Does Flow Rate Matter? - Complete Watercooling Guide - Part 15
Вставка
- Опубліковано 22 сер 2024
- Re-running our 10 years old written flow rate test in video review format. Investigating the watercooling system flow rate impact on cooling performance.
Stress Test SW: AIDA64, Furmark
NewSys1: GAZ270XGaiming8/i5-7500/GTX980
NewSys2: GADesignare/TR1950X/RTX2080ti
OldSys1: EVGAP55LE/i5-750/R4870
OldSys2: GAX58/i7-920/2xGTX285
►Subscribe and Share our videos on forums
►Visit our store @ www.dazmode.com...
►Follow on Facebook and Twitter / dazmode ... / dazmode
------------------------------------------------------------------------------------------
Respect each other in the comments.
Daz Mode continues to be the best for no bullsh1t watercooling advice.
Thanks for vote of confidence :)
Glad to see you back Daz. I always wait for your videos even that I already learned everything about LC from you in past few years - I still enjoy it. About flow, I have my D5 set to 1,and it's enough to me. In past builds I used flow meters, but in my current one I decided not to. If something is wrong, you can always notice it with higher temps. Wish you all the best and thanks for video.
R5-5600x 4.8Ghz with a cheep pump 2.5x slower (week) than D5 on max, and just order a D5 with top and 10L water for my build and will oc up to 5.3Ghz for my CPU. I have 360mm and 140mm radiator with noctua fans.
Hi Daz, Great to see you back posting videos. This one was very helpful and proved out what I was getting out of my loop. Basically showing that as long as the D5 was running above one third of it's speed it has enough flow rate to keep my CPU and both GPU's at around 45c while gaming. I find running my single D5 at 2800rpm gives an excellent trade off between noise and cooling.
Your data showing water flow rates and fan speeds was very interesting.
If noise the issue - slowing down D5 or DDC almost no cooling performance penalty.
Der8auer just did a video today on flow rate. Cool that we got videos from both of you guys!!
Thanks! Checking it now. ... He is talking more about loop order. Good video.
@@DazModeWatercooling I had to do a double-take after seeing Der8auer's vid as well. Both great to watch as it is great information for those of use who don't have the gear to test those things.
@@dandel351 What the odds? The same subject same day.
@@DazModeWatercooling ... Insert "Thinking Emote" here.
happy you're back
People mounting fans directly to radiators is more of an issue, it seems forgotten over the past decade that shrouds should be used to space the fans out from your rads by about 30mm
Yeah people do not understand how static preasure works
The i7 920 and the 4870 was the first set up i saved up for and got during highschool. Literally spent all my money at the time on it. Fun to see it mentioned again. Both parts were such a steal for the price and ability to overclock
Increasing the flow rate is the only way to bring the radiator temp closer to room temp given the fan speed is constant or if there are no fans at all. Basically you have two fluids - air and water. They are touching each other indirectly via the fins in the rad. The more you can make those two mediums exchange heat, the better the performance. The way to do it is to mix them as much or as often as possible. More flow - more water, more fan speed - more air. Once the maximum heat exchange rate has been reached, mixing the fluids more violently/faster will do nothing. Some people have experimented with car radiators and they tend to perform better, but up to a point. Basically you can put 6 3000 RPM fans on a huge car radiator and that will perform roughly the same as a 360 rad with two 1300 rpm fans (as an example). In order for the car rad to do anything useful past the 100-400 watt load with any type of fan, you need to increase the load - which is pointless / impossible with a computer system. So the bottleneck is then how well the rad can exchange heat between the water and the air.
