Pc's thermal performance is very important, this point I believe everyone can understand, and to test the THERMAL performance of the PC there are many ways, the most common is to run a variety of copying software, and then through some monitoring software to see the CPU, GPU and motherboard temperature, you can intuitively see the PC's cooling ability. Such a heat dissipation test method is simple to operate, it can be said that there is no technical threshold, each player can run on their own PC, from the perspective of personal use, it is still very good to use.
However, with our hardware evaluation for more than ten years, whether such a test method is reliable, we are more and more skeptical, and there are always some doubts in our minds:
- Why is the temperature difference between different motherboards running out of the same CPU so much?
- Why is the temperature of different CPUs running out of the same board motherboard different?
- Why can the same type of motherboard run out of the same CPU at different temperatures?
We have also been asked more than once by netizens, why are they all similar hardware, and the heat dissipation efficiency he tested will be completely different from the results of our test? With the increasing power consumption and heat generation of the new generation of CPUs and GPUs, the paper results of the heat dissipation test are becoming more and more ugly, and such problems are increasingly appearing in our evaluations. We did a series of studies on this.
Test platform settings
Judging from the current stage of research, the results obtained by such a test are really not very reliable. We may wish to look at three experiments, namely the same motherboard with multiple CPU, the same CPU with different styles of motherboards and the same CPU with the same model of multiple motherboards, each test uses the same CPU heatsink, to see what is the difference between the temperature and power consumption test results.
In this test, we used three Core i9-12900K processors, as well as Asus ROG Maximus Z690 HERO, Gigabyte Z690 AERO G DDR4, MSI MEG Z690 ACE and other three motherboards, of which MSI MEG Z690 ACE used 2 (BIOS version and settings are the same), the heatsink uses Abee APEX Plus i360 AIO water cooling, the amount of silicone grease used and the application method during the test is maintained as much as possible. The copier project used for the test is the familiar AIDA64 FPU stability test.
Test Results: Replacing the motherboard or CPU will change the test results
The same motherboard is paired with multiple CPUs of the same model
First of all, we look at the same motherboard with the same model of different CPUs, the CPU temperature and power consumption scores obtained, you can see that even if the same type of CPU is tested on the same motherboard, their temperature and power consumption are somewhat different, the temperature is higher and often the power consumption is higher, and from a digital point of view, the difference in power consumption of different CPUs will be more obvious.
Different motherboards are paired with different CPUs of the same model
If the variables of the motherboard are also integrated into the test, then the test results are even more "confusing", because on another motherboard, the temperature and power consumption of different CPUs may have greater differences, which is obviously not worthy of cross-platform comparison.
The same CPU is paired with two motherboards of the same model
It is worth mentioning that replacing the motherboard will cause changes in CPU temperature and power consumption, which is not only effective for different types of motherboards, but also for motherboards of the same model. In the same test environment, just by switching to the same model of motherboard, the temperature difference of the CPU has reached 3 ° C, which is placed on the heatsink test, which is already a grade difference, which will obviously bring additional errors to the evaluation.
Conclusion: The actual platform tests the radiator, and the score is unique
Based on the above test results, it is not difficult to conclude that even the same CPU has different temperature and power consumption performance between each other, and the same CPU on different motherboards, even the same motherboard, will have different temperature and power performance.
This is enough to show that all the thermal test results done using the actual hardware platform are unique, and the results are only valid for the current platform, and cannot be used for cross-platform comparison, even if another set of platforms uses the same hardware. Therefore, the radiator test results made by others, in most cases we can only look at it, his results only exist on that set of platforms, and there is not much reference value for our own platform.
We built a dedicated CPU heating simulator for radiator testing several years ago
So how to test the radiator in the end to be reliable? That's why we started using a specially built CPU heating simulator a few years ago to do radiator testing, including radiator testing and chassis heat dissipation testing, all in order to ensure that the same power consumption of the same motherboard and the same CPU is reproduced, the variables are controlled to a minimum, the product performance can be more realistically displayed, and there is a fair comparison environment between different products.
It's just that ordinary players may not have such conditions, so if you want to compare the performance of different heat sinks, you have to use the same motherboard and the same CPU, and even to lock the CPU on the fixed power consumption and then test, although this practice can not eliminate the temperature and power consumption differences caused by different CPUs or different motherboards, but the error can be greatly reduced, more reflect the performance difference of the heatsink, and the test results obtained on such a platform are more referenced.