Revisiting Noctua’s Best Coolers – NHD15S and U12A 2023 Re-Review

Test System & Methodology

Idle tests are recorded after 10 minutes of the system being left alone after boot with no additional background processes or software running. All load tests are performed in Cinebench R2, first using a single speed run to generate the Cinebench R23 score, then a 10-minute loop to determine the overall all-core average load temperature for the relevant cooler.

Clock speeds are left at the default factory settings provided by the CPU and our NZXT Z690 N7 motherboard. Optimised settings lock the CPU P-Cores to 4.9Hz and the E-Cores to 4GHz at a voltage of 1.3v. Fan speeds for the case are set to 30% at all times. Pump RPM is fixed to 100% for liquid coolers, and CPU cooler fans are left to the motherboard “automatic” and “normal” profile unless otherwise stated.

Acoustics are measured from 2ft above the NZXT H7 Flow PC case, using a decibel meter, with all side panels and filters in place on the case and as stated, the case fans are left at 30% (just to the point where they become comfortably audible over ambient), to represent a real-world usage scenario. We are no longer testing coolers on an open-air test bench for this reason.

As of 20/07/2023, all coolers featured on these charts were retested to reflect our improved methodology, and as such, reviews prior to this date can not be fairly compared to the new results. However, we have been periodically retesting older coolers, which is why you may see their performance figures here, and why they may not align with previous test results from the same product in their respective reviews.

Test system:

• NZXT Z690 ATX
• Intel Core i9-12900K @ stock and & 4.9 GHz all P-Core 4 GHz E-Core 1.3v
• Aorus RTX 2080
• 16GB Crucial DDR4 4400 MHz XMP
• 512GB NVMe SSD
• Thermaltake Tough Power 750W
• NZXT H7 Flow RGB Case
• All testing is conducted using NT-H1 thermal paste

We’d like to say a big thank you to Gigabyte, Noctua, Crucial, Intel, Thermaltake, NZXT and more for providing us with the above testing equipment and their ongoing support.

Testing Methodology

• We always use Noctua Noctua NT-H1 thermal paste to make sure testing reveals the efficiency of the tested coolers not the efficiency of the bundled thermal paste
• Cinebench R23 is run for 10 minutes to calculate “load” results
• Multiple readings are taken with the average of the package temperatures calculated for our recorded result
• Fans are left to operate at default PWM profile speeds unless otherwise stated
• For water cooling tests, all pumps have been operated at 12 volts unless otherwise stated
• Ambient temperatures should be between 21-23 degrees in all our tests unless otherwise stated
• Acoustic measurements are taken 60cm vertically away from the CPU cooler
• Stock tests are performed using “out of the box” settings for the CPU
• All coolers were tested under identical settings unless otherwise stated.
• There is approximately a 1-degree Celsius margin of error in our temperature recording software CPUID HW Monitor
• There is approximately a 1.5dBA margin of error with our Benetech GM1351 decibel meter
• In all these graphs we may have a few “reference” results of particular products that do not fit within that category for comparative purposes.

Software Used

• CPUID HWMonitor
• Cinebench R23
• Unigine Superposition