So, does this mean trying to do clock for clock comparisons with AMD is/was completely inaccurate, and by Intel's design can never be accurate?
Phoronix: Intel Publishes New Linux P-State Driver
Dirk Brandewie has published a new P-state driver for Intel Core CPUs on Linux. This new cpufreq performance state scaling driver initially is supporting just Sandy Bridge processors but will be expanded to handle other Intel hardware...
http://www.phoronix.com/vr.php?view=MTI5Mzc
So, does this mean trying to do clock for clock comparisons with AMD is/was completely inaccurate, and by Intel's design can never be accurate?
This is the new reality for most vendors. The downside of maximizing performance within a power or thermal envelope is that it makes benchmarking much more of a challenge because things like ambient temperature and heat accumulation in the system from previous test runs starts to have a big effect on clocks and voltages. The upside, of course, is that real world performance can go up without drawing any more power.
This should make for some interesting conspiracy theories![]()