CPU Water Block Flow Direction Test - Does It Matter?!

[Tam3n]

Long story short: I did some changes to my custom water loop, didn't pay attention to the cpu water block flow direction, noticed today (after weeks) that I have the flow going backwards, decided it's a great chance to do some water cooling science!

 

Lets just cut to the results. Ambient during tests 22~23C, other details below:

And 30 mins later:

 

System:

CPU: Intel i7 3970X @ daily 4.6 GHz ~1.37 V

GPU(s): 2x Gigabyte R9 290OC (Water cooled)
Motherboard: Asus P9X79
RAM: Kingston HyperX 32 GB 1866 MHz DDR3
PSU: Corsair AX860i

Water Cooling:

CPU block: Alphacool NexXxoS XP³ Light - Acetal (it does have "IN" carved at the center port and "OUT" at the top port)

GPU blocks: EK-FC R9-290X - Acetal+Nickel (Rev.2.0)

Tim: GC- Extreme

Pump: EK-5D Vario (X- RES 100)

Radiators: Phobya Xtreme NOVA 1080mm (9x Bitfenix Spectre push) + 480 mm (4x Alphacool NB-eLoop push) + AlphaCool UT60 360mm (3x noctua fans push)

 

Notes:

- Only the CPU was loaded, 1st GPU was at idle and 2nd GPU was powered off.

- Prime95 small FFTs (with AVX) was used to load the CPU.

- Pump and fans were run at the max speed.

- AX860i with Corsair Link software was used to log the most of the data, GPUz was used for GPU temp & VRM1 & VRM2 temp.

- Ambient temp was read from a cheap thermometer... It showed around 22 C for the "reversed" test and 22.5 C for the "correct", but is unreliable.

- This was just a quick test to see how much of an effect does the flow direction have on the CPU temps, not conclusive on the matter or system performance.

- To change the flow direction, pump inlet and outlet were swapped, which reverses the direction of the flow in the whole loop, so technically it could have affected the temps in other ways than just the flow inside the CPU block.

- The loop had still small air bubbles in it after swapping the tubes, but nothing that should have impacted the performance.

 

Well after analysing the results myself, I came to the conclusion it doesn't matter a whole lot. Maybe, maybe (if you believe hard enough) 1 C better with the "correct" flow direction, but it's well within the margin of error. 

 

Other things I observed: "MB CPU" temp came down with the "correct" flow direction, but I don't know what that is and it looks to be a bit random anyway. Looks like the GPU VRM1 temp came down ~2 C, probably because the flow divides more easily into the VRM channel inside the GPU water block with this flow direction (or it's just random variation).

 

 

What we learnt? Once again, don't sweat the small details, just add more radiators!

[Lurick]

Where were you a few months ago when I was wondering this! ;-;

Good to know though, I saw minimal change in temp after swapping it around myself although it was a good excuse to redo part of the loop and make it look better  

[W-L]

I just want to mention it can vary depending on the waterblock's design mainly the way the jetplate forces the fluid around the fins, I have seen a difference as much as ~10C difference from reversing the flow on a CPU block on a Supremacy EVO.

[Tam3n]

9 hours ago, W-L said:

I just want to mention it can vary depending on the waterblock's design mainly the way the jetplate forces the fluid around the fins, I have seen a difference as much as ~10C difference from reversing the flow on a CPU block on a Supremacy EVO.

10C holy batman!

 

And I was running this backwards without a measurable difference.