Tricky Parallel Loop questions (A fun puzzle for all ages)

[CMart]

Hello everyone,
I am new to these forums and was hoping you guys might be able to answer a few questions. This is like my 5th liquid cool build and I am trying some new stuff out. I have read through a lot of the reference material on the forums and didn't see anything about my situation so figured I would just ask. Tried to give everyone the essential info, but if you have more questions don't hesitate to ask.
I plan to run a parallel loop on the GPU waterblocks & need to find the best option to keep temps as low as possible. I have listed the parts & questions below, and even included a diagram of the two parallel config I was thinking to use for the radiators that are apart of this loop.

Parts:

-Graphics Cards:

x2: MSI GeForce GTX 1080 DirectX 12 GTX 1080 SEA HAWK EK X 8GB 256-Bit GDDR5X PCI Express 3.0 x16 HDCP


-Pump:

EK-XTOP Revo Dual D5

-Radiators:

Option 1:
EK-CoolStream PE 360 w/ 4 Fans
EK-CoolStream XE 360 w/ 6 Fans

-OR-
Option 2:
EK-CoolStream PE 360 w/ 4 Fans
EK-CoolStream PE 360 w/ 6 Fans


Graphics card Loop Diagram:
https://drive.google.com/open?id=1j260A1n2j770swj9kUkVGubjVt9kpLxP

Questions:
1. Would it be more efficient to have a serial loop using Option 1 (see parts list) or a parallel radiator setup Option 2(see parts list)?
2. (See Diagram) I have two parallel loops concepted out and was trying to learn which would keep the flow rate at its highest. Option A I would y split and y merge the flow into and out of both radiators. Option B I keep the flows separate off each radiator and run a tube, of exact same length, from each radiator to each graphics card.
3. (See Diagram) I’ve read a lot about tube size and changing tube size to keep flow rates up if you split your initial flow from the pump. Check the diagram for reference. I was thinking if I use 12/16 mm tubing for the tubes leading up and into the radiator and then 10/12 mm on the outputs of each radiator. I was hoping this would essentially slow the flow rate in the radiators giving the liquid more time to cool, but still maintain high flow rate by using the 10/12mm tube to restrict flow. Note: I wasn't sure if I would have to use the 12/16mm tubing or 10/12mm after I split the flow from the pump to achieve a slower flow rate in the radiator.

Thanks for your time and energy! Once I get this beast ordered and put together I will for sure be sharing baby pictures

 

[rts2468]

Hello everyone,
I am new to these forums and was hoping you guys might be able to answer a few questions. This is like my 5th liquid cool build and I am trying some new stuff out. I have read through a lot of the reference material on the forums and didn't see anything about my situation so figured I would just ask. Tried to give everyone the essential info, but if you have more questions don't hesitate to ask.
I plan to run a parallel loop on the GPU waterblocks & need to find the best option to keep temps as low as possible. I have listed the parts & questions below, and even included a diagram of the two parallel config I was thinking to use for the radiators that are apart of this loop.

Parts:

-Graphics Cards:

x2: MSI GeForce GTX 1080 DirectX 12 GTX 1080 SEA HAWK EK X 8GB 256-Bit GDDR5X PCI Express 3.0 x16 HDCP


-Pump:

EK-XTOP Revo Dual D5

-Radiators:

Option 1:
EK-CoolStream PE 360 w/ 4 Fans
EK-CoolStream XE 360 w/ 6 Fans

-OR-
Option 2:
EK-CoolStream PE 360 w/ 4 Fans
EK-CoolStream PE 360 w/ 6 Fans


Graphics card Loop Diagram:
https://drive.google.com/open?id=1j260A1n2j770swj9kUkVGubjVt9kpLxP

Questions:
1. Would it be more efficient to have a serial loop using Option 1 (see parts list) or a parallel radiator setup Option 2(see parts list)?
2. (See Diagram) I have two parallel loops concepted out and was trying to learn which would keep the flow rate at its highest. Option A I would y split and y merge the flow into and out of both radiators. Option B I keep the flows separate off each radiator and run a tube, of exact same length, from each radiator to each graphics card.
3. (See Diagram) I’ve read a lot about tube size and changing tube size to keep flow rates up if you split your initial flow from the pump. Check the diagram for reference. I was thinking if I use 12/16 mm tubing for the tubes leading up and into the radiator and then 10/12 mm on the outputs of each radiator. I was hoping this would essentially slow the flow rate in the radiators giving the liquid more time to cool, but still maintain high flow rate by using the 10/12mm tube to restrict flow. Note: I wasn't sure if I would have to use the 12/16mm tubing or 10/12mm after I split the flow from the pump to achieve a slower flow rate in the radiator.

