What is the speed of the CPU meausured in?

Answer 1

There is no one measurement of CPU speed. Different CPU's may perform differently on various tasks. For instance, you might find something like this:

CPU 'A' runs program 1 in 5 seconds

CPU 'A' runs program 2 in 4 seconds

CPU 'B' runs program 1 in 4 seconds

CPU 'B' runs program 2 in 7 seconds.

It may appear that CPU 'A' is faster over all in this instance... But it may be the case that program '1' is run 100 times more often than program '2' which would make CPU 'B' the faster one.

There are some very complicated ways of actually figuring out which is faster, but that would require looking deeply into its architecture. And the problems your CPU will be working on. The best way to get an idea about what processor is actually faster is to look at sites that offer processor benchmarks such as TomsHardware.com. Otherwise you will be looking into MIPS (Millions of instructions per second.) along with GHz, and today you will even need to see how much a parallel architecture will speed up your program.

Don't let anyone tell you there is one simple number that will tell you how fast a computer is. Many people have tried to use GHz as a measure in the past, but this just isn't valid on its own. If you have 20 pipelines stages, your throughput will be lower for the same clockrate. The only time clock rates can be compared is between processors with the same architecture. (And I don't just mean x86 here!) They need to be developed with the same caches, branch prediction etc.

Answer 2

The speed of a CPU can be measured in a number of different ways.

One way to measure the speed of a CPU is according to the clock speed of the computer. This is not a really accurate measurement because the CPU may use varying numbers of clock cycles to perform varying operations.

Another way to measure the CPU is in terms of operations per second, in OPS.

Another way to measure the CPU is according to the time it takes to perform some particular task, for example, how long it takes to multiply to perform floating point operations, which could be measured in floating point operations per second, or FLOPS.