performance of hardware dynamic scheduling
|
From: Bill Todd on 20 Apr 2008 01:25 Stefan Monnier wrote: >>> There is nothing barring us from making a laptop you can work on for >>> a day or so with normal battery cappactities. 6Ah @ 14V is about the >>> limit, or 84Wh. That means we would need to limit power for the system >>> to an average of 8W. > >> I have no idea whether that's feasible or not, but what *does* seem clear is >> that it's not at all 'normal': a quick review of the current under-$1200 >> Thinkpads at the Lenovo site indicates that their standard (6-cell, hence >> presumably nominally 11.1v.) batteries are rated 2.4 Ah, or less than 27 Wh. > > AFAIK, normal battery capacities for laptops is around 50Wh. > It may be as low as 20-30Wh and go as far up as 80-90Wh, but 50Wh seems > to be about average. It turns out that the Lenovo information was misleading: when it characterized its lower-end batteries as 6-cell, 2.4 Ah units, it was referring not to the battery capacity but to the capacity of each cell (so given that it uses two 3-cell parallel stages the actual *battery* capacity is 4.8 Ah * 10.8 v. = 51.84 Wh rather than the 26- Wh which they seemed to be saying). So indeed average standard battery capacity seems to have about doubled since the days of my 570E Thinkpad (which has a 2.6 Ah, 10.8 v. battery) - though perhaps not for units that use only 4-cell batteries with 8 cells as their upgrade option (at 14.4 v.), nor for the newest models that use 3-cell batteries (and LED backlights). That's still a lot less than 80 Wh for a 'normal' battery capacity, but a lot closer than I thought it was. > Note also that some systems are able to work with less than 8W indeed. I'll have to take slight exception to the use of the word 'work' above - see below. > That usually require ultra-low-voltage CPUs which run a bit slower, like > 1.2GHz (and is measured in "optimal" (aka optimistic) conditions such as > very low screen brightness and almost no CPU activity). > See http://www.thinkwiki.org/wiki/Idle_consumptions for examples. Hmmm. The first sub-8-W example there has the processor idling, the disk spun down, and the backlight turned completely off - is that actually a 'working' situation? And while it squeaks in at 7.5W with the backlight at minimum brightness, that's still with the processor idling and the disk spun down. What it does illustrate, though, is that 100% CPU utilization takes a lot more power than the backlight does for this laptop (not that this says much about normal laptop relative power requirements, since the processor is seldom working very hard then, but it's still an interesting data point for that particular display and CPU - a PII 366). The next sub-8-W example similarly has the backlight off, blank screen, idle processor, and spun-down disk. Then there's another that just squeaks in at under 8 W with minimum backlight and otherwise idle, followed by one that actually has the hard disk spinning (but is undervolted)... The only actually promising sub-8-W example is the X21 at 5.5 W: yes, it's only a 12.1" screen and it's still not doing anything, but it would be nice to know more specifics about it. - bill
From: Ken Hagan on 21 Apr 2008 06:52
On Sun, 20 Apr 2008 06:25:16 +0100, Bill Todd <billtodd(a)metrocast.net> wrote: > Hmmm. The first sub-8-W example there has the processor idling, the > disk spun down, and the backlight turned completely off - is that > actually a 'working' situation? It might be quite representative for someone using the laptop as a highly mobile thin-client for applications running on a central server. There probably aren't very many people who do that, but if only they had the server and the necessary skillset to configure it all, there's quite a lot of people whose computer usage *could* be adapted to that pattern. Think how many people read a book in bed, or on the sofa. |