• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

RATIONAL custom loop for 14900K and 4080S

So, can you remember the difference in temps that was so alarming?
Instead of 95-96c in Prime95 it was 100c and throttling. I needed it to be around 320w stable for the test and it was dropping down to 260w because of the throttling. Not very helpful info. Just that the wrong direction can have a impact a higher wattages.
 
Well thank you, but I do not see the coolant temperature anywhere?

Don't think it can be monitored by msi AB and tbh I've never really worried about coolant temps as 99% of the time my CPU & GPU temps are fine so I find coolant temp negligible but each to their own loop ;)
 
It is alive!

The old setup with 4070:

View attachment 333661

The wet GPU upgrade:

View attachment 333662

View attachment 333663

I let it run for a while with 250W GPU and 70W CPU load and the intake of the rad felt like 40°C, that is 20°C over ambient., that is probably not good.

Those are 280 rads, not 240. I think I probably should add another fans to the top rad and exchange fractal case fans from the front rad for P14 fans too.

It seems it will not be silent.... :(

Hi!
From the looks of it, you have just enough or just barely enough radiator.
I'm not a pro, but basic numbers like 100w per 120.1 worth of rad max rule of thumb. You're covered essentially 400w with average case fans.

There could be ways to optimize the cooling, a few degrees may sometimes be better than nothing. Things like stripping paint from rads running bare metal, to spacers between the fans and rads to reduce the dead spot in the center. Trying to maximize the surface area. In a pinch deal, maybe something to try ect. I have a single delta on a 360 rad. It cools 14700K nicely. Cause it pushes many CFM at high pressure. Maybe some fan upgrades. Or more rads to be quiet bout it.

Really, warm rads means the block is doing good heating the water. You just want to see the outlet temp lower if you can. So maybe with not quite enough radiator, the path might matter. I do pump to cpu to gpu to radiators. Inlet rad bung should be warmer than the outlet. You know if you're doing this by feel anyways.

I tape the probe from the multimeter where I want to measure temps. Does your multimeter do temp probing? This might help.

Damn water cooling thread got you guys going liquid. Something I've observed, pretty cool.
 
100w per 120.1 worth of rad max rule of thumb. You're covered essentially 400w with average case fans.
I don't know if these estimates ever meant anything, CPU/GPUs have different cooling needs today vs 10 years ago even if they use the same amount of power.

I think a better rule of thumb is too simply fit as many rads as you can reasonably have, it really doesn't make sense trying to get stingy at amount of radiators, the thing is going to cost a ton anyway.
 
I noticed something interesting about the CPU block: it is a new revision now, it has different internals. The old one fired water in the middle of the base slots. Now the new version has a different water jet assembly in it, which splits incoming water and fires it from both sides of the slots and it exits in the middle.
Hi,
Intake should shoot into the center chip and out both sides of the cooling fins and flow to outlet port so no conflicts besides hitting the cold plate in the center.
Reverse does the opposite and has two stream of flow hitting each other in the middle to get to the outlet.

Hard to tell but the block you have now doesn't look like it cares flow goes from one end and out the other
1707400359171.png
 
Hi,
Intake should shoot into the center chip and out both sides of the cooling fins and flow to outlet port so no conflicts besides hitting the cold plate in the center.
Reverse does the opposite and has two stream of flow hitting each other in the middle to get to the outlet.
That is exactly what they did.
 
That is exactly what they did.
That is what the assembler did by swapping inlet ports.
Case and point what happens if the inlet were the opposite ?
 
They have now two inlet slots, I quickly drew old and new version:

new block revisin.png
And yes, the more distant slot from the inlet is larger. I could not see exactly what is happening at the ends of the slots, I did not want to disasseble the block, because is uses just self tapping screws in the plastic.

I am talking about Corsair XC7 block. Now I am not sure if the traditional construction on the left is not an entirelly different product, instead of a previous revision of the XC7, as I thought!

