Noise output is also central to a useful cooler evaluation. A loud cooler can perform better at full speed than a quiet one, but many users want effective cooling at a low noise level. We therefore publish noise-normalized results alongside full-speed results, and use THETA to express the measured performance on the named test platform.
We named THETA after the Greek letter theta, the symbol engineers use for thermal resistance. This Cybenetics standard rates a cooler's Overall Performance at 30 dBA normalized noise output.
We define Overall Performance (OP) as OP = W × F / T. W is the average CPU power in watts measured at the EPS connectors by Powenetics v2, F is the average CPU frequency in MHz, and T is the average CPU temperature in °C. We calculate OP from these component averages over the last 4 minutes, rather than averaging instantaneous OP values.
The product of power and frequency divided by temperature rewards a better cooler on both platforms: AMD runs at its 95 °C temperature limit, so the improvement appears as sustained power and clocks, while Intel is power-limited, so the improvement appears as a lower temperature.
We chose 30 dBA because two fans at PHI's 25 dBA reference level each combine to about 28 dBA, while three combine to about 30 dBA. Every cooler we have tested so far has a measured result at this level, or at 25 dBA under the route below for coolers that cannot reach 30 dBA. A performance level is awarded only when that result meets at least the Bronze requirement of the cooler's class.
Air coolers and AIO liquid coolers are rated in two separate classes, each with its own requirements, and each class has one scheme for each test platform: four schemes in all. The badge names the class and the CPU vendor and socket, for example THETA Gold · Air · AMD AM5 or THETA Gold · AIO · Intel LGA1851. The report also names the test CPU: AMD Ryzen 9 7900X or Intel Core Ultra 9 285K, respectively. There is no year on the badge.
A cooler that cannot reach 30 dBA is rated at its own measured 25 dBA point against the 25 dBA requirements. The report states “rated at 25 dBA”. This route requires a measured 25 dBA point; we do not estimate the cooler's performance at 30 dBA. A cooler that cannot run as quietly as 30 dBA receives no performance level.
The 25 dBA requirements express the same performance levels at the lower noise output. We derived them by dividing each 30 dBA threshold by the median OP30/OP25 ratio of the coolers measured at both points when we set these requirements, 1.0140 for AMD (45 coolers) and 1.0212 for Intel (11 coolers), and rounding the result to the nearest 50. These ratios and the listed requirements are fixed for this revision. The ratios set the requirements only; the rating always uses the cooler's own 25 dBA measurement.
An AIO liquid cooler with a large radiator moves considerably more heat at the same noise output than an air cooler, so each cooler is compared with coolers of its own kind. We set the Diamond requirement of each class about 2.5% above the best cooler of that class we had tested when this revision was issued, so that Diamond is a target for the very best coolers. On AMD, the lower levels follow the measured spread of each class.
THETA – Air · AMD AM5 Performance Levels/Requirements – 30 dBA Normalized Noise Output
| Performance Level | Requirement |
|---|---|
| Bronze | ≥9,000 & <9,500 |
| Silver | ≥9,500 & <10,000 |
| Gold | ≥10,000 & <10,250 |
| Platinum | ≥10,250 & <10,500 |
| Titanium | ≥10,500 & <11,000 |
| Diamond | ≥11,000 |
THETA – Air · AMD AM5 Performance Levels/Requirements – 25 dBA Normalized Noise Output
| Performance Level | Requirement |
|---|---|
| Bronze | ≥8,900 & <9,350 |
| Silver | ≥9,350 & <9,850 |
| Gold | ≥9,850 & <10,100 |
| Platinum | ≥10,100 & <10,350 |
| Titanium | ≥10,350 & <10,850 |
| Diamond | ≥10,850 |
THETA – AIO · AMD AM5 Performance Levels/Requirements – 30 dBA Normalized Noise Output
