CPU Frequency, Power, and Temperatures
The information below is taken from HWInfo64, which takes the information from the motherboard and CPU sensors. A 10 minutes loop of Cinebench R23 was used to load the system.
In this test, the GIGABYTE Z690 GAMING X DDR4 and the Core i9 12900K never exceeded Intel’s default power limits. However, it exceeds them slightly using GIGABYTE’s own default power limits. What’s interesting is that the increase in power resulted in almost no change in CPU temperatures despite a 15w difference in power consumption.
Under load, both configurations hit 99 to 101c. The latter of which is the thermal throttling limit. However, even under Intel’s only slightly more conservative values, there isn’t much difference in temperature or overall performance as we’ve seen. There are some rare cases favoring the Intel values but the vast majority show either no improvement or a significant improvement for that extra 15w of power.
Here we can see the result of the different power values on our clock speeds. As you can see from the above graph, using GIGABYTE’s default power limits results in a 100MHz increase in maximum clock speed but results in a lower average speed. Interestingly, this did result in greater performance in some cases or did nothing in a lot of other cases. Only one or two benefited from Intel’s more conservative values.
However, using Intel’s default power limits gained us a better average clock speed even if the maximum clocks were lower. In both cases, we were primarily limited by the AIO cooler more than anything. There is more performance that can probably be squeezed out of this setup just through some more refined tuning and improving the cooling system.