what-is-cpu-clock-speed-does-it-matter


It's no secret that marketers really like numbers whether it's horsepower on a car, screen size on a TV or the number of items on the buffet there's this general idea that people are drawn to large numbers and computer processors are no exception.

If you're shopping for a new system it's pretty easy to find retailers that prominently listed the processors gigahertz there's just one small problem it's basically an utterly meaningless metric these days and has been for at least the last 13 years.

Now we touched on why you shouldn't just go buy raw speed when you're buying a processor in this article but we're gonna go into it in a little more detail today.

First off, speed in megahertz or gigahertz measures how many clock cycles a CPU goes through each second and a cycle is basically what happens when an electrical pulse hits your CPU allowing it to execute instructions and keep its activity in sync with the rest of your system.

1 Hertz = 1 cycle per second

1 MHz = 1 million cycles per second

1 GHz = 1 billion cycles per second 

So this means that a processor rated at 4 gigahertz can handle a billion more cycles per second than one rated at 3.

What's simple enough to understand but things get more complicated when you consider what happens during a clock cycle one big determining factor of a CPUs performance is how many instructions that can execute per clock cycle.

Modern processors can work on more than one instruction at a time - thanks to pipelining which you can think of as being similar to how an electronics factory can turn out a huge number of smart phones per day even though it might take a long time for one phone to make it all the way through the assembly line.

So different stages of production are pipelined to take place simultaneously so that the guy that installs the screen isn't constantly waiting around for the next phone to work on

Pipelining in CPUs is similar in principle but quite a bit more complicated and engineers use different methods to allow the processor to chop up instructions into parts that can be worked on simultaneously or even combine smaller operation to complete them in fewer cycles increasing efficiency.

Exactly how this gets accomplished depends on the specific microarchitecture of the processor which can learn more about up here but bringing things back to the topic at hand it means that because the number of instructions per clock cycle and the pipelining efficiency can very enormous ly between CPU brands and even the individual models from a single brand clock speed can end up being very deceiving

CPUs also need to be able to read ahead on the page so to speak to work efficiently

Modern Intel and AMD CPUs can process around 30 percent more instructions per clock than Sandy Bridge from 2011.

Modern programs aren't all that linear and need to be able to respond to many different user inputs quickly I mean just think about how fast you expect your PC to show you the right thing after firing your weapon after someone else throws a grenade in a high budget video game

So CPUs are designed to examine out of order instructions and make sure that they're executed such that they don't interfere with other parts of the program as well as to anticipate what instructions might be needed next a feature called a Branch Prediction.

CPUs with better branch prediction can have significantly better performance and again this is something you simply won't be able to determine just from looking at the clock speed

And there are many other factors that tie into the performance as well like the type of RAM that is supported the types of instructions that are supported and CPUs can come with different types and amounts of Cache Memory you can learn more about it up here.

But basically it's this super-fast memory that stores small pieces of data that the CPU thinks it will need in a hurry so a processor with a well implemented cache can fetch things it has to work on much more quickly

Finally if you're working on CPU intensive tasks like video encoding or file compression having many cores to spread the workload out and crunch more data in parallel can speed things up more than a clock speed boost

Now don't get me wrong clock speed shouldn't be ignored completely if it was totally irrelevant there wouldn't be a thriving community of overclockers trying to squeeze as many megahertz as they can out of there chips for dem performance gains and it can be a useful indicator of performance as long as you are only using it to compare CPUs with the same microarchitecture features and number of core’s or if you are compensating appropriately for the other differences that exist.

The point is just that it shouldn't be a huge factor in your buying decision just like you probably shouldn't choose a car based on how many cupholders it has.