I do keep an eye on CPU use at idle as I move up in kernels but haven't done any battery run-down tests in a long time. It's kind of a bear to keep the variables down enough to get meaningful data without a LOT of work. On my workhorse second generation i5 laptop the idle use is down to 1% to 3% depending on the pup and pretty stable throughout the 4.17/4.18 set of kernels I've updated through. At those levels I couldn't see a 20% change at any rate. Where I do see a real and quantifiable difference in idle CPU use is between the upupbb/upupcc pups (I run them with pcmanfm integrated without GVFS and with a pretty stock JWM) and the slacko based LxPups. Being careful about governors, tray contents, Lxtask update setting etc. the upups independent of kernel always come in at 1% idle cpu while the LXDE/slacko based pups come in at 2% to 3%. Is it LXDE overhead vs JWM? Maybe, but my feeling is that the real culprit is the GVFS, my longtime nemesis. That difference might show up in battery use if a laptop was left at idle but do we ever really do that? I just lid-suspend even for a brief pause, really cutting power use, and the newer kernels have gotten good enough at that so whatever I'm doing resumes perfectly. Also hard to quantify, but a lot of work has been done on governors and power management in the more recent kernels and that plays very well on my 'hair trigger fan' Fujitsu laptops. IMO those changes are real and positive and would swamp any CPU idle changes.Moat wrote:I recently was reading about some newly developed CPU use (idle, mostly?) efficiency improvements added to the =>17.x kernels - wondering if you guys are seeing any noticeable positive differences in that regard...? (Marv?) Apparently up to 20% increase in laptop battery life in some use cases.
Bob
Just my thoughts,