[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]
On 06/09/13 11:27, Gordon Henderson wrote: > > Well, not quite, but what's it doing!!! > > top - 11:19:05 up 6 days, 14:36, 2 users, load average: 0.08, 0.26, > 0.22 > Tasks: 175 total, 2 running, 171 sleeping, 2 stopped, 0 zombie > %Cpu(s): 2.3 us, 2.7 sy, 0.1 ni, 95.0 id, 0.0 wa, 0.0 hi, 0.0 > si, 0.0 st > KiB Mem: 3831912 total, 3509196 used, 322716 free, 271272 buffers > KiB Swap: 0 total, 0 used, 0 free, 890016 cached > > PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND > 13071 gordon 20 0 1898m 1.3g 14m S 6.7 35.5 414:59.27 gnome-shell > 5369 gordon 20 0 1018m 393m 23m S 6.3 10.5 209:48.38 firefox > 30282 gordon 20 0 570m 356m 17m S 0.0 9.5 2:20.91 Fritzing > 25137 gordon 20 0 247m 30m 6240 S 0.0 0.8 2:11.95 soffice.bin > 3489 root 20 0 97756 27m 14m S 1.0 0.7 104:51.99 Xorg > 4011 gordon 20 0 298m 20m 5328 R 1.0 0.6 18:42.48 xchat > > So much bloat - all for some nice eye-candy )-: 1.8GB of RAM used for > what? > > Gordon > Just as Tom said, it's Linux doing it's job properly - the modern kernel(s) cache pretty aggressively when you have free RAM, and why not? Unlike arbitrarily smashing the wear levelling on your SSD by swapping to it constantly, there's nothing to be gained by *not* utiltising all that lovely free RAM in your machine. Even so, it tends to cut off at a certain, pretty predictable level - this machine is pretty much "at rest" just running my usual stuff - a terminal or two, firefox & thunderbird, a couple of file browser windows, etc and of course gnome-shell. This is before I fire up KVM/VMWare/VBox, start compiling or doing anything heavy, so at the moment my system is doing pretty much what yours is doing. However, this machine has 16Gb of RAM instead of your 4Gb, and even so, it's memory usage is very similar: top - 15:23:52 up 104 min, 2 users, load average: 0.14, 0.33, 0.26 Tasks: 261 total, 2 running, 259 sleeping, 0 stopped, 0 zombie %Cpu(s): 5.7 us, 0.4 sy, 0.1 ni, 93.2 id, 0.1 wa, 0.3 hi, 0.2 si, 0.0 st KiB Mem: 16403432 total, 4761536 used, 11641896 free, 196504 buffers KiB Swap: 0 total, 0 used, 0 free, 2708928 cached In my experience, until I start doing seriously computationally expensive work, no matter how long I leave this machine running memory usage never really increases past this point - it doesn't ever just keep bloating and run out (I mean, that would be an absolute disgrace in a 2013-era modern sophisticated OS). My i7 has 32Gb of RAM and runs an almost identical setup, except with the Awesome WM instead of Gnome3 and "at rest", it's memory stats are almost identical: in fact, slightly lower. The kernel on the i7 doesn't go mad and just suck up all available RAM for caching, it stays pretty constantly at 1-2Gb until I start pushing it. Like pretty much all systems. I've seen this so many times over the last few years as relatively speaking crazy-fast hardware has crept down in price and become mainstream - 15 years ago who would have thought we'd have affordable quad/hex core CPUs and 16-32Gb of RAM in our sub-£1k PCs? Yet people constantly complain that *shock!* *horror!* their sophisticated OS actually utilises all that power intelligently. For some reason otherwise very technical Linux users/admins seem offended that 25% of their available RAM is properly utilised by their OS for caching stuff: why? For those who may not know, Windows and Mac OS both do this as well (Win7/8 are both *very* aggressive at caching). Funnily enough, if you then start a "make -j8" in /usr/src/linux whilst watching htop you can see in realtime as your OS, which isn't stupid, immediately tosses all your cached soffice/firefox/etc crap straight out the window and immediately re-prioritises it's memory allocation. In short: this is modern Linux on modern hardware, behaving exactly as it should. What's the point in having a powerful system that never uses it's available resources? It's only memory bloat when it's inefficient/incorrect and your system actually runs out of physical RAM and starts paging - that sucks. But I bet it never happens on your machine, even if you leave firefox/gnome-shell running for weeks on end. Unless you've made some kind of manual "correction" and ruined it, of course... For those that are that worried about it, fear not! This is linux after all, so you can always roll up your sleeves and start screwing about with different schedulers, setting the tunable virtual memory parameters through /proc/sys/vm and recompiling your kernel with different parameters but speaking as someone who actually has to do this (primarily configuring *much* bigger boxes running stuff like Oracle 11G or in-memory DBs which have very, very specific parameters to adhere to) I can honestly tell you that for a general purpose home PC without multiple CPU sockets, >128Gb RAM and costly SLAs to maintain you would have to be literally out of your mind to even bother trying. TL:DR - you're worrying about nothing. Sit back, relax, and have some faith in the Linux gurus. If not the half-arsed distro weenies, at least people like Torvalds, who I think can be generally trusted not to randomly introduce completely mad memory eating tendencies into the mainline kernel. Sure, you have to keep an eye on those Gnome-shell coders who are a bit, umm, erratic, and the firefox plugin-container (thanks to stupid Flash) can sometimes runaway with itself a little, but that's it. Keep calm and carry on caching! Regards -- The Mailing List for the Devon & Cornwall LUG http://mailman.dclug.org.uk/listinfo/list FAQ: http://www.dcglug.org.uk/listfaq