load Ryzen core but don't realize it's the split load of Intels vs single load of Ryzen.