You can see the outline of the inlet manifold in the picture:

1707400359171.png

I just noticed that a review of this block is on the site, it has detailed photos of the internals:

 
Last edited:
Weird looks like a void in the middle on the right image
Left is traditional design.
 
There have to be, no matter how fast the flow is, that's how thermodynamics work.
Since you couldn't be bothered to quote my entire statement, I will repeat: Yes, there has to be a temperature differential, simple physics.

However, if that differential is large enough, compared to the operating temperatures of the loop, that the order of components impacts the performance of your loop, then there is something wrong with your loop.

"Practical" or not, it is not worth caring about. If you ever were in a situation where it was "worth caring about", then fix the correct thing (flow rate) rather than getting bent out of shape about negligibles (loop order). Simple engineering.
 
Yes, there has to be a temperature differential, simple physics.
You admit the logic and physics are sound yet you refuse to accept that it does matter, bizarre.
"Practical" or not, it is not worth caring about. If you ever were in a situation where it was "worth caring about"
There is no reason to not put the radiators before the component that's going to dump the most heat unless it's a hassle to do so, that way you have the largest drop in coolant temperature possible before the component that needs it the most. That's actually the simple engineering here, it's the logical thing to do.
compared to the operating temperatures of the loop,
There is no one operating temperature of the loop, it's going to be a gradient in-between places where heat is dumped or put into the system.
 
Last edited:
They have now two inlet slots, I quickly drew old and new version:

View attachment 333736
And yes, the more distant slot from the inlet is larger. I could not see exactly what is happening at the ends of the slots, I did not want to disasseble the block, because is uses just self tapping screws in the plastic.

I am talking about Corsair XC7 block. Now I am not sure if the traditional construction on the left is not an entirelly different product, instead of a previous revision of the XC7, as I thought!

You can see the outline of the inlet manifold in the picture:

View attachment 333737

I just noticed that a review of this block is on the site, it has detailed photos of the internals:

Hi,
Yeah thanks for the review link I looked briefly and didn't find much

I'll stick with weird lol
Review confirms restrictive with this weird cold plate plate flow conflicts going on not sure if the second three slot channel would be better or worse :fear:
1707405187248.png

Edit after thinking about it the three slot would likely be better.
 
Last edited:
This is a loop build much like my own.

No vanity nonsense, no hard tubing, no pointless bling - just a D5 pump and as much radiator as I can fit in the loop as possible without adding inconvenience, cost, or complexity.

I used to run an NZXT H440 with 140 slim and 280 thick radiators in the back and roof, respectively - I did this rather than running a larger radiator on the intake because I felt that having a radiator dump heat into the case made the other radiators less effective.

When I actually got around to testing it with a spare 240 radiator in the front, temps did go up slightly and flow rates went down slightly. I haven't really done enough testing to give meaningful answers but I suspect that the priorities should be, in order;
  1. Get enough radiator surface area to cool your heat output at a fan speed you're satisfied with.
  2. Get those radiators cool air that hasn't been spat out of another hot radiator already.
Effectively, if you can cool your system adequately with just a 360mm intake radiator, or just a rear and top radiator, then you don't need to mix and match intake/exhaust radiators because the radiators dealing with hot air aren't really adding anything of value other than increased complexity, cost, and flow restriction.
 
... simply fit as many rads as you can reasonably have, it really doesn't make sense trying to get stingy at amount of radiators, the thing is going to cost a ton anyway.

I dont know man, the CPU block was pretty cheap and the GPU block came "free" on the graphic card... I actually thought about putting the second 280 rad on the floor and the 360 thick rad in the front, but the cost of the loop would jump by 25% and a thick 360 rad filled with water also is not exactly lightweight.
 
I dont know man, the CPU block was pretty cheap and the GPU block came "free" on the graphic card... I actually thought about putting the second 280 rad on the floor and the 360 thick rad in the front, but the cost of the loop would jump by 25% and a thick 360 rad filled with water also is not exactly lighweight.
Fair enough but a standard 280 alphacool radiator is like 80$, I just don't think it's that big of a deal for a build that's probably around 2K$ already.
 
