Yes using powertop, ive been able to get my last intel laptop (thinkpad t440) to ~8W and my latest amd laptop (a hp 745 g3 carrizo) to ~10W. Initial value on the intel laptop from memory was ~24W and on the amd laptop it was 21W.
This is a good start http://linrunner.de/en/tlp/docs/tlp-faq.html Depending on your HW/distro/DE, you might want to start playing around with powertop --calibrate and TLP. Highly suggest switching to a lightweight DE like i3.
I've never gotten useful results from powertop. The model it builds to correlate system events vs. observed power usage always diverges with reality very quickly. In the good case it'll just do something a little silly like conclude that the CPU fan is the ultimate source of all power consumption. In the bad case it'll be something like think that a pair of linux veths uses 3kW (not a typo) of power.
I was the same - mixed results at best. I figured that manually playing with this all the time was the wrong solution and installed TLP instead which seems to work pretty nicely. Not sure if it is mentioned in the linked article, so if not then check it here: http://linrunner.de/en/tlp/docs/tlp-linux-advanced-power-man...
I nearly doubled my battery life on linux by zooming on facebook until the chat bar on the right hand side disappeared. Didn't even need to close the facebook tab, just that bar was keeping the CPUs running hot. That's been effective enough for me not to bother with decent trace tools just yet.
I do use the chat though, and every time you click on the chat it brings back the sidebar and requires clicking a tiny icon and a menu option to get rid of it again.
Little-known tip (I think): messenger.com offers a web UI for Facebook chat similar to their iOS/Android apps, also ad-free and significantly cleaner than the actual Facebook UI. No idea whether it's actually better on power consumption though.
I didn't make a real trace, but was impressed when I saw the amount of CPU was being wasted for the the three cloud drives that I had: Microsoft OneDrive, DropBox and Google Drive. They all offer some free extra space when you login from a new machine and I accepted all of them. When I disabled them all my machine were much more responsive.
You'd be amazed at the amount of disk IO wasted by these programs, then.
Can somebody tell the Dropbox developers that right after boot is the exact wrong time to kill disk performance with a massive number of very random, very short reads? Jesus. People want to use their newly booted PC, not wait for Dropbox to realize nothing has changed after all while it chokes the browser being started.
I figure they don't notice because they have two PCIE 512GB SSD disks in a raid0 in their dev boxes. They need to test this stuff with spinning rust.
I'm a developer for a large organization and every year or two I snag a laptop on the way out of the bottom of the replacement cycle. When I have it out people sometimes come by and are shocked, "You're a developer, what are you doing with that thing? It must be the oldest laptop in the whole agency!"
And I tell them, "Exactly. If I make my app run well on this laptop, I know for sure it will run acceptably on every machine here."
A company I used to work for had a slow DSL line, with a ~7 year old pc, that they'd use to test their web-apps. That way, if it was responsive on that, then it'd be responsive on the customer's pc. I wish more companies tried this, instead of pushing the limit of what their expensive work-provided desktops could do.
I was going to recommend traffic manipulation and traffic shaping with netem and tc. I set this up at my last employer in the lab for use with our networked IP radios and such. Stimulating poor network conditions has been a mainstay in my testing ever since, even now that I work on web apps and distributed services.
For this reason on my MBA I've uninstalled Google Drive and now i only use Dropbox on demand, as in I start it every now and then to sync the files and then make it quit.
Genuinely wasn't meant to be a pointed criticism. I was curious to learn more about the use, which I just found different to what I'm used to.
I guess in this case I would read "wastage" as the reduction of available power over time. That is, the power was wasting away. Rather than the "waste" of power (frivolous use) by too many CPU cycles.
"Bryan: [...] But what I do have are a very particular set of skills, skills I have acquired over a very long career. Skills that make me a nightmare for people like you. If you let my daughter go now, that'll be the end of it. I will not look for you, I will not pursue you. But if you don't, I will look for you, I will find you, and I will kill you."
The worst localized title I know is "Die Hard: With a Vengeance", mysterious called (literally) "Die Hard - Mega Hard" in Denmark. Schwarzenegger's "Red Heat" is (translated) "A Russian cleans up Chicago".
