Question Flow rate severely restricted

[lingarajug]

I built a system sometime ago and recently added a sensor to detect temp and flowrate thats made by tt, have xspc d5 170 combo, from the pump to cpu block hard rube about 2.5ft long with 2 90 bends then a ft long tube to 360 rad above with 2 90 bends and then to gpu block and it is ft long with 2 90 bends and to another 360 rad above and it is ft long tube and then 280 rad below with 4ft long tube with sensor in between to a T fitting and it has 3 90 bends apart from T fitting and then 1.5ft long tube with 2 90 bends to reservoir, now this pump is rated at 1200lph flowrate but i am getting only 130lph and my loop seems to restrict flow by 90%, i have 7900x ocd to 5ghz and a 2080ti ocd to 2000mhz so my hardware needs lot of cooling and i have a very big case where my mobo asus x 299 apex lays flat on tray parallel to floor, i wanted 2 rads to exhaust hit air and the bottom rad to provide cool air into frame along with front 300mm fan also pushing in cold air and the top 2 rads have 6 fans with push pull to remove have better, at this point i believe flow rate is limiting my cooling performance, how can i improve flow rate, thanks.

 

[Gmoney06ss]

The flow rate might be 1200, but that's before any restriction. So basically an open pump. While 130 does sound low, it may be sufficient. What are your temps and do you feel it's overheating? You may add a secondary inline pump to add more flow rate. Do you have any pics of the setup? That wall of text makes it a little hard to picture your setup.

 

[lingarajug]

The flow rate might be 1200, but that's before any restriction. So basically an open pump. While 130 does sound low, it may be sufficient. What are your temps and do you feel it's overheating? You may add a secondary inline pump to add more flow rate. Do you have any pics of the setup? That wall of text makes it a little hard to picture your setup.

Gpu stays pretty cool never goes past 40c, its the cpu that reaches between 80 to 85c, although its a safe temp for cpu i like it to be below 65c, i might add another pump, where should i add it between cpu and rad or between rad and gpu or between 2 rads, my direction is pump combo>cpu>360 rad>gpu>360 rad>280 rad>pump combo. Total of 11 90 degree bends and cpu and gpu block.

 

[Gmoney06ss]

It may be pretty hard to get such a hot cpu down that low, especially at 5ghz. But my guess is the most restricted part of the system is the cpu block and since you already have one pump feeding you could try it after it to get the flow back out. The 90 bends aren't much of a restriction at all, it's the plates in the blocks that add up.

Keep in mind this may not help, as that is a hot cpu, and your temps aren't too high as is. And since the gpu temps are well within great range, flow may not be the issue.

Are you delidded? May see more of a benefit there then adding a second pump.

 

[lingarajug]

It may be pretty hard to get such a hot cpu down that low, especially at 5ghz. But my guess is the most restricted part of the system is the cpu block and since you already have one pump feeding you could try it after it to get the flow back out. The 90 bends aren't much of a restriction at all, it's the plates in the blocks that add up.

Keep in mind this may not help, as that is a hot cpu, and your temps aren't too high as is. And since the gpu temps are well within great range, flow may not be the issue.

Are you delidded? May see more of a benefit there then adding a second pump.

Yes my 7900x is delidded and conductonaut with custom copper ihs and sitting under ekwb block that covers both cpu and vrm, with so much restriction i am not sure if pump is working too hard to push fluid and adding more heat to the loop and worried if pump might fail in the long run.

 

[rubix_1011]

How do you know you are getting only 130lph? Is that what the 'sensor' is? Your loop isn't that complex, but can you provide a photo?

Have you checked temps without CPU overclocked, just to validate as a baseline? Most people forget to do this, so they do not have any comparison.

What CPU block is being used? What GPU block? Are you using the correct inlet/outlet ports?

Also, very difficult to read the posts...just a lot of run-on sentences. Can you use punctuation to clean up the flow a bit?

 

[lingarajug]

How do you know you are getting only 130lph? Is that what the 'sensor' is? Your loop isn't that complex, but can you provide a photo?

Have you checked temps without CPU overclocked, just to validate as a baseline? Most people forget to do this, so they do not have any comparison.

What CPU block is being used? What GPU block? Are you using the correct inlet/outlet ports?

Also, very difficult to read the posts...just a lot of run-on sentences. Can you use punctuation to clean up the flow a bit?

@rubix_1011 and @Gmoney06ss i get my lph value from the sensor as seen picture, i use ekwb monoblock made for asus rampage vi apex that cools cpu and vrm and i use full cover gpu block made by bitspower, without cpu over clocked temps are around 45 to 55. Pics of setup in the link, https://www.dropbox.com/s/7slkifn9xeo7z8x/IMG_20190319_185307.jpg?dl=0

 

[rubix_1011]

Ah, thank you, that helps.

