CPU Benchmark: Multi-core vs. Single core comparison?

[jayskiko]

In my current role, we get requests from customers complaining about their "slow" single core processors (Pentium 4 - 3.0Ghz; or Pentium M D740 1.73GHz). We've begun (or I have at least) to use these benchmarks to compare the older systems to the dual-core processors in our current systems (Core 2 E8400; Core 2 P8400). While I know many of our applications may not be designed to utilize both cores of the processor, could I get a more 1:1 "rough" performance comparison by dividing the scores of the E8400 or P8400 by two to get a "score per core"?

Hopefully this makes sense...

Thanks.

[David (PassMark)]

You should be able to test this yourself if you have a dual core CPU. Just set the number of processes from the Edit / Preferences window.

From memory an extra core doesn't double the CPU mark figure.

So instead of dividing by 2 for dual cores, you can divide by maybe 1.8.