To be able to cover the higher TDP classes of processors in the future, I upgraded my testbench from a Ryzen 5 3600 to a Ryzen 9 3950X. However, so that the compact or budget coolers are not completely overtaxed, testing is done in three stages. This article serves as a reference and basis for future comparisons so that tested air and water coolers can be classified according to their cooling performance
Methodology
An AMD Ryzen 9 3950X has to be cooled in various levels of power consumption, joined by 16GB Crucial Ballistix Sport (2x8GB) DDR4 memory on an MSI B550 Tomahawk (BIOS 7C91vAB). The Asus ROG Strix GTX 1060 6GB runs in zero-fan mode for the CPU stress tests only and a Fractal Design Ion Gold 750W serves as the power source. The whole thing is housed in the Thermaltake Core P3 as an open benchtable.
The test system runs on Windows 10 Pro 22H2 (build 19045.2251). All data sets are recorded with HWInfo64 v7.34-4930 – 5000ms logging, the specified temperature results from the sensor CPU Die (Average). Since I don’t have air conditioning, the values in the graphs are the delta between room temperature and CPU temperature. This allows all cooling systems to be compared fairly, even if the ambient temperature is different. The thermal paste included with the cooler or pre-applied is used in each case.
A TDP limit of 100, 150 and 200 watts is set via Ryzen Master software to cover different processor performance classes.
I determine the sound level with a Voltcraft SL-200 meter at a distance of 40cm, which operates in 125ms cycles in Lo mode.
Subsequent results will be supplemented with new coolers as they are tested.
Air cooler
In the following overview, all coolers were set to a loudness of 40 dB(A) at a distance of 40cm from the first fan:
Notes:
- Xilence XC081 could not hold the 200W load, Thermal Throttle reduced the power target to 195W. Result therefore removed.
- be quiet! Pure Rock 2 could not hold the 200W load, Thermal Throttle reduced the power target to 185W. Result therefore removed. Even at maximum speed, only 39.6 dB(A) is achieved instead of 40.0 dB(A).
- Endorfy Spartan 5 Max ARGB could not hold the 200W load, Thermal Throttle reduced the power target to 193W. Result therefore removed. Even at maximum speed, only 36.1 dB(A) is achieved instead of 40.0 dB(A).
- Scythe Fuma 3 Rev. A could not hold the 200W load, Thermal Throttle reduced the power target to 199W. Result therefore removed.
These results would be achieved at the respective maximum speed (100% PWM), thus showing the highest possible cooling performance without considering the noise level:
Water cooler
In the following overview, all coolers were set to a loudness of 40 dB(A) at 40cm distance from the center of the pump and fan on the radiator:
These results would be achieved at the respective maximum speed (100% PWM), thus showing the highest possible cooling performance without considering the noise level:
55 Antworten
Kommentar
Lade neue Kommentare
Urgestein
Urgestein
Moderator
Moderator
Moderator
Veteran
Moderator
Urgestein
Urgestein
Mitglied
Veteran
Urgestein
Veteran
Veteran
Urgestein
Moderator
Moderator
Urgestein
Alle Kommentare lesen unter igor´sLAB Community →