Question about what determines CPU Performance

[mwforrest7]

I just read the announcement of the Ryzen chips and how they're more performant than the 9700K despite having lower clock speeds. I am wondering how that works.

For example:

3700X is an 8 core CPU with base clock of 3.6 and a turbo of 4.4ghz.

9700K is also 8 cores, but with a base clock of 3.6 and turbo of 4.9ghz.

Same # of cores, same base clock, but the 9700K has a much higher turbo clock.

I am very curious, how does the 3700X manage to be more performant then? Is it memory related? I've always just assumed # of cores x clock speed was the most important thing.

 

[zir_blazer]

You're missing the entire IPC (Instructions per Clock) thing.

You may want to read things like this and this, which are extremely ancient but pretty much tells you tangentially what you want to know, the rest you can figure it out. Internally, different Processor architectures work very differently, thus clock speed alone means almost nothing if you don't know how much clocks an instruction takes to execute. And that is pretty much an average, which the first link can tell you due to how different optimizations are depending on what a Processor is good at, and the second link provides a table with execution latency for those x86 Processor families, where you will see that numbers are all different (In some instructions it improves, in others it worsenes). Apply that to our modern days, and it is basically the same.

 

[Campy]

In the example you mentioned it probably has a lot to do with the Intel CPU not having hyper threading.

 

[mwforrest7]

Ahhhhh! That explains a lot and makes total sense. Thank you for filling me in on the missing puzzle piece!