Jump to content
  • entries
  • comments
  • views

Follow-up to the Airflow Mod

Mira Yurizaki



This is a follow-up to my blog on the airflow mod I made. It was brought to my attention that the sound card might play a role in affecting how well my video card remains cool. My presumption is that it's not doing much to affect the cooling potential because the issue was moving hot air away from the video card area and that while the video card is sucking air from the rear, there was enough airflow that it wouldn't make much of a difference.


So today I decided to test whether or not the sound card had an impact along with retesting whether or not the airflow mod had an impact. The numbers I got earlier I pondered about and wondered if another factor had something to do with it: the intake from the front and the exhaust in the rear and top.

The Setup

Taking all of these combinations together, I've narrowed down the parameters to:

  • Fan speed of the top and rear of the case (they're both tied to the same fan controller channel)
  • Fan speed of the two front intake fans
  • If NZXT CAM was running or not
  • If EVGA Precision XOC was running or not
  • Whether or not the airflow mod fan was installed
  • Whether or not the sound card was installed

To avoid having a bajillion combinations to test, I've eliminated the following variables by setting them to a default

  • If NZXT CAM is running, both the exhaust and intake fans are at 50%
    • If NZXT CAM is not running, then it's likely the controller keeps them fixed at a lower RPM range. Due to my observation the other day with the new LED fans "blinking", I can tell without NZXT CAM running they're not being varied because no blinking occurs.
  • EVGA Precision XOC will always be running with a custom fan profile

Basically, the only variables I'm testing then are:

  • NZXT CAM is running or not
  • Airflow mod fan is installed or not
  • Sound card is installed or not

The following is the software setup when EVGA Precision XOC and NZXT CAM running:



Also of note, the ambient air temperature in the room was about 68°F or 20°C


What I'll be doing

  • Run the Final Fantasy XIV Stormblood Benchmark three times. Between each run, there will be about a 10 second delay before the next. The logging captures each configuration, but not each test. i.e., the same log file will be used for all three runs.
    • This benchmark is used because this game is a frequent use case. It lasts for about 4-5 minutes per run.
    • I did not want to run something like FurMark because that's not a realistic test and I'm not interested in workloads I won't be subjecting the video card to.
  • Use GPU-Z to log GPU clock speed, temperature, and performance cap reason

And the results

Table Format

(Note that some of the formatting didn't carry over)



Charts of Interest


Average Clock Speed (higher is better)


Maximum Clock Speed (higher is better)


Average GPU temperature (lower is better)


Performance Cap Reason Counts

This one needs explaining. The goal is to have 0 thermal hits. Vmax should not be encountered because I've not allowed the card to push past default voltage limits. Util just means the GPU was not given a hard enough workload. The other two then, Power and Vrel, are for hitting the TDP and not being able to clock faster due to hitting the stock voltage limit. Now the question becomes is one better than the other? And for the purposes of this test, actually, there is! Power delivery circuitry becomes less efficient as it heats up. It's not by much, but when you have hardware that's being pushed to the edge, that "not by much" becomes "actually an appreciable amount" This is why it's important to cool off your VRM if you plan on pushing the overclocks.


tl;dr, better is:

  • 0 Thermal
  • Lower Power
  • Higher Vrel

This chart only has Power and Vrel, as no Thermal reasons were hit and the other two, Vmax and Util, don't matter.




Despite my initial presumption, having no sound card is effective in decreasing the temperature of the video card as much as having the airflow fan mod. So this means there could be more air flow coming into the rear video card fan. Having both the air flow mod and no sound card however, resulted in significantly better cooling. I'm not sure if having NZXT CAM was a factor or not since in some cases it helped a lot and in others it didn't seem to help at all.


Some other conclusions I can make regarding this:

  • microATX cases, at least ones of this size and configuration, are not ideal for having an "open air" style cooler video card if you are planning to use another expansion card. Though this may be mitigated if you don't mind turning your intake fans up higher if you don't want to lose the expansion card or do a airflow mod.
  • Having a mid-case air circulator helps regardless.
  • They make L brackets for case fans!




It's interesting to see that without the sound card and the added fan, performance is significantly better. 


I also have noted how limiting both MicroATX cases and mITX cases with PSU shrouds (or even just close proximity to things below them like a sound card or the PSU) could potentially cause performance  issues  Especially in mITX, but even in many Micro ATX applications where many motherboards are putting the main PCI-E x16 slot in the second slot down instead of the very top. AS a perfect example, Blue Build 2018 uses a mATX case with a 2nd slot PCI-E motherboard layout, and there's a significantly lower amount of room than there would be without a PSU shroud (for one fan) or if the motherboard moved the PCI-E slot upwards. 


I understand WHY motherboard manufacturers do this (away from CPU air cooler, better M.2 placement for cooling, etc) especially on ATX where this airflow problem is (usually) a non issue, and unless you're one of us trying to squeeze every last performance out of your system for your dollar, most users wont notice. But it's also good for science and to be better builders in the future :) 



Link to comment
Link to post