If you really can’t figure out how to avoid polling then be smart about it. Waking up every second may seem very conservative, but if every long running program does it then your CPU may get woken up dozens of times per second.
The rules on timer coalescing are complex and they depend on the details of the timers. So, the timers may get coalesced, but it's hardly guaranteed, especially if the different wakeup frequencies are slightly different.
I suspect very few programs are using SetCoalescableTimer, but I'll add that as a suggestion.
Don't forget to disable Windows Defender which does lots of realtime scanning and hogs CPU resources and disk IO. A complete piece of trash that automatically deletes false positives. Never have I been so disappointed with Windows 10 than after upgrading and seeing this thing enabling itself and running.
(To the downvoters: I would love to hear feedback about why I'm mistaken here. Windows 7 didn't need such an authoritarian defender (auto-deleting things, scanning, etc.) and it felt just as safe)
I see Defender as a symptom - as it monitors the files, it only checks files that some other process opens. Go solve a problem of second process regularly accessing files on idle PC.
If you're on a Mac and using OSX, I highly suggest checking out Turbo Boost switcher https://github.com/rugarciap/Turbo-Boost-Switcher It's given me at least an hour more of battery life on my aging Macbook Pro 13 (2011). Another handy tool on OSX is CoconutBattery which is basically the closest I've found to PowerTop for monitoring power usage (I know ActivityMonitor somewhat shows power usage BUT it makes my fans & CPU spin like crazy when it's gathering info!)
I can just imagine a sql server team meeting where they discuss sql server's effect on laptop battery life. Given sql server's intended environment, I almost wonder if its activity is intentional.
Power consumption is a concern even in data centers, but yeah, the power cost of this inefficiency is probably inconsequential in situations where sqlservr is actually being used.
But, VS installed and started sqlservr on millions of developer machines so they should have paid some attention to this issue. Super sloppy.
it's also installed by millions of other applications - a sony laptop I had once shipped with some "music management" app installed by default that used it.
So true, if only more users realized this. I for one really appreciate your work this.
Users don't understand that computer resources like CPU time and battery level are really their property. It is an awful like paying a hidden poor quality tax.
It is not unreasonable to expect software to not waste CPU and battery time.
Windows Longhorn (Vista before the development reboot) actually shipped with SQL Server Express for WinFS. Though dotNet based shell and SQL Server Express based WinFS was too slow and consumed too much CPU resources.
WinFS was too slow and consumed too much CPU resources.
I'm genuinely curious what your source is for this. From the following uncited text on Wikipedia?
"An early revision of WinFS was also included, but very little in the way of a user interface was included, and as such it appeared to early testers to be nothing more than a service that consumed large amounts of memory and processor time."
Well, it's Express. Five years ago, when most development happened on a desktop, I'd agree. These days most developers have laptops. I guess it's time for them to revisit their Express config.
Waking up every minute is fine. Spending 1.5 s of CPU time doing nothing... I think I could improve on that. That's enough CPU time to drag down overall performance slightly when the machine is not idle.
I would have liked to see bit more discussion/analysis on CPU power states, wakeups and frequency scaling. My understanding CPU usage might not be always perfect proxy for energy usage.
Yeah, I thought about that, but I wasn't sure I could say anything authoritatively enough beyond "CPU consumption wastes power, and so does waking up". Apple's energy monitor seems to use that for its model so I guess it's not too bad.
On (recent) Macs, they have the Energy Impact numbers on the Activity Monitor. It can give a metric on which applications are taking up the most energy. However it doesn't show many of the system processes...
In 10.9 it was a simple function of on-CPU time and wakeups with each wakeup counted as 500µs of CPU time, in 10.10 a bunch of new counters were added (with machine-specific conversions to CPU time estimates) variables were added to the mix: whether the process was in "background" QoS, the number of disk read/writes, GPU usage and network activity (in packets#)
The menu of the battery status indicator also shows a summary of this, if there are applications that are using lots of power (see https://support.apple.com/en-us/HT202776)
After they added the menu telling me which 3rd party apps were wasting energy I reported bugs to them all and within about a year they had all removed whatever polling code and they dropped dramatically in power usage, today Google Chrome is the only 3rd party app that I use from time to time that drains my battery.
