Clock Frequencies
The following chart shows how well the processor sustains its clock frequency and which boost clock speeds are achieved at various thread counts. This test uses a custom-coded application that mimics real-life performance—it is not a stress test like Prime95. Modern processors change their clocking behavior depending on the type of load, which is why we provide three plots with classic floating point math, SSE SIMD code, and modern AVX vector instructions. Each of the three test runs calculates the same result using the same algorithm, just with a different CPU instruction set.
The rated 5.7 GHz top speed is easy to reach, but only for a short time, with a light load, on the two best P-Cores. As you can see, even with two threads active, or a single heavy thread, the average is already down a little bit.
The E-Core side of this chart (threads 9 to 24) can be somewhat misleading. It looks like the cores run at increasingly slower frequencies as the load goes up, but that's not the case. The frequency plotted is the average of all the P-Cores active, plus more and more E-Cores, which all run at a lower frequency than the P-Cores, dragging down the average.
That's why I tested E-Core frequency scaling separately in the chart below.
The E-Cores are basically running at 4.6 GHz all the time, unless the processor goes into some kind of throttle state.
This screenshot from the BIOS confirms the clock scaling:
- 1 core active, under 70°C = 5.7 GHz
- 1 core active, over 70°C, or 2 cores active = 5.6 GHz
- 3 or more cores active = 5.4 GHz
Overclocking
Overclocking the Intel Core Ultra 9 285K is pretty easy, thanks to the unlocked multiplier. While temperatures and keeping them away from the throttle point have been a big gotcha with overclocking in the past, this is a complete non-issue now—the processor is easy to keep cool. Intel has also increased the TjMax to 105°C, which adds an extra +5°C in thermal headroom. You can manually increase this limit to 115°C. On AMD AM5 CPUs the limit is 95°C, with no option to increase.
You can now adjust the voltage for the P-Cores and E-Cores separately, and also their overclocking frequencies. On top of that there are additional dials for the inter-tile clock frequency and others.
The OC testing in this review is designed to represent what a typical user can achieve without in-depth knowledge of secret overclocking techniques or dozens of hours to tweak the system for a long time. I set the voltage offsets to +150 mV for both core types and started increasing the CPU multiplier for the P-Cores only.
Here's another change, the frequency steps are much more fine-grained, down to 16 MHz. For the testing in this review I went in steps of 200 MHz first, and then 100 MHz, until the CPU was unstable. At this point (5.6 GHz), I tried to give it another bump in voltage, but I couldn't get it stable without risking damage to the CPU due to too high voltage, so I settled for 5.5 GHz, but that ended up with a little bit of thermal throttling, so I picked 5.4 GHz. This is considerably lower than the various boosts, which reach up to 5.7 GHz, so this all-core overclocking only helps with highly multithreaded workloads. It's important to note that Intel CPUs support per core overclocking, which can help with finding an overclocking profile that doesn't compromise.
Next, I started working on the E-Cores. However, I first lowered the P-Cores' multiplier to ensure that any instability wouldn't disrupt the E-Core overclocking.
E-Core overclocking topped out at 5.0 GHz—impressive, but it was a tiny bit unstable, so I dialed the clocks down another 100 MHz to 4.9 GHz.