Home
| Calendar
| Mail Lists
| List Archives
| Desktop SIG
| Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings Linux Cafe | Meeting Notes | Linux Links | Bling | About BLU |
On Tue, Mar 06, 2012 at 04:27:07AM -0500, Shirley M?rquez D?lcey wrote: > The fact that the correct strategy for maximum performance may be > different from the best power management strategy is likely to require > new OS capabilities (changing the scheduling strategy based on power > management settings). If I have four things running on a desktop system > I'd likely prefer to spread them among all four modules for maximum > speed, but on a laptop I might prefer to put them on only two modules > and take the performance hit to get the power reduction of shutting off > the other two modules. I'm not so sure you would gain anything by shutting off modules. It was found with CPU frequency governors that slowing down a processor actually used MORE power, because it took longer to complete running tasks. It is better to have the CPU always run as fast as it can while there are running tasks, and then halt the processor when it is truley idle. If you have enough tasks that could take advantage of multiple modules by being scheduled simultaneously, you may finish those tasks sooner and be able to put all the modules to sleep sooner, saving more power than if you artificially limit yourself to a subset of modules. I guess it remains to be seen and tested if someone implements what you suggest, but in the end it may not be worth it.
BLU is a member of BostonUserGroups | |
We also thank MIT for the use of their facilities. |