Temperature Check for Custom Loop

[Pikatchu]

I have recently made a custom loop cooling for my machine, consist of Ryzen 9 7900x and a RTX 3080 FTW3.

My question is are the temperatures normal for custom loop? CPU temperature does not seem too far off from air cooling.

 

My CPU is at a -10 curve optimizer offset, and EXPO setting.

My GPU is at max power slider with +120 core +300 memory.

 

CPU on air:

Cinebench All Core - 92C | 180~190W

Gaming - 70~75C | 100W~

Idle - 45~50C

 

GPU on air:

Furmark - 70C | 400W~

Gaming - 65C | 400W~

 

CPU on custom loop:

Cinebench All Core - 90C | 180~190W

Gaming - 60~65C | 100W~

Idle - 45~50C

 

GPU on custom loop:

Furmark - 55C~ | 350W~

Gaming - 55C~ | 350W~

 

EDIT: The loop is cooled by one Corsair 360mm slim radiator.

[PDifolco]

32 minutes ago, Pikatchu said:

I have recently made a custom loop cooling for my machine, consist of Ryzen 9 7900x and a RTX 3080 FTW3.

My question is are the temperatures normal for custom loop? CPU temperature does not seem too far off from air cooling.

 

My CPU is at a -10 curve optimizer offset, and EXPO setting.

My GPU is at max power slider with +120 core +300 memory.

 

CPU on air:

Cinebench All Core - 92C | 180~190W

Gaming - 70~75C | 100W~

Idle - 45~50C

 

GPU on air:

Furmark - 70C | 400W~

Gaming - 65C | 400W~

 

CPU on custom loop:

Cinebench All Core - 90C | 180~190W

Gaming - 60~65C | 100W~

Idle - 45~50C

 

GPU on custom loop:

Furmark - 55C~ | 350W~

Gaming - 55C~ | 350W~

What was your previous air cooler ?

Ryzen idle temp is a kinda fake calculated temp it's always in the 40-45C, so I wo'nt bother much with that

Seems the loop is working with -10C in gaming and Furmak

With Cinebench did you have same clocks on both ? The loop may push the CPU higher then hotter

[Pikatchu]

6 minutes ago, PDifolco said:

What was your previous air cooler ?

Ryzen idle temp is a kinda fake calculated temp it's always in the 40-45C, so I wo'nt bother much with that

Seems the loop is working with -10C in gaming and Furmak

With Cinebench did you have same clocks on both ? The loop may push the CPU higher then hotter

NH-D15 for CPU.

With Cinebench, I did have slightly higher scores.

 

Loop order is Res -> Rad -> CPU -> GPU -> Res.

 

Oh right forgot to mention, all of these are cooled by one Corsair's 360mm slim rads.

[NorKris]

so u added a 3080 to the 360

cpu needs a 240 and gpu "needs" a 240 

[johnno23]

Considering you are using a single 360 rad your temps look pretty good. you are however really pushing the limits there.

I do suggest you monitor the temps on a regular basis however as over time any possible build up of residue will get caught in the micro fins on the CPU block which will reduce flow over time and then you are faced with a possibility that its efficiency to cool will drop.

I would save up and invest in an extra rad but you did not mention your resevoir and pump ?

You can help reduce the issue if the case has room for a larger res if yours is on the smaller side. By increasing the volume of fluid in the system you can help slow down the process of the fluid heating up.

These days I monitor things but also use a digital temp sensor in my loop to read the fluid temps. thats my main go to source of info about the system.

 

 

[PDifolco]

22 minutes ago, Pikatchu said:

NH-D15 for CPU.

With Cinebench, I did have slightly higher scores.

 

Loop order is Res -> Rad -> CPU -> GPU -> Res.

 

Oh right forgot to mention, all of these are cooled by one Corsair's 360mm slim rads.

Huh only 1 rad ? That's not much, you should add a second one (240 will do)

[johnno23]

3 minutes ago, PDifolco said:

Huh only 1 rad ? That's not much, you should add a second one (240 will do)

indeed and using a slim one as well. slim line rads can be efficient but the fluid volume is also lower. six months from now the temps could well be back to air temps and possibly worse if unliucky with any residue build up on the micro fins in the CPU block.

