Hello, On Tuesday, I was all happy that I got APM working on my pre-2000 Compaq Deskpro following these instructions: http://forums.freebsd.org/showthread.php?t=4619
On Wednesday, after configuring the router to wake the server whenever DHCP leases are touched (quick&dirty hack), I was disappointed to learn that "suspend" mode saves only ~1watt over the "inherent" use of the HLT instruction by the kernel. I was expecting a savings of ~6 watts due to the disk spinning down. Approx power consumption (+- 1W): 51 Watts (busy; disk + CPU IIRC; not retested with DVD activity) 35 Watts idle 34 Watts suspend 3 Watts off (Measured using Kill-A-Watt model P4400) Part of the problem may be that I am not using the "on board" IDE controllers: I am using a Promise (Ultra100TX2) PDC20268 I realize the memory can't be shutdown without Hibernation support, but the disks can be spun down manually (using atacontrol): http://forums.freebsd.org/showthread.php?t=1012 However, when I try to do that, I find that the disk wakes within 2 seconds of spinning down. I noticed that the spindowns are logged. Could the log being written be causing the drive to spin up again? apm(4) says that apm gets around that problem by logging the suspend event AFTER waking up. I suppose it would be tricky to concurrently log to spin down of several disks that way. For example: Say disk with /var/log spins down at 00:00:05, but the rarely-used /srv drive spins down at 00:00:07. Should the logging drive defer recording BOTH spin-down messages, or spin-up, then spin-down again at about 00:15:20? Not that important for a 1W savings, but apm says my BIOS supports the following capabilities: global standby state // Supported sleep modes global suspend state resume timer from standby // Resume timer allows sleep to last resume timer from suspend // specific period of time? RI resume from standby // Wake on interrupts, i assume RI resume from suspend Would it be possible to coordinate the cron dameon with the suspend timer? Ie: wake 15 sec- 5min before next cron job? Not worth it without hibernate support though. apm(4) does not mention suspend timers at all. acpi(4) mentions "timer" as a sub-device and feature that can be disabled. Regards, James Phillips __________________________________________________________________ Connect with friends from any web browser - no download required. Try the new Yahoo! Canada Messenger for the Web BETA at http://ca.messenger.yahoo.com/webmessengerpromo.php _______________________________________________ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to "freebsd-questions-unsubscr...@freebsd.org"
