Re: VIA EPIA 5000 and ACPI Cx levels
On 10/10/10, Bruce Cran br...@cran.org.uk wrote: On Sunday 10 October 2010 21:49:30 b. f. wrote: If it has an i8254, that can also be used in one-shot mode if hint.attimer.0.timecounter=0 is used, since r212778. Thanks, I didn't know about that. After enabling it things are quite different: kern.eventtimer.periodic is now 1, and setting hw.acpi.cpu.cx_lowest=C2 results in 100% time being reported as being spent in C2 mode according to dev.cpu.0.cx_usage - using C3 causes the system to hang. Shouldn't a fully loaded CPU spent more time in C1 state though? When I run a program that results in 0% idle time cx_usage still reports that no time was spent in C1 state. I'm not sure what is going on here: if you set hint.attimer.0.timecounter=0 and kern.eventtimer.timer=i8254 in /boot/loader.conf, then the system should try to use the i8254 in one-shot mode, unless you've specifically set periodic mode. If kern.eventtimer.periodic=1, then you are _not_ using one-shot mode. If it was 0 before your latest changes, then you were previously using one-shot mode. But, as I wrote earlier, for kern.hz128 and kern.eventtimer.singlemul=1, periodic mode may result in more sleeping than one-shot mode, though at a price. This may be what you are seeing. The C-state used is determined in acpi_cpu_idle() in src/sys/dev/acpica/acpi_cpu.c, if you are using ACPI. I think that if the latency for the C2 state is low enough, the number of callouts and interrupts sufficiently low, and the scheduler quanta large enough, it's possible for your machine to mostly use C2 rather than C1. You can take a look at the algorithm, and make some experiments. Note that bus mastering activity, which can include routine USB polling, may prevent the use of C3. This or the high latency of C3 may account for your machine not using it. Also note that you shouldn't use a LAPIC timer if you are using C3 or deeper sleep states. b. ___ 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
Re: VIA EPIA 5000 and ACPI Cx levels
On Thu, 14 Oct 2010 19:43:36 + b. f. bf1...@googlemail.com wrote: I'm not sure what is going on here: if you set hint.attimer.0.timecounter=0 and kern.eventtimer.timer=i8254 in /boot/loader.conf, then the system should try to use the i8254 in one-shot mode, unless you've specifically set periodic mode. If kern.eventtimer.periodic=1, then you are _not_ using one-shot mode. If it was 0 before your latest changes, then you were previously using one-shot mode. But, as I wrote earlier, for kern.hz128 and kern.eventtimer.singlemul=1, periodic mode may result in more sleeping than one-shot mode, though at a price. This may be what you are seeing. Sorry, that was a typo: kern.eventtimer.periodic=0 after setting hint.attimer.0.timecounter=0. I'm seeing 145 interrupts per second now, so setting hz=100 and using singlemul mode would further decrease it - but I guess it's not something you would want to do on a router or desktop. This or the high latency of C3 may account for your machine not using it. Also note that you shouldn't use a LAPIC timer if you are using C3 or deeper sleep states. Thanks. It sounds like things are working as they should then. Good to know the new timer code is working properly on this more unusual hardware! -- Bruce Cran ___ 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
VIA EPIA 5000 and ACPI Cx levels
I recently upgraded to HEAD on my VIA EPIA C3 box, and had thought about trying out the new one-shot timer mode. Reading mav@'s email it seems that since it doesn't have LAPIC or HPET timers it won't work. However I thought I should still get power savings by using higher Cx levels, but setting hw.acpi.cpu.cx_lowest to C3 I'm only seeing 0.8% time spent in C2 mode, the rest being in C1. The CPU's idle so I don't understand why it's not going into deeper sleep modes. Is this likely to just be an ACPI bug? -- Bruce Cran ___ 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
Re: VIA EPIA 5000 and ACPI Cx levels
I recently upgraded to HEAD on my VIA EPIA C3 box, and had thought about trying out the new one-shot timer mode. Reading mav@'s email it seems that since it doesn't have LAPIC or HPET timers it won't work. However I thought I should still get power savings by using higher Cx levels, but setting hw.acpi.cpu.cx_lowest to C3 I'm only seeing 0.8% time spent in C2 mode, the rest being in C1. The CPU's idle so I don't understand why it's not going into deeper sleep modes. Is this likely to just be an ACPI bug? If it has an i8254, that can also be used in one-shot mode if hint.attimer.0.timecounter=0 is used, since r212778. But for low values of kern.hz, I've found that periodic mode can result in fewer interrupts (albeit increased latencies and lower accuracy in accounting) than one-shot mode, if kern.eventtimer.singlemul=1. As for the power-saving states, are you using a simple 'sysctl dev.cpu.0.cx_usage' to find the percentages? If you're doing something more involved, you may be affecting the measurements. Also, does the system think that the deeper sleep states are available on your machine? If so, what are their latencies? If they are high, they may be used less often, or not at all. Did you follow some of the other recommendations to allow more sleeping, like at: http://wiki.freebsd.org/TuningPowerConsumption ? Or apply mav's extra patch to further reduce interrupts: http://people.freebsd.org/~mav/tm6292_idle.patch ? b. ___ 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
Re: VIA EPIA 5000 and ACPI Cx levels
On Sunday 10 October 2010 21:49:30 b. f. wrote: If it has an i8254, that can also be used in one-shot mode if hint.attimer.0.timecounter=0 is used, since r212778. Thanks, I didn't know about that. After enabling it things are quite different: kern.eventtimer.periodic is now 1, and setting hw.acpi.cpu.cx_lowest=C2 results in 100% time being reported as being spent in C2 mode according to dev.cpu.0.cx_usage - using C3 causes the system to hang. Shouldn't a fully loaded CPU spent more time in C1 state though? When I run a program that results in 0% idle time cx_usage still reports that no time was spent in C1 state. But for low values of kern.hz, I've found that periodic mode can result in fewer interrupts (albeit increased latencies and lower accuracy in accounting) than one-shot mode, if kern.eventtimer.singlemul=1. As for the power-saving states, are you using a simple 'sysctl dev.cpu.0.cx_usage' to find the percentages? If you're doing something more involved, you may be affecting the measurements. Also, does the system think that the deeper sleep states are available on your machine? If so, what are their latencies? If they are high, they may be used less often, or not at all. I'm just using dev.cpu.0.cx_usage to check the Cx level usage. According to dev.cpu.0.cx_supported, I have: C1/0, C2/90, C3/900 Did you follow some of the other recommendations to allow more sleeping, like at: http://wiki.freebsd.org/TuningPowerConsumption I haven't done anything extra yet, since I was mainly interested in seeing if one-shot mode worked on this box. I can't use powerd because running at 266MHz (I only have 533 and 266 available) results in too much of a slowdown. -- Bruce Cran ___ 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