[Pikatchu]

10 minutes ago, PDifolco said:

Huh only 1 rad ? That's not much, you should add a second one (240 will do)

 

5 minutes ago, johnno23 said:

indeed and using a slim one as well. slim line rads can be efficient but the fluid volume is also lower. six months from now the temps could well be back to air temps and possibly worse if unliucky with any residue build up on the micro fins in the CPU block.

 

I am currently using Openbench table as my "case".

It could fit more than 1 radiator, but fitting more than 1 will make it significantly harder to move it around...

 

I guess I will be looking on moving back to a metal box if I want to add more radiators.

[NorKris]

but there u go, 2c better than NHD15 with  a 3080 added to the mix  

[johnno23]

37 minutes ago, Pikatchu said:

I guess I will be looking on moving back to a metal box if I want to add more radiators.

your system is running ok right now and temps are ok. If you enjoy using the open bench just add a small digital flow temp sensor to the loop. they are very inexpensive these days. Not like in the past and so easy to read the fluid temps. if the fluid remains at temps that are not too high you are fine. Plus the flow meter whilst never 100% accurate really is a good indication of whether there is any build up of tiny particles that get caught in the micro fins.

I clean my system about every 9 months or so. I am using a full water block on the 3080 and an EK velocity on the AMD 5900x. the velocity gives me issues over time as the jet plate above the fins has a very narrow slit to force the water thru which in turn creates the turbulence that in turn enhances the cooling efficiency. on cleaning my flow is around 4.5 Lmin but after about 9 months it is reduced to around 2.6Lmin. 

As far as I know so long as it does not drop under 2.5Lmin then it is fine.

when I clean the block it is a tiny amount f residue caught in the micro fins directly under the jet plat where the slit is.

Takes me about 30  minutes in total to drain detach the block open her up clean repaste CPU and reassemble then refill. on an open bench you could do it much faster

[Pikatchu]

46 minutes ago, johnno23 said:

your system is running ok right now and temps are ok. If you enjoy using the open bench just add a small digital flow temp sensor to the loop. they are very inexpensive these days. Not like in the past and so easy to read the fluid temps. if the fluid remains at temps that are not too high you are fine. Plus the flow meter whilst never 100% accurate really is a good indication of whether there is any build up of tiny particles that get caught in the micro fins.

I clean my system about every 9 months or so. I am using a full water block on the 3080 and an EK velocity on the AMD 5900x. the velocity gives me issues over time as the jet plate above the fins has a very narrow slit to force the water thru which in turn creates the turbulence that in turn enhances the cooling efficiency. on cleaning my flow is around 4.5 Lmin but after about 9 months it is reduced to around 2.6Lmin. 

As far as I know so long as it does not drop under 2.5Lmin then it is fine.

when I clean the block it is a tiny amount f residue caught in the micro fins directly under the jet plat where the slit is.

Takes me about 30  minutes in total to drain detach the block open her up clean repaste CPU and reassemble then refill. on an open bench you could do it much faster

May I ask what case are you using?
 

O11D is nice but kind of too big of a foot print.

Lancool 3 looks nice and is probably my go to.

 

If I were to get a case, I will probably do 2 360mm.

[johnno23]

be quiet base pro 900 as I have it inverted and wanted to use a 420mm radiator

the case is quite nice and will handle 2 x 360 absolutely no issue. the 3 fans that come with the case were the silent wings 3 so that makes the case much less expensive when you look what those cost. it also has a couple of simple led strips for colour and the fan module is unobtrusive and needs no extra software. Also has a slider on the front behind the door to micro manage fan speeds on the fly should ambient temps change dramtically with the weather so no need to continually jump into any settings. etc.

as you can gather i am a fan of the Be quiet case and used them for the last 12 years or so.

I am using a push pull 420mm 6 fans on the front and a 280 rad on the top

full water block GPU and velocity for the CPU. I also used a dual D5 pump housing with a large cylinder resevoir. Lots of fluid as I live in Southern China and it get really hot in the summer months here.

