Chicago
Track Your Order
0
It's not just clock speed: Here's what makes one CPU better than another
When you're comparing two of the best CPUs, the first spec that comes to mind is clock speed. It quite literally represents how fast the CPU can churn through clock cycles, and naturally, you'd assume a higher clock speed means higher performance. Today, however, clock speed doesn't say much. It can still represent performance differences between two CPUs, but when choosing a processor for your rig, there are several other considerations you should keep in mind.
What clock speed says about your CPU
And what it doesn't say
First, a definition. The clock speed of your CPU defines how many cycles it can complete per second. It's just a frequency, not dissimilar from an audio frequency, and that's important to keep in mind. Clock speed doesn't say anything about the number of instructions that the CPU can execute in a second. It just concerns the number of clock cycles that can be completed in a second.
The more important number is instructions per clock, or IPC. Newer architectures can either support more instructions per clock cycle, or they execute them more efficiently. A great example of that is the branch predictor in Zen 5 CPUs like the Ryzen 9 9950X, which is a two-ahead branch predictor that can speed up execution by prefetching data.
Here's the bottom line, though: Clock speed is how many cycles are completed in a second, not how many instructions are completed.
Because of that, comparing the clock speeds of CPUs from different brands or generations doesn't say a lot. You can generally expect a newer processor with a higher clock speed to perform better than an older processor with a slower speed, but that's largely a result of architectural improvements, not clock speed improvements. Further, the applications that leverage your CPU may be more or less sensitive to clock speed, which is something you should consider when looking at two CPUs.
Decades ago, when single-core processors were all that was available, clock speed ruled the roost. Only a single core was executing instructions, so if that core could complete more cycles in a second, that translated directly to extra performance. As CPU designs have pushed toward additional cores and threads, applications have followed suit. You'll find that most applications today care more about having a certain number of cores rather than how fast those cores are running.
Going beyond clock speed for performance gains
OK, so what else is there?
I've already touched on a few differentiators when looking at two CPUs, from core count to architecture. Third-party reviews are always the best when looking at CPUs -- we have those here at XDA if you're interested -- and you can use standardized benchmarks like Geekbench and Cinebench to see how two CPUs compare to each other. Going purely on specs, here's what you should look at:
- Architecture -- Newer architectures almost always offer better performance than older architectures.
- Core count and configuration -- How many cores do the CPUs have? Are they the same core design, or are they split between different core types?
- Cache -- More cache means less time fetching data from system memory.
- Instruction extensions -- Some workloads require specialized instructions.
- Connectivity -- A motherboard chipset gives you some connectivity, but your CPU largely determines your USB and PCIe options.
Let's get some easy ones out of the way first. Architecture and connectivity are two things you don't need to dive too deep into. Newer is almost universally better with both. The only major thing to pay attention to with connectivity is USB and PCIe support. If you want Thunderbolt 4, for example, Intel's Core Ultra 9 285K supports that feature natively. Meanwhile, AMD's Ryzen 9 9950X doesn't natively support Thunderbolt 4. You'll need to get it through the motherboard's chipset. Although PCIe is backwards compatible, you'll want to look for the latest version if you want bleeding-edge storage capabilities. The number of PCIe lanes supported can be important, too, but only if you plan on connecting several PCIe devices to your motherboard.
Elsewhere, core count and configuration can represent some major differences between two processors, and not always in the way you think. Generally, more cores is better, but you need to keep the core configuration in mind. For example, the Ryzen 7 9800X3D with eight cores actually outperforms the 16-core Ryzen 9 9950X3D in games by a hair. That's because the Ryzen 7 part uses a single, eight-core core complex die (CCD), while the Ryzen 9 uses two. There's some latency when going between the two CCDs.
Beyond CCDs, Intel has pushed a heterogeneous architecture for the last several generations. CPUs like the Core i9-14900K boast 32 cores, but they're split between high-power cores and high-efficiency cores that are significantly less powerful.
An increasingly important spec for CPUs is cache. This is a small bit of storage that's directly next to the CPU, and it's extremely fast as a result. A large pool of cache means that more data can be stored right next to the CPU, decreasing the number of times the CPU needs to go out to your system memory to fetch data. By comparison, your system memory is orders of magnitude slower than the cache in your CPU. AMD has leveraged additional cache to great effect with its X3D CPUs like the Ryzen 9 9950X3D to enhance gaming performance.
Finally, there are instruction extensions, which are only important if you have a specific workload in mind. The core instructions of most CPUs are based on the x86 instruction set architecture, but Intel and AMD extend the capabilities of their chips with instruction extensions like AVX. Recently, CPUs that are a decade old (or more) ran into issues in Helldivers 2, as that game requires a CPU with AVX2 instructions. Similarly, AMD supports AVX-512 instructions with a native 512-bit data path on Zen 5 CPUs, while Intel does not. These instructions can speed up workloads like PS3 emulation and some AI tasks.
You shouldn't rely solely on specs
Only do so at your peril
Looking at specs is great for understanding the capabilities of a CPU, and they serve as a touchstone for comparing two CPUs from the same brand in the same generation. The Ryzen 5 has six cores while the Ryzen 7 has eight, so maybe you should go with the Ryzen 7 for some extra multithreaded grunt. Still, there isn't a single spec that says anything directly about a processor's performance. Reducing a processor down to one spec -- or even a small list of them -- is a quick way to make a bad buying decision.
You need to understand the workloads you want to run, too. For example, additional CPU cache can represent a big boost in gaming performance. Meanwhile, most games don't scale beyond eight cores. On the other hand, tasks like video editing and encoding favor a ton of cores on your CPU. Additional cache on your CPU doesn't do much to speed up these workloads, though. It's important to understand the specs of a CPU, but it's even more important to understand how to apply those specs to applications you'll actually use.
At the end of the day, the best way to find out how one processor performs compared to another is to look at reviews. That way, you can get your head out of the specs and see how they translate into real-world performance.