When you’re purchasing new PC components, there are a ton of different numbers thrown at you besides price. Memory capacity, bandwidth, wattage, temperatures; it all plays into marketing essentially all parts of a computer, but the one number I’ll always ignore on the spec sheet is clock speed.

It doesn’t matter if it’s a CPU running at 5.6GHz or a GPU boosting past 3,000MHz, that frequency figure tends to grab attention, and it’s easy to assume that a higher clock equals a faster chip. For a long time, that might’ve been the case, but today, that spec has become one of the least meaningful numbers to judge performance by.

What clock speed actually means

Not a whole lot anymore

At its core, clock speed just measures how many cycles per second a processor can execute, with one hertz equaling one cycle per second, and one megahertz equaling a million. Back when all CPUs were based on roughly similar architectures, this number made for a fair comparison across different platforms and even generations, within reason. An Athlon XP running at 2.0GHz was, generally speaking, slower than a Pentium 4 at 3.0GHz.

But that’s not how it works anymore. Modern CPUs and GPUs differ so much architecturally that comparing clock speeds across brands and generations will tell you almost nothing about real-world performance. For example, my current CPU, the Ryzen 7 7800X3D, cannot turbo higher than an i9-12900KS, but it will routinely beat it in gaming workloads thanks to its massive cache and architectural advances. Similarly, GPUs like AMD’s RX 7900 XTX can hit higher core frequencies than NVIDIA’s RTX 4090, but still trail in framerates.

Why architecture beats raw frequency

It all comes down to how it’s all laid out, not how fast it can go

Every clock cycle, a processor executes a certain number of instructions. That number, called IPC (instructions per clock), is what really determines how much work gets done. If two chips run at the same clock speed, but one has higher IPC, it will outperform the other every time.

This is why new architectures can get dramatically faster without increasing their frequency. AMD’s Zen 5 doesn’t necessarily have higher clocks across the board than Zen 4. Apple’s M-series chips are another great example. Their clock speeds don’t skyrocket from one generation to the next, but their efficiency and throughput do.

The same story plays out on GPUs. Architectural design, cache hierarchy, and memory bandwidth have a far bigger impact on gaming performance than how high the core clock can spike. That’s why a GPU with a lower base frequency can still crush another that “boosts higher.” If GHz was all that mattered, you’d have seen that number continually rise well past the 5GHz mark we’ve stabilized at now. The first 5GHz CPU was the AMD FX-9590, and that was over 10 years ago.

Boost clocks aren’t what you think, either

They’re a best case scenario

Even if you focus only on a single product, that “boost clock” listed on the box doesn’t mean what you think it does. It’s not even a guaranteed speed. It should be thought of as more of a best-case-scenario number that represents what you can expect to hit under ideal conditions.

In real gaming workloads, clocks constantly fluctuate based on temperature, voltage, and power limits. CPUs from both Intel and AMD use dynamic algorithms to adjust frequency and voltage on the fly, and GPUs do the same through similar algorithms.

So even if your chip is advertised as being able to hit 5.5GHz, it doesn’t mean it lives there most of the time. What matters is the sustained frequency under load, which depends on cooling, power delivery, and architecture. A cooler and quieter PC that sustains its clocks for longer will almost always beat out a system that spikes high and throttles.

Clock speed still matters sometimes

It’s just not a great metric for buying

There are still a few niche cases where clock speed has a measurable impact. Certain simulation or strategy titles that are heavily single-threaded can benefit from increases in raw frequency. Some productivity software can also benefit, but it might be very slight.

Looking at raw frequency numbers when buying components, though, doesn’t give enough context, and doesn’t give you a good picture of the performance you could expect.

Clock speed just isn’t terribly important anymore

There once was a time when pushing a CPU past 5GHz meant something, but now, architecture and voltage curves matter a lot more than raw frequency. I’ll always find hardcore overclocking cool, but when I’m shopping for components, the clock speed is the last thing I consider.