[TPCEA]

2 hours ago, johnno23 said:

your system is running ok right now and temps are ok. If you enjoy using the open bench just add a small digital flow temp sensor to the loop. they are very inexpensive these days. Not like in the past and so easy to read the fluid temps. if the fluid remains at temps that are not too high you are fine. Plus the flow meter whilst never 100% accurate really is a good indication of whether there is any build up of tiny particles that get caught in the micro fins.

I clean my system about every 9 months or so. I am using a full water block on the 3080 and an EK velocity on the AMD 5900x. the velocity gives me issues over time as the jet plate above the fins has a very narrow slit to force the water thru which in turn creates the turbulence that in turn enhances the cooling efficiency. on cleaning my flow is around 4.5 Lmin but after about 9 months it is reduced to around 2.6Lmin. 

As far as I know so long as it does not drop under 2.5Lmin then it is fine.

when I clean the block it is a tiny amount f residue caught in the micro fins directly under the jet plat where the slit is.

Takes me about 30  minutes in total to drain detach the block open her up clean repaste CPU and reassemble then refill. on an open bench you could do it much faster

I'm curious. If you use distilled water,  will that reduce sediment in the water even with whatever you add?

[johnno23]

I use the alphacool premix as it is not so expensive and lasts a year no problem. should last longer but the amount of residue that reduces flow is really a tiny amount.

Possibly just miniscule particles from manufacturing of the rads ? possibly just a few bits of dust that get into the system when cleaning refilling etc.

Lots of people do use distilled water then add what they need afterwards but as stated the alphacool premix which I use is a very reasonable price and save the messing around.. I just use what was in the system after draining into a plastic bowl to clean things as the micro fins housing internals etc with a soft toothbrush

I also do not use any colourants......led lighting is all that is needed so adding colour to the liquid I just find a waste of time....personal opinions of course.

 

[TPCEA]

40 minutes ago, johnno23 said:

I use the alphacool premix as it is not so expensive and lasts a year no problem. should last longer but the amount of residue that reduces flow is really a tiny amount.

Possibly just miniscule particles from manufacturing of the rads ? possibly just a few bits of dust that get into the system when cleaning refilling etc.

Lots of people do use distilled water then add what they need afterwards but as stated the alphacool premix which I use is a very reasonable price and save the messing around.. I just use what was in the system after draining into a plastic bowl to clean things as the micro fins housing internals etc with a soft toothbrush

I also do not use any colourants......led lighting is all that is needed so adding colour to the liquid I just find a waste of time....personal opinions of course.

 

Some colorants are also made with precipitates, so that's really bad, especially the viscous stuff. 

[johnno23]

yup,,,,and many of those colourants leave a lot of rubbish behind.They have improved over time but simple clear fluid in the resevoir picks up the colour of the leds from any led fans and even the smaller leds on the motherboard so I simply never felt the need to use them. Also many windowed side panels these days have a smokey appearance so colours would be altered.

I think the colourants were much more popular in the days before the led lighting became so prevalent. After all it was a way to dress up the water cooling and make it look more impressive so many years ago.

[TPCEA]

4 minutes ago, johnno23 said:

yup,,,,and many of those colourants leave a lot of rubbish behind.They have improved over time but simple clear fluid in the resevoir picks up the colour of the leds from any led fans and even the smaller leds on the motherboard so I simply never felt the need to use them. Also many windowed side panels these days have a smokey appearance so colours would be altered.

I think the colourants were much more popular in the days before the led lighting became so prevalent. After all it was a way to dress up the water cooling and make it look more impressive so many years ago.

I eventually want to build a custom loop, but I'm not interested in hard tubing unless I put the loop outside of the case. In that case, I'll have some fun with it. 

[johnno23]

I used the black tubing from EK. this type https://www.ekwb.com/shop/ek-tube-zmt-matte-black-19-4-12-5mm

Inside a black interior it is perfectly fine and very easy.

As My case is inverted so the the case window is on the right as the corner of the room where I sit has wall facing me and to the left and behind me on that left hand wall is a window.

