Clock speed split in /proc/cpuinfo

February 6, 2011 at 23:10:00
Specs: Linux powerpc, 1.33 GHz/512 MB RAM
Before asking my two questions, I want to clarify the i686 computer I'm writing this on works properly, and that I have been coding on and using Linux for a while, and understand it clearly.

The anomaly lies within an iBook G4 1.33 GHz 7477A Altivec model. The board make is a PowerBook6,5 which is correct for the model (and it has NOT been messed with--the 6,5 *is* the right board); however, within an app that I wrote to tell me basic system information about the computer that I allowed the powerpc to run (memory, processor, computer name, OS name, and logged in user) it reports the clock speed at the mysterious speed of 666 MHz--which is, logically, 1332 MHz divided precisely by 2. Upon further investigating this in the shell, it gets weirder, as *all* results divide 1332 by 2. There's no other logical explanation for why all of the outputs report 666 MHz.

I'm NOT the ONLY iBook or PowerBook user that has had problems with /proc/cpuinfo, where the clock speed is cut down the middle. Is the G4 an early dual-core processor? This can't be, because my Apple is a single-core, so was it a concealed dual-core that was implemented with the Altivec instructions? The G4, oddly enough, was classed by Apple back in the day as a "supercomputer".

What do you think is causing or causes this funky anomaly with grepping /proc/cpuinfo (this is where my app derives the processor information)?


See More: Clock speed split in /proc/cpuinfo

Report •

#1
February 7, 2011 at 22:40:42
My guess would be some sort of frequency scaling, a common practice in laptops. I wasn't able to find much information on it, but according to this page (its for the 12" model), the processor supports frequency scaling and one of the steps is in fact 666 MHz.

Report •

#2
February 8, 2011 at 09:08:09
Thank you for replying! I honestly appreciate it. I actually thought this question was odd enough no one would touch it. :D

Yes, I also believe it to be processor scaling as well, because of the way that the Altivec 7447 series was built. I've programmed for a while, and understand that it's possible that Linux could be looking at the speed in error, as we both know that the G4 took registers in "gulps", supposedly to make it quicker. While waiting, I actually did quite a bit of research from Apple, IBM, and G4 articles to research this.

Strangely, however, this has never happened on an iBook G3--it would simply report 450 MHz, or whatever, flat. Same with working on i586 and i686 machines, so... this is a phenomenon that I think I've finally solved. :)

Thanks for your assistance.


Report •
Related Solutions


Ask Question