I dont know man, the CPU block was pretty cheap and the GPU block came "free" on the graphic card... I actually thought about putting the second 280 rad on the floor and the 360 thick rad in the front, but the cost of the loop would jump by 25% and a thick 360 rad filled with water also is not exactly lightweight.
Hi,
You don't show a region so can't say what cpu might do better
Optimus in the US or TechN in the EU.
 
Don't think it can be monitored by msi AB and tbh I've never really worried about coolant temps as 99% of the time my CPU & GPU temps are fine so I find coolant temp negligible but each to their own loop ;)

I highly recommend using a temp sensor stop fitting. You can connect it to the motherboard and have hwinfo readout and exported to msi AB.

All the fans should be controlled by the coolant temp, which warm up and cool down very slowly, leading to very smooth fans transitions.

31anGy0wfOL.jpg
 
I don't know if these estimates ever meant anything, CPU/GPUs have different cooling needs today vs 10 years ago even if they use the same amount of power.

I think a better rule of thumb is too simply fit as many rads as you can reasonably have, it really doesn't make sense trying to get stingy at amount of radiators, the thing is going to cost a ton anyway.
The thing about estimate or a guesstimate is that no accuracy is involved.

As much radiator as possible sounds good to me. I had a Silverstone TJ07 case packed full of radiators. Probably too many actually.

Recycling heat isn't good. The temp rise is exponential, his system will overheat in due time.

Either way, just trying to help. :)
 
I let the PC run full blast (all fans and pump at 100%, opened front door of the case and removed top panel and dust filter, and these are the temperatures with 340W heat load:

temps full blast.png

I presume that they are quite good?
 
I highly recommend using a temp sensor stop fitting. You can connect it to the motherboard and have hwinfo readout and exported to msi AB. All the fans should be controlled by the coolant temp, which warm up and cool down very slowly, leading to very smooth fans transitions.
I have this fitting fitted on the pump, unfortunately my motherboard has no connector for temp sensor! I am thinking about getting Corsair iCUE Commander CORE XT fan controller, but that would require installing Corsair software...
 
I let the PC run full blast (all fans and pump at 100%, opened front door of the case and removed top panel and dust filter, and these are the temperatures with 340W heat load:

View attachment 333752

I presume that they are quite good?
Hi,
Not really man 81c and 53c is large core temp spread assuming those are p cores only ?
Really need an image of all your P core temperatures because this will give the best gauge of how well your block is making contact plus how good it is at large.
E cores temps are useless scrambled into this

1707433045691.png
Give you an example your highest and lowest P cores will tell you how good of contact you have.
Usually if those numbers are less than 10c apart you've done pretty well

With all this water cooling stuff I ended up installing my optimus foundation water block again with mx-6 timm
I only have a 1 old ek 280pe rad with gpu and cpu on it
I ran CB 2024 and left it at stock but remove the timers restrictions so lowest the clocks went is 4.9 plus a hair.
Score sux lol but my core temp spread is at 75-67c apart so that is 8c difference so the mount is making contact pretty good.

1707433088140.png
 
81°C is a maximal distance to temp limit recorded, the temp of the CPU was highest, when the distance from temp limit max was smallest. Temp limit minus 47, 100-47 is 53, the maximal CPU temperature was 53°C, as is written in the third column first row of the picture.
 
.... you don't need to mix and match intake/exhaust radiators because the radiators dealing with hot air aren't really adding anything of value other than increased complexity, cost, and flow restriction.

I measured the performance of my 280 rad, see a thread in overclocking section. Here are the least wrong numbers I produced so far:

280 rad perf.png

Let us say we have ambient temp 20°C, coolant temp 35°C, one rad blowing air in the case and one rad blowing the warm air out of the case.

