from benchmarks to actual systems

[stetson-grey]

Hi there,

First I just want to say passmark is a great site. I use it all the time to try to estimate the relative performance of different CPUs and GPUs. Thanks!

I have a question about translating the benchmark values to actual computer systems performance. I know that benchmarking is hard and translating that to real systems is even harder. I'm not looking for exact formulas but more like general expectations. For example, based on CPU benchmark scores alone, you might expect an i5-7500U (score 5381) based system to outperform an i5-6267U (score 4882) based system. However if the former is built with 1866MHz DDR3 memory and the latter with 2133MHz DDR3 memory, would expectations change?

Ideally you could run the benchmark s/w on both systems and compare, but that is not always possible.

Thanks.

[David (PassMark)]

Some of the CPUs tests use a fair amount of RAM. Meaning the RAM (and CPU cache) has some impact on the CPU results. Memory speed doesn't have a huge impact on system performance. A 30% increase in memory speed might result in a 5% increase in overall speed.

So yes, I would expect the i5-7500U to be slightly faster than the i5-6267U, all else being equal. But the CPU results are close enough that other factors, like the type of HDD, SSD used will have a much bigger impact on how fast the overall system feels.

FInally you can see overall results by loading up 'baseline' files in the PerformanceTest software.