Question 5900x Cooling

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.


I like Clocktuner-2-for-ryzen personally. That and Dram Calculator. But not in default modes. Using them can give you an idea, it's upto you to tailor them to your particular system.

Ryzens aren't Intel. They work differently. As temps go down, performance goes up, whether there's higher boosts or not. The cpu literally can 'think' faster. In broad terms, the higher the temp, the higher the resistance, so the slower data moves through the cpu. My Ryzen gets better single and multi scores in cinebench with an all core 4.28GHz at 62°C than it does with an all core 4.4GHz at 83°C. PBO is disabled simply because there's no use for it, my Ryzen doesn't reach beyond stock limits using Balanced power plan.

The best cpu temps that you can achieve are with a front mounted aio. If used in pull configuration, the rad doesn't act like a diffuser and at @ 1300rpm or less, pull is more efficient than push for most fans, so it's a win-win as the fans are moved that much closer to the gpu, which receives full direct air from the fans.

A couple of fans at rear/top exhaust provides best airflow. For most of what is exhausted will be gpu exhaust, case temps will be slightly higher, but that's not a real concern as long as there is flow.

AIO's need air, not necessarily air Flow, that's predominantly a requirement for air cooled cpus, which with a front mount is moot as you get direct ambient air through the rad. Gpu temps will be slightly higher than otherwise, but unless you are brutalizing that 3070 with 4k outputs, anything lesser isn't going to put a max load on the gpu anyways, other than bench tests.

Using cinebench, be careful of results. As you check voltages and limits, while multi score might go up, single thread can go down, and thats not a good thing. Games tend to use a Master thread, all the other threads working on frames, but the Master ties everything together to ship to the gpu. So undervolting to get better temps might also lower fps, a result of lower single thread ability, even though multi thread went up.

The trick is to find the best balance, regardless of actual temps, where performance is boosted across the board, not necessarily maximized in one area. Cinebench is good for that, just make small changes and see what the tests show, either a positive boost to performance numbers or a negative.
Reactions: Crosslhs82x2