This video here kinda proves that as long as your flow rate is greater that 3.5 - 4 you're good
I think parallel pumps give you more flow. Connecting them in series increases "head" - which is how high the water can rise (pressure). Your method of pinching the pipes to decrease flow by increasing the resistance in the loop and makes it "harder" for the pumps to push through. Ideally I think you should have tested only with various pump speeds and having otherwise identical resistance in the system. Anyway, IDK much about fluid dynamics but the theory is parallel = more flow afaik. I suspect that I can beat a 80+ dollar D5 or DDC pump with 3-4-6 cheap aquarium pumps in parallel that cost under 5 dollars each, but then again if it were that simple/easy other people would have probably have done it by now. Or no one wants to bother with aquarium pumps and external water tanks...
Nice case, I liked that model so much I bought two, LOL!
Well known by experienced builders but nonetheless it is good to see one more verification with test. My personal sweet spot is 100 l/h cuz this is easy to achieve with pretty much any pump and with minimal noise and yet it has some safety gap from 60-70 l/h where flow rate starts to matter.
So, (3) D5’s in series at 100% speed (4800RPM) can push 7LPM or 420LPH?
I think a great experiment (that I really want to do) is hooking a cheap 40 x 60 cm aluminium car radiator to cheap aluminum blocks and pumping the water with 4-6-8 cheap aquarium pumps in parallel on minimum speed from a tank of 5-10 liters. No fans, or possibly a single fan. The total price would be ~ 230 USD with fittings, tubes and fan included. I think that would match or beat the performance of a "custom" overpriced system AND be quieter (It's really not custom, but purpose made to fit in a computer case). I think the convenience / practicality of having all the tubes and the water inside the case is slim or none unless you move the case a lot - which I have no intention of doing. The only draw back of the system I'm talking about is the looks. But a "nice/tidy" system with two blocks can cost as much as 800 USD and at that point it's dumb because you are paying 50-30% on top of the the cost of the original cpu/gpu for a 10-15% increase in performance. Or you are paying for the cosmetics of it. For low/mid-range components it's ultra dumb if you focus on performance per dollar.
its called custom because you can add change and connect however you want. unlike the closed loop which you cant do anything with
i went from the i7 870 to a 2700k :) my first cpu was a Q8200
The problem I am having with high flow rate is that my reservoir needs to be bigger to accommodate the pressure. In my current rig around 2lpm is perfect, any higher and I start sucking in air.
What was the system setup to do the testing?
Was it under full load? If so what software were you running?
Good topic for debating lol.
I'll update specs in descr later. I used AIDA64 and AIDA64 + Furmark. Using Furmark for GPU made no difference for me.
Why are the temps higher when using 2100 RPM vs 1500 RPM? Are the graphs labeled incorrectly?
It is not pump RPM
Hey Daz do you know who makes the the highest quality flow sensor that plugs into a fan header? I don't need accuracy I just need it to trigger the CPU fan error when there is no flow on boot.
I like Alphacool's flow sensor because its inline but I heard it has problems making noise as does Auqatuning, Barrow, and Bitspower but I think a plastic washer from a milk jug can fix those.
Mechanical all can click, squeak and etc. I only used Koolance and Aquacomputer myself- the noise was not an issue.
such an informative channel! keep it up
how you doing @DazMode i hava an unknown, I have the connection somewhat different from what is recommended, a friend of mine who was the one who did the configuration assured me that it would have a better temperature if the water entered the input part of the processor block from the direct radiator, My connection is as follows,
from the outlet of the radiator the outlet of the pump is connected so that the pump pushes the water through the radiator first and from the radiator it goes out through the inlet channel towards the inlet of the processor and the outlet of the processor. processor then goes into the pump canister, which seems wrong to me, can anyone help me?
Corsair 7000d build Watercooling Threadripper
Hi guys, hoppe good
This is my first Watercooling Threadripper build.