I actually measured the temp of the warm air from the intake rad to be 32°C.

The intake rad works with delta 15°C and has cooling power 466W. The exhaust rad works with delta of only 3°C (using 32°C air for cooling) and has cooling power 100W (see in the measurement thread how I got this number from the thermal resistance I meaured!).

Total cooling power of these rads would then be 466 + 100 that is 566W.

Let us say we are really loading these rads with 566W. If we made them both use the ambient cold air and one rad would be dealing with 283W, the corresonding delta is 9°C. The coolant temperature would then be 20+9, that is 29°C.

Recapitulation:

With heat load of 566W, using one rad as an air intake in the case and the second as an air exhaust would lead to coolant temperature of 35°C. Using both rads as air intakes would lead to coolant temperature 6°C lower.


Adding a rad on the case that works with warm air produced by other rads in the case really has a limited effect, in this example improving cooling power only by 20%.

The difference is in the coolant temperature - with 566W load,

one rad would cause 18°C delta,
one rad as an air intake and one rad as an air exhaust 15°C delta
both rads as air intakes 9°C delta.


Now I actually swapped the top rad to be an air intake too, the temps really improved, and now my case is swamped with warm air and I have just one 140mm fan and holes in the case to get rid of it. I do not like that too much...
 
Last edited:
81°C is a maximal distance to temp limit recorded, the temp of the CPU was highest, when the distance from temp limit max was smallest. Temp limit minus 47, 100-47 is 53, the maximal CPU temperature was 53°C, as is written in the third column first row of the picture.
Hi,
Yeah I saw the tjmax after I posted the rest is just language clash on what the entries are.

Issue is you need to expand the Core temperatures section like I did so we can see all of them not just 1
Preferably after running CB 2024 or blender opendata this will show how well the water block is mounted until then the piping setup is pretty much a silly discussion
Like others have said water temperature equalizes so routing is just which ever pipes the cleanest.

But Yes indeed intake air on rads is what I always do they love cool air plus they already have screen filters.
 
I measured the performance of my 280 rad, see a thread in overclocking section. Here are the least wrong numbers I produced so far:

View attachment 333863
Let us say we have ambient temp 20°C, coolant temp 35°C, one rad blowing air in the case and one rad blowing the warm air out of the case.

I actually measured the temp of the warm air from the intake rad to be 32°C.

The intake rad works with delta 15°C and has cooling power 466W. The exhaust rad works with delta of only 3°C (using 32°C air for cooling) and has cooling power 100W (see in the measurement thread how I got this number from the thermal resistance I meaured!).

Total cooling power of these rads would then be 466 + 100 that is 566W.

Let us say we are really loading these rads with 566W. If we made them both use the ambient cold air and one rad would be dealing with 283W, the corresonding delta is 9°C. The coolant temperature would then be 20+9, that is 29°C.

Recapitulation:

With heat load of 566W, using one rad as an air intake in the case and the second as an air exhaust would lead to coolant temperature of 35°C. Using both rads as air intakes would lead to coolant temperature 6°C lower.


Adding a rad on the case that works with warm air produced by other rads in the case really has a limited effect, in this example improving cooling power only by 20%.

The difference is in the coolant temperature - with 566W load,

one rad would cause 18°C delta,
one rad as an air intake and one rad as an air exhaust 15°C delta
both rads as air intakes 9°C delta.


Now I actually swapped the top rad to be an air intake too, the temps really improved, and now my case is swamped with warm air and I have just one 140mm fan and holes in the case to get rid of it. I do not like that too much...
Sounds about right - thanks for the more in-depth testing, it's not really very complicated from a logic point of view;

If your intake radiator gets coolant down to 40C and heats up the inside of your case to 40C (fairly believable, ordinary values) then the exhaust radiator breathing in 40C air to cool down 40C coolant at the do nothing, because it's working on a ΔT of zero, and anything multiplied by zero is still zero.
 
Back
Top