Thanks for your time and energy! Once I get this beast ordered and put together I will for sure be sharing baby pictures

From my experience with parallel loops, it does not lower temperatures, also you have it backwards when it comes decreasing the tube diameter. if you lower the diameter of the tube coming out of the radiator, you will increase the flow rate( think putting your finger in front of a hose). You will increase both head pressure and flow rate, and it will equalize once everything is at equilibrium. so if you decrease after radiator, and then increase before, you are basically cancelling it out(since it is in parallel). If you want to change flow rate, it has to be in serial. I may be wrong, but from my experience, you will not see any noticeable difference in temps, with that concept. Main(and mostly only way) to lower temps in more CFM Fans, and more surface area(More radiators) when it comes to transfer of heat from the liquid to heat spreader of the radiator, you have to remember that the liquid is always in contact with the radiators, so it doesn't really matter if its flowing faster or slower. all that matters to get closest to room temp is fan CFM and surface area. So if you only have 1 radiator per gpu, it will be hotter than if you had 2 gpus with 2 radiators, you drawings does not really show clearly if its 1 gpu/radiator or if your going parallel only from the pump to the gpus back into a Y fitting into the 2 radiators. hope this helps.

 

[CMart]

From my experience with parallel loops, it does not lower temperatures, also you have it backwards when it comes decreasing the tube diameter. if you lower the diameter of the tube coming out of the radiator, you will increase the flow rate( think putting your finger in front of a hose). You will increase both head pressure and flow rate, and it will equalize once everything is at equilibrium. so if you decrease after radiator, and then increase before, you are basically cancelling it out(since it is in parallel). If you want to change flow rate, it has to be in serial. I may be wrong, but from my experience, you will not see any noticeable difference in temps, with that concept. Main(and mostly only way) to lower temps in more CFM Fans, and more surface area(More radiators) when it comes to transfer of heat from the liquid to heat spreader of the radiator, you have to remember that the liquid is always in contact with the radiators, so it doesn't really matter if its flowing faster or slower. all that matters to get closest to room temp is fan CFM and surface area. hope this helps.

Ok ty for the reply! I understand now. I will probably run a serial on the radiators but still probably do the graphics cards in parallel.

 

[CMart]

so the surface area of the XE 60mm radiator would probably be a better use of that space then a smaller radiator run in parallel? I was thinking just the size difference from the PE 38 mm to the XE 60 mm adds like 170 mL and that alone makes me favoring serial if that is the case.

 

[rts2468]

so the surface area of the XE 60mm radiator would probably be a better use of that space then a smaller radiator run in parallel? I was thinking just the size difference from the PE 38 mm to the XE 60 mm adds like 170 mL and that alone makes me favoring serial if that is the case.

This is where it gets complicated. as well as your preference.
Q1: Does fan noise a concern?
Q2, does your case support 60mm radiators (w/ push/pull config)
You need to answer those questions, for me to be able to help. Because if you case can support the 60mm push/pull, that will work best, but if it does not, then the 38mm in push/pull will out perform the pull or pull of the 60mm, unless the push or pull fans have the same CFM as the push/pull of the 38mm,, then the single push OR pull will out perform the 38mm, if that makes any sense.

 

[GTXJackBauer]

Just run it all one loop as those dual D5s will surely handle it.

As for rads, I'd go with their XE 360 P/P (6 fan) configuration since it's 16 FPI as opposed to the 19 FPI PEs.

Fans, I'd go with their Vardar's F4-120s or F3-120s unless you're interested in their RGB version, either or should be fine. I should also note, it's the static pressure of the fans that should be focused on and not based on CFM. CFM is more for case air flow. Static Pressure is for pushing through things like radiators.

