Fan upgrades - more power!

[Petelock]

Hi all,

I was reading the thread on 10 fans connected to mobo (http://forums.overclockers.co.uk/showthread.php?t=18713287) and saw the OP's configuration images. This got me thinking.

I'm at work, so not able to upload pictures of current set up, but have mocked up a few quick images in excel of my current and proposed upgrades.

Here is the current configuration:

[/URL][/IMG]

This has been giving me acceptable temps on circe 50c on the CPU under gaming load and circa 55c on the GPU (with the GPU noctua fan on 100%) If I run it at 75% I get around 60c.

The revisions I am considering are as follows:

[/URL][/IMG]

Here the CPU rad fans are moving to push/pull, all case fans upgraded to Noctua 120mm/140mm.

The notable change is on the GPU that will be changed to the 2000rpm industrial Noctua's in push pull. I was also wondering whether it may be a possibility to bring the air into the case to help with the cooling under load....

My case is an Enthoo Luxe with both of the two removable bays taken out. Air flow should be pretty decent from the front/bottom, but I am aware that the position of the GPU in PCIE lane 1 (and the top 5.25 inch bay) act as natural barriers between air coming in from the front/bottom and reaching the top rear of the case.

I'm not using a fan controller right now, but have a Sata powered 5 PWN splitter for the case fans that are on the CPU opt header. The CPU header is used for the CPU rad fans (when this moves to push/pull it will likely use a SATA powered splitter for the 4 fans). The two AIO pumps are on system headers running at 100% and the GPU rad fan is on a system header. I have this manually set to 50% when not gaming and up the rpms in MSI command centre when I'm implementing my AB overclock profile for gaming.

I appreciate that airflow is really a science of trial and error with the specific set up etc, but am I making a stupid mistake to change the GPU rad fan air direction?

As always, any other suggestions are welcome for alternative solutions to improve flow and efficiency.

 

[Cenedd]

As a general rule, you'll find that faster fan speed equals more noise. Whether that bothers you or not is down to you. You will get better cooling from more air flow - but it will be noisier. There's a caveat to this - isn't there always?! - you cannot get more air flow if the fan cannot move the air. By switching the GPU to intake, you will be pulling colder air in from outside the case and that should benefit the GPU temp. The CPU rad will be getting warmer air in as it's already passed through the GPU rad so CPU temps may rise a bit. You may also find that with 5x intake and 3x exhaust that the GPU cannot pull as much air in as the pressure will rise and 'fight back'.
These, however, are speculations. Without hideously complicated fluid dynamics calculations you're only going to know by trying it and monitoring the temps. Worst case, you can spin the GPU fans round again.....and a fan controller could bring back peace and quiet if necessary. Personally I like the Aquaeros for their autonomous control (fan curves), monitoring and alarms....but they're pricey.

 

[Petelock]

As a general rule, you'll find that faster fan speed equals more noise. Whether that bothers you or not is down to you. You will get better cooling from more air flow - but it will be noisier. There's a caveat to this - isn't there always?! - you cannot get more air flow if the fan cannot move the air. By switching the GPU to intake, you will be pulling colder air in from outside the case and that should benefit the GPU temp. The CPU rad will be getting warmer air in as it's already passed through the GPU rad so CPU temps may rise a bit. You may also find that with 5x intake and 3x exhaust that the GPU cannot pull as much air in as the pressure will rise and 'fight back'.
These, however, are speculations. Without hideously complicated fluid dynamics calculations you're only going to know by trying it and monitoring the temps. Worst case, you can spin the GPU fans round again.....and a fan controller could bring back peace and quiet if necessary. Personally I like the Aquaeros for their autonomous control (fan curves), monitoring and alarms....but they're pricey.

Aye, I was accepting that I would experience a db increase by changing up to the 2000rpms on the GPU. One of the reasons why I currently ramp them up for gaming only (with gaming sound it offset's the additional fan noise). I can live with that, but appreciate that for some it's the main objective (silent running).

Alos, appreciate that I would be pulling additional warm air into the case via the GPU rad. That was the item I was a bit on the fence about. I see where you;re coming from on the pressure ratio from 5x in 3 x out. I agree that it really is a trial and error. I guess the main reason for my post was down to whether it's worth the financial outlay for additional Noctua's.

My other idea was to also add the 2000rpm NF-F12 industrials onto the CPU rad. That additional array of 4 further fans running at higher rmp with better airflow (71.7 CFM) 'may' offset the 5/3 in out fan set up?

 

[Disco_P]

From what you have to what you propose you might see a couple of C drop in temps. Maybe.

Push/pull barely affects temps, especially with fans already running at high RPMs. I have it on my front 60mm thick rad with 1000rpm fans and without the push fans installed it doesn't change temps.

Fan orientations is also trial and error. Sometimes it doesn't make a lick of sense. I have the aforementioned 360mm in push pull in the front pulling air in and a 140mm in the back exhausting. I also have a 240mm in the roof. If I swap between pulling air in or pushing it out nothing changes. Component temps and internal air temps stay the same despite the change from 3 in 3 out to 5 in 1 out.

I wouldn't put any more money into more fans. I don't like noctuas anymore - they had their day but now the same performance to noise can be had for a fifth of the price and with a choice of colours.

With your current setup I'd experiment with flipping the GPU fan. Remember that you are using a 120mm rad to dissipate heat from a 250W GPU and a 240mm for a 140W CPU.

 

[Petelock]

With your current setup I'd experiment with flipping the GPU fan. Remember that you are using a 120mm rad to dissipate heat from a 250W GPU and a 240mm for a 140W CPU.

This is a very good point that I had not considered. Sounds like the best first experiment. I needed to open up the case to install my braided cables so will try this at the same time.

 

[Petelock]

So GPU fan flipped to pull air in. Noted about a 2c improvement in temps with the card oc'd to 1512/1949vram. Peaked at 54c under gaming load.

CPU temps remained steady though. I've already ordered 3x 120mm vardar f-4's to go on the rad's. I had a voucher knocking about to use. I'm going to move some of the existing rad Noctuas to replace the generic case fans.

Not expecting much of a change, but it will be interesting to see what the extra rpms may do to the running temps.