Watercooling specification
Rads:
front: standard fan from corsair case, but planing on put it on an lian li white 120 mm v2 x4
side: Water Radiator EK-Quantum Surface X480M White + 4x Noctua NF-F12 iPPC 3000 PWM, Heavy Duty Cooling Fan, 4-Pin, 3000 RPM (120mm, Black) With an NA-SAVP1Chromax. White Anti-vibration pads
top: nothing but planing on put it on an lian li white 120 mm v2 x4 or lees
Back: standard fan from corsair case but waitting for a new one 140 mm lian li white Fan
CPU Block: EK-Quantum Magnitude sTRX4 D-RGB - Nickel + Plexi
Thermal Paste: Thermal Grizzly Kryonaut Extreme The High Performance Thermal Paste for Cooling All Processors, Graphics Cards and Heat Sinks in Computers and Consoles (2 Gram) used mor than 2gram....
GPU : ASUS ROG Strix AMD Radeon RX 5700XT Overclocked 8G GDDR6 HDMI DisplayPort Gaming Graphics Card (ROG-STRIX-RX5700XT-O8G-GAMING), with thermal pads upgradeit.....
Res/Pump: EK-Quantum Kinetic TBE 300 D5 PWM D-RGB - Plexi
Flow meter: i want to buy one
Tubing: Thermaltake V-Tubler PETG Tube 16mm (5/8") OD 1000mm 4-pack
Fittings: EK-Quantum Torque Satin Titanium
Temperature sensors: no one but hoppe to have it
Bulkheads: i would llike in the future 2x Alphacool Apex Fillport - chrome or Satin Titanium
Water mix: distilled water without dye to avoid conductivity
PC specification
CPU: AMD Ryzen Threadripper 2da gen 2920x
GPU: ASUS ROG Strix AMD Radeon RX 5700XT Overclocked 8G GDDR6 HDMI DisplayPort Gaming Graphics Card (ROG-STRIX-RX5700XT-O8G-GAMING), with thermal pads upgradeit.
RAM: G.Skill Trident Z RGB 4x8 32 GB, 3200MHz, CL14
MBoard: Asus ROG Zenith Extreme AMD Ryzen X399 Threadripper E-ATX Gaming Motherboard TR4 With 10G lan card
PSU: Seasonic PRIME TX-1000, 1000W 80+ Titanium, Full Modular, Fan Control in Fanless
Case: Corsair 7000D Airflow
Cable riser: not yet but will have one
Storage System M.2: 1) SAMSUNG 970 PRO SSD 512gb PCIe NVMe Gen 4 Gaming M.2 Internal Solid State Hard Drive Memory Card, Maximum Speed, Thermal Control, For OS Windows..., 2) Samsung 970 EVO Plus SSD 512gb NVMe M.2 Internal Solid State Hard Drive w/ V-NAND Technology, Storage and Memory Expansion for Gaming, Graphics w/ Heat Control, Max Speed, 3) 4tb hd Seagate ST4000DM005 4TB SATA Hard Drive, 4) Western Digital Purple 4TB Surveillance Hard Disk Drive - 5400 RPM Class SATA 6 Gb/s 64MB Cache 3.5 Inch - Western Digital40PURZ
Phobya DC12-400 12Volt is what I use and it sucks when I need to tryhard on oc for my r5-5600x just for now stuck in 4.8Ghz and want to oc to 5.3Ghz for my mmo and csgo thats what I order 10L water + D5 for maximum performance, I will not use liquid metal again on copper. 360mm + 140mm with Noctua fans for now and will get 480mm when I get better GPU.
Well, does it truelly matter ? Really ? Fan speed have much more impact on temps. Less than 2 degrees between 1LPM and 1.8LPM.... What i do is i set the max speed to be barely audible on everything mechanical in the bios and i overclock depending on that cooling capability. I hate noise.
Interesting results, but you should really plot against inverse flow rate (minutes per liter), because according to what I could make out from your 2010 data, and your current Threadripper data, the core temperatures have a basically linear correlation with minutes-per-liter, which makes sense because that's how quickly fluid _is removed_ from the block
Я правильно понял что скорость потока оказывает не столь существенное влияние на температуру как скорость вращения вентиляторов?
Da. >1L/min
Bing chat AI is learning from your videos