Review AMD R7 5700X cooled with Scythe Ninja 5 RevB (no fans), Graphite Thermal Pad, and 3 case fans.

[Golgatha]

I have a computer I primarily use as a Plex and file server in my home. I've built it around the ideas of maximum reliability and stability, with a secondary want for it to run on as little electricity as possible while still doing what I want it to. To that end, I made this build run with as few moving parts as possible. Even the graphics card, a Gigabyte 1660 Ti, only spins up its fans when under a substantial load. Also, I've used 2666Mhz ECC RAM and recently I swapped out a Wraith Prism for a very large tower cooler; the Scythe Ninja 5 Rev.b, after viewing a video comparing several HSFs running without any fans attached. On the reliability side, I'm using a graphite thermal pad instead of thermal compound to ensure I never have to worry about the thermal grease migrating, drying out, etc. I did have to source the cooler from eBay, as it was not readily available for purchase Online otherwise. As a bonus, it nicely matches the black and silver aesthetic of my case too.

Fanless PC comparison video -

Cooler used - https://www.amazon.com/Scythe-Cooler-Single-LGA1700-LGA1151/dp/B09XPXC7TG

Case fans are 2x 92mm Noctua NF-B9 pulling in air low and in the front of the case, and 1x 120mm Noctua NF-P12 plus the stock 120mm Seasonic Focus 850w power supply fan exhausting out the top rear of the case. The case is a very old Zalman Z-Machine GT1000 and it is made entirely out of 4mm aluminum panels.

I did 3 consecutive runs of Cinebench R23 to test the cooling effectiveness. Running the system any longer than 30 minutes did not increase the temperature any higher than what was observed during the 3rd run.

Run 1 - Max temp running Cinebench R23 for 10 minutes is 68.5°C, ran around 3.9Ghz all-core, and the score is 12,869.
Run 2 - Max temp running Cinebench R23 for 10 minutes is 69.0°C, ran around 3.9Ghz all-core, and the score is 13,002.
Run 3 - Max temp running Cinebench R23 for 10 minutes is 70.4°C, ran around 3.9Ghz all-core, and the score is 12,979.

For reference, the CPU Idles around 40°C max and the networking equipment plus the computer uses about 82w total power. Power draw under 100% CPU load running Cinebench R23 is around 135w for this same setup.

 

[Brainonska511]

Does a fan on the heatsink really add that much electricity use? I feel like running it fanless could impact your primary goal of reliability.

 

[Golgatha]

Does a fan on the heatsink really add that much electricity use? I feel like running it fanless could impact your primary goal of reliability.

No, it adds almost nothing in terms of electricity usage. Ideally, I'd like the system to keep chugging along even if all the fans went out, and I think it could in it's current configuration. The other part is by changing from the Wraith Prism, I eliminated a point of failure, because that HSF can't properly cool the CPU without the fan being operational. The other thought is that even under 100% load, I'm still like 15°C shy of any type of thermal throttling kicking in. Under normal operation I'm at least 40°C under thermal throttling. So, why put a fan on it when it's not needed? Also, a CPU fan would cause dust to accumulate in the fins of the CPU's heatsink. That's a concern because it runs 24/7/365, dust also collects around the VRMs surrounding the CPU socket, and it's just one more chore to have to clean it regularly.

 

[Brainonska511]

No, it adds almost nothing in terms of electricity usage. Ideally, I'd like the system to keep chugging along even if all the fans went out, and I think it could in it's current configuration. The other part is by changing from the Wraith Prism, I eliminated a point of failure, because that HSF can't properly cool the CPU without the fan being operational. The other thought is that even under 100% load, I'm still like 15°C shy of any type of thermal throttling kicking in. Under normal operation I'm at least 40°C under thermal throttling. So, why put a fan on it when it's not needed? Also, a CPU fan would cause dust to accumulate in the fins of the CPU's heatsink.

Wouldn't having a fan on the heatsink reduce the risk of failures overall? If the one of the system fans went out, you'd still have airflow through the heatsink. And as long as air moves through the case, you're always going to have dust accumulation.

 

[Golgatha]

Wouldn't having a fan on the heatsink reduce the risk of failures overall? If the one of the system fans went out, you'd still have airflow through the heatsink. And as long as air moves through the case, you're always going to have dust accumulation.

Good point about still having airflow if a case fan fails. I do plan to validate the system can run without any fans plugged in later. On the dust accumulation, I've found the dust buildup to be highly localized around the CPU socket and in the fins of the CPU. The other bit is it's super fine dust, as the front fan screens catch the large majority of dust. I use a handheld air blower to clean those screens from time to time. I think the slightly higher CPU temps are worth not having dust buildup around the CPU socket and in the fins. This is more long-term thinking, as my last home file server lasted me about 10 years before I finally upgraded it. I really want to minimize airflow and fans, but I also want everything capable of running at 100% without overheating. Cinebench R23 is just a benchmark, but I do run Handbrake from time to time, and that will get the CPU cooking.

Air Blower - https://www.amazon.com/gp/product/B0B867L89R