On 29 Nov, 2004, at 10:37, James Bucanek wrote: > Michael Winter wrote on Monday, November 29, 2004: > On Nov 26, 2004, at 9:06 AM, James Bucanek wrote: >> >>> And I've never heard a compelling argument as to why an idle CPU >>> should be kept idle when you have a process that's ready to run. >> >> The only reasons I've heard for doing this is to artificially slow >> down >> the processor to reduce power use on a portable running off a battery >> or to keep the processor running cooler. I've never had to consider >> doing this, but those are the reasons I've heard others mention. > > But don't processors designed for laptops usually implement some kind > of "low power" mode that runs the processor at a reduced clock speed > to conserve power and reduce heat? It would make a lot more sense to > implement some kernal extension that would simply "down shift" the > processor when running tasks with a lower priority. > > What I'm really saying is this: A problem with laptop power > consumption should be solved for laptops. Don't introduce a whole new > class of resource management mechanisms to be (ab)used where there > isn't any rational need. The change in the clock speed of the CPU is implemented in the hardware, not software, and has nothing to do with resource utilization. It was developed strictly for the power consumption needs of the laptop. The system really does run slower when invoked. The problem with this approach is that it effects EVERYTHING across the board, including the scheduler. It also takes "time" to achieve. Attempting to "downshift" and "upshift" during a single time-slice would introduce massive overhead that would significantly impact overall performance. It is just "context switching" on a different level. The argument for keeping the CPU idle or only allowing a particular process a certain percentage of CPU usage has to do with funding of enterprise-class, time-sharing systems. The issue is quite common in Academia and Service Bureau situatoins. If a "box" costs X dollars, and two or more entities share in the cost of purchasing that box, each is expected to "use" the box in proportion to their share of the expense. If one entity is using the resources of the entire box, from say 9am to noon, and another entity is complaining that their response time has gone to never-never-land, things can get very "dicy" when trying to get folks to spring for a bigger box. T.T.F.N. William H. Magill # Beige G3 - Rev A motherboard - 768 Meg # Flat-panel iMac (2.1) 800MHz - Super Drive - 768 Meg # PWS433a [Alpha 21164 Rev 7.2 (EV56)- 64 Meg]- Tru64 5.1a # XP1000 [Alpha EV6] magill at mcgillsociety.org magill at acm.org magill at mac.com whmagill at gmail.com