Re: CPU Perf Power Mgt
On 20 Aug 2015, at 17:44, Justin Hibbits jr...@alumni.cwru.edu wrote: There was a working group at BSDCan this year on power management, and what we need to / can do to bring it up to par with the modern world. Unfortunately, I haven't had any time lately to work on it, but you can read the notes at https://wiki.freebsd.org/201506DevSummit/ClockDomains In short, the goal is to add infrastructure to the kernel to support overall power management of the system, scaling beyond cpufreq/powerd. Looking for volunteers who could do some of this, due to my lack of time to work on it. We also had a power management meeting at BSDCam this week. Robin presented some slides (though the download seems to be broken for me at the moment): https://wiki.freebsd.org/201508DevSummit?action=AttachFiledo=gettarget=power-management-for-freebsd-bsdcam-2015.pdf This includes the design that ARM is trying to push into Linux, which looks like it would be adaptable to FreeBSD (and, on the plus side, our complete lack of existing power management in the kernel already means we don’t already have the legacy stuff that they are fighting in Linux). The goal is to have the scheduler drive all of the (CPU) power management. Any volunteers to work on this are welcome - ARM people are happy to provide advice. David ___ freebsd-current@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org
CPU Perf Power Mgt
Hello, I’ve been thinking about CPU performance and power management on FreeBSD recently. As a user it seems like there has been little activity in this area and I wanted to try and understand what the situation was. From the publicly available information on powerd [1], the wiki [2] and my attempts to optimize hardware power/performance; it seems the current approach is quite old and laptop-focused. Recent CPU designs can control the state and frequency of individual cores very quickly. In the case of a single heavy thread, a multicore CPU might power-gate all but one core so the active core can be pushed to a higher frequency. This doesn’t seem to be possible on FreeBSD at the moment: powerd is userland (~250 ms poll) and can only control the frequency of all cores together. I understand this opens a can of worms as the CPU core states, frequency and scheduler would all need to co-operate. However, I think it’s important that this does happen. Without this functionality FreeBSD is leaving performance on the table and consuming more power than other operating systems. At BSDCan I heard that there was work going on for arm systems, but didn’t manage to get any details and whether it was relevant to amd64 too. TIA, Will PS. I was interested to see Intel announce at IDF that they'll be working with open source projects to implement Speed Shift Technology”, which leaves responsibility for p-state management on the CPU. [1] https://www.freebsd.org/cgi/man.cgi?query=powerd [2] https://wiki.freebsd.org/TuningPowerConsumption ___ freebsd-current@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org
Re: CPU Perf Power Mgt
[snip] and if you're interested in doing some stuff with the scheduler and interrupt handling then please poke jhb/i. -adrian ___ freebsd-current@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org
Re: CPU Perf Power Mgt
On Thu, Aug 20, 2015 at 5:44 AM, Will Green w...@sundivenetworks.com wrote: Hello, I’ve been thinking about CPU performance and power management on FreeBSD recently. As a user it seems like there has been little activity in this area and I wanted to try and understand what the situation was. From the publicly available information on powerd [1], the wiki [2] and my attempts to optimize hardware power/performance; it seems the current approach is quite old and laptop-focused. Recent CPU designs can control the state and frequency of individual cores very quickly. In the case of a single heavy thread, a multicore CPU might power-gate all but one core so the active core can be pushed to a higher frequency. This doesn’t seem to be possible on FreeBSD at the moment: powerd is userland (~250 ms poll) and can only control the frequency of all cores together. I understand this opens a can of worms as the CPU core states, frequency and scheduler would all need to co-operate. However, I think it’s important that this does happen. Without this functionality FreeBSD is leaving performance on the table and consuming more power than other operating systems. At BSDCan I heard that there was work going on for arm systems, but didn’t manage to get any details and whether it was relevant to amd64 too. TIA, Will PS. I was interested to see Intel announce at IDF that they'll be working with open source projects to implement Speed Shift Technology”, which leaves responsibility for p-state management on the CPU. [1] https://www.freebsd.org/cgi/man.cgi?query=powerd [2] https://wiki.freebsd.org/TuningPowerConsumption Hi Will, There was a working group at BSDCan this year on power management, and what we need to / can do to bring it up to par with the modern world. Unfortunately, I haven't had any time lately to work on it, but you can read the notes at https://wiki.freebsd.org/201506DevSummit/ClockDomains In short, the goal is to add infrastructure to the kernel to support overall power management of the system, scaling beyond cpufreq/powerd. Looking for volunteers who could do some of this, due to my lack of time to work on it. - Justin ___ freebsd-current@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org