One thing that is easy to forget is that most modern OS use any "free" RAM for disk cache (aka buffers), so you will always look like you're using 100% RAM.
What's really important is to look for processes in the D state under top, (which are doing I/O) and also the amount of swap actually being used. But something has really gone awry on modern GUIs - my wife swears at (not by) the KDE interface running on my Linux box when things slow to a crawl (usually when I'm doing something slightly I/O intensive), whereas my trusty (and perhaps krusty) old fvwm is as lithe and responsive as ever. I used a Windows 7 machine at a previous client's with 4GB of memory and quad core. The machine would always slow to a crawl if I had Visual Studio, Firefox and Eclipse all open at the same time (and of course Cygwin), and even without eclipse running, I would need to restart Visual Studio and Firefox on the order of once a day to reclaim leaked memory. I got good at selectively killing processes so that I didn't need to do a full reboot every time. The CPU might be quad core, but never saw the load average go much above about 1.5, even with multiple parallel C++ compiles happening - the machine was far too I/O dominated. Something is wrong with Virtual Memory handling in modern UIs - it doesn't seem to matter which OS you're using. It's one good reaon keeping me using Linux, because I have the choice to use a minimal window manager that gets out of the way and lets you use the machine. Cheers On Fri, Jul 05, 2013 at 02:45:53PM -0600, Owen Densmore wrote: > O > n my non-SSD mini, before the change to SSD, I often had the experience > you mention. Since then, no. But likely the behavior is still the same, > just that the SSD manages it better. > > Before SSD, I had to run "purge" in a terminal to get the memory back. > I'll try starting lots of apps and see what happens on the new mini/SSD. > > Would be nice if Apple, finally, learns to handle swap space better. Maybe > Mountain Lion did so? > > -- Owen > > > On Fri, Jul 5, 2013 at 2:27 PM, Steve Smith <sasm...@swcp.com> wrote: > > > Owen ( and other OSX fanbois) - > > > > I'm guessing that a few others here will be interested in the technical > > details of this topic... > > > > I did not become interested in OSXs memory management until about 6 months > > ago when my PBPro with 4G running 10.6.x started throwing me the rainbow > > frisbee of death (or at least tedium) often. I began to look at the > > process table (via Activity Monitor) and noticed that **all** my > > applications seemed to be bloating up with memory... as if each and every > > one had memory leaks. > > > > Firefox, Thunderbird and Skype were the most notable. I kind of assumed > > that the problem was a system library that they all shared, and aggravated > > by the fact that they were all naturally wanting/needing/using lots of > > their own internal cache (well, maybe not Tbird so much?)... I also > > assumed that I had not updated my system properly (I tend to be pretty > > cavalier about keeping up with suggested updates, but trust the system (at > > large) to know what needs to be updated and not leave anything in the > > cracks)... > > > > I recently finally buried that machine after stripping it down to replace > > the charging port only to find afterwards that the problem was NOT that my > > battery was zeroed and my charge port too fried to take power... I finally > > gave up and blamed the easy/last-resort "logic board failure". I give my > > machines a lot of abuse. One of the SFX interns inherited the one my wife > > ran over in Iowa (shattered screen... he used it with an external monitor). > > > > Anyway... back on topic. The 15" 2010 MBP I bought to replace it had 8G > > and Mountain Lion installed. I assumed (hoped futilely) that my problems > > would evaporate with a full (up to date) fresh system (10.8.4 install and > > max memory). I didn't fret about it much but within a few days I started > > noticing (mostly because my previous machine had taught me to compulsively > > check the Memory Usage monitor) that I was operating on virtually 0 free > > memory as before. The big difference was that I was not getting the > > whirling frisbee of death very often and nearly 1/2 of the memory is > > labeled "Inactive", though under the 4G 10.6 circumstance I also had > > significant "Inactive" memory available at all times... > > > > I am postulating (very tentatively) that this new machine/configuration is > > more efficient at reclaiming "Inactive Memory" just-in-time... perhaps > > because it has the quad-thread version of the duo core or perhaps 10.8 > > fixed it up, or because my old system was just poorly configured (memory > > management libraries out of date?). > > > > One thing I am wondering is if others have had this problem (saturating > > physical memory and NOT getting efficient reclaiming of Inactive memory)? > > Or if others understand whether this is a real problem or just my lame > > understanding of how the memory management is supposed to work (I would > > sort of expect the Apps themselves to be managing memory more effectively > > than they seem to themselves, not just trusting the VM to keep them out of > > trouble?). > > > > - Steve > > > > ============================================================ > > FRIAM Applied Complexity Group listserv > > Meets Fridays 9a-11:30 at cafe at St. John's College > > to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com > > > ============================================================ > FRIAM Applied Complexity Group listserv > Meets Fridays 9a-11:30 at cafe at St. John's College > to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com -- ---------------------------------------------------------------------------- Prof Russell Standish Phone 0425 253119 (mobile) Principal, High Performance Coders Visiting Professor of Mathematics hpco...@hpcoders.com.au University of New South Wales http://www.hpcoders.com.au ---------------------------------------------------------------------------- ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com