Processor Speed comparison

komodosp

This might sound like a bit of a beginner question but...

In the old days you could tell how good a processor was from just its clock speed - e.g. 100 MHz was faster than 66MHz and that was nearly all you needed to know.

But now it seems there's a lot more to take into account than the Hz - Quad core, dual core and all this - an Intel 2.4 GHz might be slower or faster than an AMD 2.4GHz, so is there any simple way of comparing each processor with the next?

I take it you can't just mulitply the speed for a "dual core" by 2 to get the effective power? or a "quad core" by 4... ?

Jonathan

It all depends on how the application is written.

A single threaded app will take roughly the same time to execute a task on a 2.4GHz dual core compared with a 2.4GHz quad core.

A multithreaded app will take roughly twice the time to execute a task on a 2.4GHz dual core compared with a 2.4GHz quad core.

Assuming the same processor architectures etc

leninbenjamin

tbh, the real worth of any computer is the number of instructions it can execute over a given time frame.

clock speed is the most important factor in this by far, however in recent years manufacturers have hit a brick wall as to increasing the clockcylcles, mainly due to heat issues.

increasing the number of cores was obviously the easiest solution to this, but even this is limited, as it greatly depends on the nature of algorithms executed as to whether they can be parallelised, and parallelisation increases overheads due to ensuring the data is consistent across cores.

so they've focused on other aspects of the architecture in recent years, namely the memory hierarchy. Instructions take time to load from memory, and the CPU is idle while it waits. So to get additional performance increases they've done things like increase the cache size (so more instructions can be loaded on the CPU at any one time, reducing the likelihood of having to wait) and increasing the bus frequency (so instructions are loaded faster), and also redesigning how the caches operate themselves to reduce the number of bad reads and writes (which are inevitable). The whole idea is to stream memory to CPU as fast as possible so it can operate on it as efficiently as possible.

so it's mainly these improvements to memory hierarchy that allow the i7 to out perform significantly the core2quads, even though their clock speeds are largely the same.

Hope this helps anyway, it's pretty brief and not entirely accurate.