When it comes to Serial or Parallel GPU configurations, dependent on the direction of flow, in serial, the first GPU to be hit will have a lower temp than the next connected GPU in the loop. In parallel, the temps will be slightly higher but both GPU's will be closer in temps. The loop will be least restrictive in a parallel GPU config.

Yes, the loop's temps will reach a equilibrium so no point in splitting things up. As long as you have a efficient running loop roughly in the 1.0-1.5 GPM, you'll be fine. All that needs to be done after is to tweak the rad fans to your liking via temps and audibles, assuming you have the right amount of flow and rad.

Tube size doesn't not matter since it's up to the pumps how much flow will go through the system. Pick the tubing size that best works for the rig.

One more thing, if that's the price you're going to pay for each of those two MSI 1080s, I'd would completely hold off for the next gen GPUs. They're basically a few months away (rumored) if you're willing to wait that long otherwise, I'd look else where for a GPU + WB or just put one yourself together. At those ridiculous scalped prices, I would rather purchase the GPU and WB separate from each other as you could literally put 3 GPUs and WBs together and still have some funds leftover.

 

[rts2468]

Just run it all one loop as those dual D5s will surely handle it.

As for rads, I'd go with their XE 360 P/P (6 fan) configuration since it's 16 FPI as opposed to the 19 FPI PEs.

Fans, I'd go with their Vardar's F4-120s or F3-120s unless you're interested in their RGB version, either or should be fine. I should also note, it's the static pressure of the fans that should be focused on and not based on CFM. CFM is more for case air flow. Static Pressure is for pushing through things like radiators.

When it comes to Serial or Parallel GPU configurations, dependent on the direction of flow, in serial, the first GPU to be hit will have a lower temp than the next connected GPU in the loop. In parallel, the temps will be a bit higher but both GPU's will be closer in temps. The loop will be least resistant in a parallel GPU config.

Yes, the loops temps will reach a equilibrium so no point in splitting things up. As long as you have a efficient running loop roughly in the 1.0-1.5 GPM, you'll be fine. All that needs to be done after is to tweak the rad fans to your liking via temps and audibles.

LOL, different words, same meaning.

 

[rts2468]

Just run it all one loop as those dual D5s will surely handle it.

As for rads, I'd go with their XE 360 P/P (6 fan) configuration since it's 16 FPI as opposed to the 19 FPI PEs.

Fans, I'd go with their Vardar's F4-120s or F3-120s unless you're interested in their RGB version, either or should be fine. I should also note, it's the static pressure of the fans that should be focused on and not based on CFM. CFM is more for case air flow. Static Pressure is for pushing through things like radiators.

When it comes to Serial or Parallel GPU configurations, dependent on the direction of flow, in serial, the first GPU to be hit will have a lower temp than the next connected GPU in the loop. In parallel, the temps will be slightly higher but both GPU's will be closer in temps. The loop will be least restrictive in a parallel GPU config.

Yes, the loop's temps will reach a equilibrium so no point in splitting things up. As long as you have a efficient running loop roughly in the 1.0-1.5 GPM, you'll be fine. All that needs to be done after is to tweak the rad fans to your liking via temps and audibles, assuming you have the right amount of flow and rad.

Tube size doesn't not matter since it's up to the pumps how much flow will go through the system. Pick the tubing size that best works for the rig.

One more thing, if that's the price you're going to pay for each of those two MSI 1080s, I'd would completely hold off on the next gen GPUs. They're basically a few months away (rumored) if you're willing to wait that long otherwise, I'd look else where for a GPU + WB or just put one yourself together. At those ridiculous scalped prices, I would rather purchase the GPU and WB separate from each other as you could literally put 3 GPUs and WBs together and still have some funds leftover.

You are right about Static pressure, its about how much air you move through the radiator, and CFM is cubic feet/minute, so sorta the same thing. but you are correct.

