On Mon, Mar 16, 2009 at 02:28:35PM -0600, michael k lang wrote: > Take a look at the slides (around 29): > http://www.cs.utexas.edu/users/cart/arch/beeman.ppt > > Should be able to control individual cores freqs/Cstates. (Unless I'm > reading them wrong.) I believe you're misinterpreting them. The C states of the invididual cores can be controlled, but it's an all or nothing proposition as far as their clocking goes. In C0, the cores will run at the package clock. In C6, the cores will be unclocked. The voltage to individual cores can be cut off completely as power domains are shut down, but voltage scaling can only be carried out at the package level. The only quirk of this is the dynamic overclocking. If all cores but one are in C6, the package can overclock the remaining core while staying within the TDP of the chip. This is only exposed to the OS as a package level P state which gives no indication of the frequency that the core will shift to (and, indeed, this may depend on things like the temperature of the chip). So yes, while you can control the individual core C states, that doesn't mean that you have any real control over the frequency of the cores - they'll either be at 0MHz (in deep C states), the package frequency (corresponding to the current package P state) or a frequency equal to or above the frequency claimed by the maximum package P state (if you're in P0 and all other cores are in C6). -- Matthew Garrett | mjg59@xxxxxxxxxxxxx -- To unsubscribe from this list: send the line "unsubscribe cpufreq" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