The upside to an inverted case is/was better airflow and lower temps over my 3080 before I converted everything to a custom loop.

So my loop was also easy to build.

dual D5s in a single dual housing with a tall cylinder res mounted to it. from the pump housing outlet direct into the digital flow meter then to the cpu/ cpu direct to gpu and gpu to the top 280 rad. 280 rad direct to the 420 rad and then 420 to the res. so the fluid has a huge area of cool down time before it has to recirculate. 

The dual D5s are not really needed but I use it as a redudancy measure should one ever fail. It does give me some extra flow but adding pumps is diminishing returns when in the same loop. Aty least in this scenarion should a pump fail I can still use the PC without issue till I have a replacement and in the meantime just having extra flow as a bonus.

When I totalled up my costs the extra pump was minor and doing dual block housing made sense. Just stay away from using exclusively EK as it is over priced. most of my kit is from Barrow and the D5s are oem made in Chna which where 95% of the pumps are made anyways.

 

[Pikatchu]

On 4/8/2023 at 12:07 PM, PDifolco said:

Huh only 1 rad ? That's not much, you should add a second one (240 will do)

 

On 4/8/2023 at 11:59 AM, johnno23 said:

Considering you are using a single 360 rad your temps look pretty good. you are however really pushing the limits there.

I do suggest you monitor the temps on a regular basis however as over time any possible build up of residue will get caught in the micro fins on the CPU block which will reduce flow over time and then you are faced with a possibility that its efficiency to cool will drop.

I would save up and invest in an extra rad but you did not mention your resevoir and pump ?

You can help reduce the issue if the case has room for a larger res if yours is on the smaller side. By increasing the volume of fluid in the system you can help slow down the process of the fluid heating up.

These days I monitor things but also use a digital temp sensor in my loop to read the fluid temps. thats my main go to source of info about the system.

 

 

Wanted to give an update, I have added a 240mm radiator to the loop, but swapped my D5 for a DDC for space issue.

I have moved my pc back into a case, Lancool 2 mesh.

 

My temperature have not changed at all which is very interesting 

[TPCEA]

17 minutes ago, Pikatchu said:

 

Wanted to give an update, I have added a 240mm radiator to the loop, but swapped my D5 for a DDC for space issue.

I have moved my pc back into a case, Lancool 2 mesh.

 

My temperature have not changed at all which is very interesting 

Please take a photo and add arrows to show the direction each fan blows air.

[Pikatchu]

All Intakes. The bubbles have settled down now, this was taken right after filling the loop.

[TPCEA]

33 minutes ago, Pikatchu said:

All Intakes. The bubbles have settled down now, this was taken right after filling the loop.

 

Well that explains it. If you have no exhaust, the heat builds up in there because you're not pulling in cold air - you're pulling in cold air that is heated by the heated water from the CPU & GPU.  Turn the fans on the top rad the other way, at least. Let us know if that helps.

[NorKris]

6 minutes ago, RevGAM said:

Well that explains it. If you have no exhaust, the heat builds up in there because you're not pulling in cold air - you're pulling in cold air that is heated by the heated water from the CPU & GPU.  Turn the fans on the top rad the other way, at least. Let us know if that helps.

misinformation

[NorKris]

what is ur GPU temps when running 10m full res furmark? 

[johnno23]

I think your solution is solid.....your temps were ok but the limits with regards to how long you could keep the system pushing to its max before the fluid temps rose too high have now been removed. I expect that you could run any tests you like and hit a stable temp and simply stay there for hours on end without any concerns.

I do not know your location but if you have crippling heat in the summer months you could help by adding a fan at the back just to assist with moving air out of the system. the fan could be mounted on the outside so it would not impact your top rad and fans.3 points to secure the fan are enough and all 3 points in the photo are easy to access. If you are in Europe you can ignore the fan idea completely. but if in a geo location where temps are extreme it is just an extra source for removing internal air rapidly from the case thats all.

Also your 7900x has I believe a target temp of 95c for optimal performance and that you are keeping it to 90c is excellent.