So, the 45-55 (assume Celsius?) is stock speeds and at loads? What are you using for loading the CPU? Something like Prime95? If so for both stock and overclocked testing, your results are probably about right. The only thing you could do is to try and minimize the vcore needed for the CPU clocks, if anything.

What is the PCIe card that is in front of your GPU?

Where did you get that little flow meter? That's pretty slick. Do you have a link?

 

[lingarajug]

Ah, thank you, that helps.

So, the 45-55 (assume Celsius?) is stock speeds and at loads? What are you using for loading the CPU? Something like Prime95? If so for both stock and overclocked testing, your results are probably about right. The only thing you could do is to try and minimize the vcore needed for the CPU clocks, if anything.

What is the PCIe card that is in front of your GPU?

Where did you get that little flow meter? That's pretty slick. Do you have a link?

@rubix_1011 yeah its C and at stock full load where it boosts to 4000 on 8 cores and 4300 on 2 cores, with oc and i used prime95 and if u run it for ling like an hour it will reach 90C too but i only run for about half hour and i am guessing coz the monoblock cools both vrm and cpu i believe cpu is not cooled well, i mostly run at stock speeds and only use oc when i do heavy video editing, my chip is not stable under 1.265 for slower cores and 1.315 for faster cores, and that sensor is on amazon for ariund 25$ i believe, i am more concerned pump dying sooner than expected due to 90% back pressure on it, i will probably clean all blocks and tubes and redo whole loop using a Y splitter to push to cpu and gpu from the pump and then to rads from blocks and then Y again to pump to reduce some tube length.

 

[rubix_1011]

Flow restriction is common in every loop. For your CPU monoblock, are you using the correct inlet port? I know on nearly all monoblocks, they require flow to only go one direction - most are not bi-directional. Usually the inlet is nearest the actual CPU IHS itself and the outlet is routed further away. This would likely be the first thing I check along with how well the monoblock is seated on the CPU.

Prime95 is excessive at loading a CPU, so while it is good to evaluate your cooling and overclock, it isn't a realistic test of daily usage and loads. I wouldn't put too much stock into running it for lengthy periods of time unless you really wish to ensure the highest levels of stability and thermal control. Also, with a healthy overclock, any CPU is going to run hotter than usual, which is the nature of adding more voltage to every clock tick at an increased rate.

I'll try and hunt down that flow meter - I would like to take a look at it to see if it is worth adding to my liquid cooling lab gear. I'd also like to test it to see how accurate it is. (do you have a link to it? Can you drop it in, if so?)

What is the other PCIe card that is in the foreground by the GPU?

 

[lingarajug]

Flow restriction is common in every loop. For your CPU monoblock, are you using the correct inlet port? I know on nearly all monoblocks, they require flow to only go one direction - most are not bi-directional. Usually the inlet is nearest the actual CPU IHS itself and the outlet is routed further away. This would likely be the first thing I check along with how well the monoblock is seated on the CPU.

Prime95 is excessive at loading a CPU, so while it is good to evaluate your cooling and overclock, it isn't a realistic test of daily usage and loads. I wouldn't put too much stock into running it for lengthy periods of time unless you really wish to ensure the highest levels of stability and thermal control. Also, with a healthy overclock, any CPU is going to run hotter than usual, which is the nature of adding more voltage to every clock tick at an increased rate.

I'll try and hunt down that flow meter - I would like to take a look at it to see if it is worth adding to my liquid cooling lab gear. I'd also like to test it to see how accurate it is. (do you have a link to it? Can you drop it in, if so?)

What is the other PCIe card that is in the foreground by the GPU?

Yes the tubing is connected to correct port on monoblock in the correct order, and that pcie card you see is asus pcie nvme enclosure, it has 2 samsung 970 pro 1tb nvme inside and the heat is transfered to the outer case and there is built in fan that cools the whole case and here is the link for the sensor, Thermaltake Pacific TF1 Cooler... https://www.amazon.com/dp/B07KR8VQQL?ref=ppx_pop_mob_ap_share

 

[rubix_1011]

Thank you for the link...I might buy this and test/validate accuracy and use it if it is actually any good.

OK, so sounds good on the block flow. Only thing I can think of is just temps due to OC or you could look into re-application of thermal paste, possibly using some high-end stuff like Thermal Grizzly Conductonaut or Kryonaut.

Personally, I think you're fine and you will unlikely only see temps like that by running Prime95. Normal usage under loads would never reach those temps.