 

[CMart]

This is where it gets complicated. as well as your preference.
Q1: Does fan noise a concern?
Q2, does your case support 60mm radiators (w/ push/pull config)
You need to answer those questions, for me to be able to help. Because if you case can support the 60mm push/pull, that will work best, but if it does not, then the 38mm in push/pull will out perform the pull or pull of the 60mm, unless the push or pull fans have the same CFM as the push/pull of the 38mm,, then the single push OR pull will out perform the 38mm, if that makes any sense.

Q1: Not really since i usually have headphones on. I was gonna go for the vardar high rpms and then use a fan control to adjust as needed if they are too loud.

Q2: I have a Core X5 thermaltake. I did all the measuring im actually putting a lot in this computer this is just 1 of the loops. The problem is it can only fit one on the top that is over the cpu the graphics cards, since mounted vertically, are too tall for the XE.

Thanks for the info!

- - - Updated - - -

@GTXJackBauer
Thanks for info. I already have the graphics cards and all the hardware. Just moving it into a more suitable case. Thanks for the info though I will look for the graphics card drop.

 

[Cnaydmancangt]

Vardars are 1800rpm, thats noisy for any fan, headphones you can still hear that unless you have $300 noise canceling headphones like i do. TBH I had the same choice between vardar and the thermaltake riing plus fans which i already had and i opted to stick with thermaltakes, because they are much lower in rpm.. 1400-1500rpm.. There is a considerdable difference in noise.. and yes 3-4db is a big difference.. the RGB on the new vardars is also a 256 color limited design... Thermaltake you can control each LED.. The differences in cooling isnt too much few C both are high static air pressure fans. I dont have time to constantly manually changing fan speeds. I set them to PWM and be done with it.. And i dont want my fans going to 1800+ rpm.. sure you can reduce that on the vardar fans.. but they are still ugly as sin fans their RGB is even uglier

Also SLI has always been a joke.. half the games dont support it.. and its something you hear less and less and less as the years go on.. Id skip getting two 1080s and get a 1080ti. Unless you already have them ?

Thats just me TBH

 

[CMart]

Vardars are 1800rpm, thats noisy for any fan, headphones you can still hear that unless you have $300 noise canceling headphones like i do. TBH I had the same choice between vardar and the thermaltake riing plus fans which i already had and i opted to stick with thermaltakes, because they are much lower in rpm.. 1400-1500rpm.. There is a considerdable difference in noise.. and yes 3-4db is a big difference.. the RGB on the new vardars is also a 256 color limited design... Thermaltake you can control each LED.. The differences in cooling isnt too much few C both are high static air pressure fans. I dont have time to constantly manually changing fan speeds. I set them to PWM and be done with it.. And i dont want my fans going to 1800+ rpm.. sure you can reduce that on the vardar fans.. but they are still ugly as sin fans their RGB is even uglier

Also SLI has always been a joke.. half the games dont support it.. and its something you hear less and less and less as the years go on.. Id skip getting two 1080s and get a 1080ti. Unless you already have them ?

Thats just me TBH

Something to think about! I noticed they took a lot of the lower RPM vardar fans down on EKWB so im starting to debate changing it up. Im building more for performance not for looks but I can understand a purdy glow when i see one.

As far as the graphics cards, I am actually running dual monitors. I do a lot of design work on this rig and the extra monitor helps. I am debating getting the 1080ti but don't really think the performance increase is worth buying 2 new cards for.

 

[maxfly]

if you want performance fans check out noctua or be quiet! fans. as far as static pressure goes they are among the best you can get. they are expensive at $20-$30 per fan.
https://www.amazon.com/NF-F12-iPPC-...D=51YcmwunkVL&preST=_SY300_QL70_&dpSrc=detail
https://www.newegg.com/Product/Prod...MIkKG3nbCS3AIVBLXACh3UXQ4VEAYYASABEgL54fD_BwE
if you already have a pair of 1080s dont waste your money on another pair of cards. your already at the top of the existing performance scale. youll be hard pressed to resell and get anything close to what you paid for your 1080s. if you absolutely have to have new cards wait for nvidias next gen cards.