| Performance Level | Requirement |
|---|---|
| Bronze | ≥10,050 & <10,500 |
| Silver | ≥10,500 & <11,000 |
| Gold | ≥11,000 & <11,550 |
| Platinum | ≥11,550 & <11,900 |
| Titanium | ≥11,900 & <12,350 |
| Diamond | ≥12,350 |
THETA – AIO · AMD AM5 Performance Levels/Requirements – 25 dBA Normalized Noise Output
| Performance Level | Requirement |
|---|---|
| Bronze | ≥9,900 & <10,350 |
| Silver | ≥10,350 & <10,850 |
| Gold | ≥10,850 & <11,400 |
| Platinum | ≥11,400 & <11,750 |
| Titanium | ≥11,750 & <12,200 |
| Diamond | ≥12,200 |
THETA – Air · Intel LGA1851 Performance Levels/Requirements – 30 dBA Normalized Noise Output
| Performance Level | Requirement |
|---|---|
| Bronze | ≥12,000 & <12,650 |
| Silver | ≥12,650 & <13,350 |
| Gold | ≥13,350 & <13,700 |
| Platinum | ≥13,700 & <14,050 |
| Titanium | ≥14,050 & <14,350 |
| Diamond | ≥14,350 |
THETA – Air · Intel LGA1851 Performance Levels/Requirements – 25 dBA Normalized Noise Output
| Performance Level | Requirement |
|---|---|
| Bronze | ≥11,750 & <12,400 |
| Silver | ≥12,400 & <13,050 |
| Gold | ≥13,050 & <13,400 |
| Platinum | ≥13,400 & <13,750 |
| Titanium | ≥13,750 & <14,050 |
| Diamond | ≥14,050 |
THETA – AIO · Intel LGA1851 Performance Levels/Requirements – 30 dBA Normalized Noise Output
| Performance Level | Requirement |
|---|---|
| Bronze | ≥13,400 & <14,000 |
| Silver | ≥14,000 & <14,650 |
| Gold | ≥14,650 & <15,400 |
| Platinum | ≥15,400 & <15,850 |
| Titanium | ≥15,850 & <17,150 |
| Diamond | ≥17,150 |
THETA – AIO · Intel LGA1851 Performance Levels/Requirements – 25 dBA Normalized Noise Output
| Performance Level | Requirement |
|---|---|
| Bronze | ≥13,100 & <13,700 |
| Silver | ≥13,700 & <14,350 |
| Gold | ≥14,350 & <15,100 |
| Platinum | ≥15,100 & <15,500 |
| Titanium | ≥15,500 & <16,800 |
| Diamond | ≥16,800 |
On Intel, Diamond follows the same rule as on AMD. We carried the lower Intel requirements across from the AMD requirements with the median ratio of Intel to AMD Overall Performance in each class, 1.336 for air coolers and 1.331 for AIO coolers, from the coolers tested on both platforms. We will review the Intel requirements once 25 coolers have Intel results. Any change will be issued in a new revision.
The requirements belong to the test platform. A rating remains valid for the vendor and socket named on its badge. When we replace a platform, we derive new requirements by testing a set of coolers on both the old and new platforms, so that each performance level keeps its standing.
We retain one control cooler for the life of each platform. We re-test it periodically and after any platform change, including the BIOS, CPU sample, mainboard or thermal paste, and track its results on a control chart.
We award THETA levels by simple acceptance: we compare the measured OP directly with the corresponding requirements. Values below the Bronze requirement receive no performance level.
THETA is a Cybenetics performance rating. It is not part of our ISO/IEC 17065 certification schemes and is outside our accreditation.
Coolers Evaluated Before Revision 2.0
We also assigned THETA levels to the coolers we evaluated before Revision 2.0, from their published results at 30 dBA, or at 25 dBA under the rule above. These coolers were not measured again. Their results come from the procedure of Revision 1.8, in which each noise level ran for 20 minutes and was followed by a 10-minute cool-down, so the earlier procedure is not identical to the current procedure. Their evaluation reports say so next to the THETA level.
With Revision 2.2 (October 2026) we divided THETA into the air and AIO classes, each with new requirements. Every cooler already rated was re-rated with the new requirements from its existing measurements; no cooler was measured again.
The complete method is described in the Cybenetics Cooling Solutions Test Protocol (Revision 2.2).