 

[CMart]

if you want performance fans check out noctua or be quiet! fans. as far as static pressure goes they are among the best you can get. they are expensive at $20-$30 per fan.
https://www.amazon.com/NF-F12-iPPC-...D=51YcmwunkVL&preST=_SY300_QL70_&dpSrc=detail
https://www.newegg.com/Product/Prod...MIkKG3nbCS3AIVBLXACh3UXQ4VEAYYASABEgL54fD_BwE
if you already have a pair of 1080s dont waste your money on another pair of cards. your already at the top of the existing performance scale. youll be hard pressed to resell and get anything close to what you paid for your 1080s. if you absolutely have to have new cards wait for nvidias next gen cards.

Hmm just check out those fans the Noctua fan looks pretty tempting. How does the Noctua SSO2 compare against a vardar at the same rpms? I just did some more research and I found that there is a new Vardar fan that has stop-start tech that will turn off and on fans as needed. Wasn't sure how viable this is and if it is a lot of work to setup the PWM with that new start-stop tech. Also check some top fan reviews and the Silencio FP120 PWM by Cooler Master seems to be a top fan by far for 2017. Not sure how much has changed in 2018 but the specs on those fans are gonna be hard to beat. 4.8 mm SP but only 27 dBA that seems unbeatable.

Silencio FP120 PWM by Cooler Master Specs:
Airflow: 76 m³/h
Static pressure: 4.8 mm/H₂O
Noise: 27 dBA
Pins: 4-pin PWM

New Vardar ER fans review
http://www.overclock.net/forum/1806...eases-new-vardar-evo-fans-0db-technology.html

2017 Best Radiator fans:
https://www.consumermentor.com/best-radiator-fans/

 

[Silver Surfer]

I agree with the comments regarding not enough temperature difference to go with the parallel and the biggest negative I see in your diagram is the Y to the radiators, because fluid will always take the path of less resistance. You can have two identical radiators but because of construction imperfection one may flow better than the other so the most fluid will be going through the one with least resistance which means the others flow is slightly compromised. This may come down to splitting hairs or be really bad depending on if there is an unknown solder obstruction in one of the radiators. Which leaves you confidently thinking everything is fine when actually it isn't.

If you are concerned with the temperature difference of the primary and secondary card, as the primary is usually always the hottest, reverse the serial flow and supply coolant from the radiators to the hottest card first.

 

[CMart]

I agree with the comments regarding not enough temperature difference to go with the parallel and the biggest negative I see in your diagram is the Y to the radiators, because fluid will always take the path of less resistance. You can have two identical radiators but because of construction imperfection one may flow better than the other so the most fluid will be going through the one with least resistance which means the others flow is slightly compromised. This may come down to splitting hairs or be really bad depending on if there is an unknown solder obstruction in one of the radiators. Which leaves you confidently thinking everything is fine when actually it isn't.

If you are concerned with the temperature difference of the primary and secondary card, as the primary is usually always the hottest, reverse the serial flow and supply coolant from the radiators to the hottest card first.

What a great though! I didn't think about any unknown blockages that could exist in the radiators out of imperfections. That alone convinces me that parallel is I huge risk and without proper tools to monitor flow I don't think I would be able to setup parallel and have peace of mind that its working properly. I really wanted to get a flow rate monitor to ensure consistent flow for all my loops, but could never really find one that wasn't like a grand and way more advanced than I needed. Does anyone out there use any devices to test their flow rates before they finalize their loops?

 

[GTXJackBauer]

It's not needed but if you want a peace of mind, have a look at this flow meter or for about $10 more roughly, you can grab this and this if you want to display it in front of the case in a single bay.

I used to use the latter but have replaced it with another flow meter since I use another company's eco system of components for monitoring and controlling of the loop.

 

[CMart]

It's not needed but if you want a peace of mind, have a look at this flow meter or for about $10 more roughly, you can grab this and this if you want to display it in front of the case in a single bay.

I used to use the latter but have replaced it with another flow meter since I use another company's eco system of components for monitoring and controlling of the loop.

Great Link! I was looking everywhere for these and couldn't seem to find one that wasn't like industrial grade lol!

 

[godevskii]

Dont know if its worth to mention but you dont need 2 cards to run dual monitors, a single card is capable of running 4 monitors. As some people already said, if you dont have the cards yet, a single 1080ti would be a good option.