On Wed, Jun 20, 2012 at 11:56 PM, Carsten Haitzler <ras...@rasterman.com> wrote: > On Wed, 20 Jun 2012 20:34:02 +1000 David Seikel <onef...@gmail.com> said: > >> On Wed, 20 Jun 2012 18:31:13 +0900 Carsten Haitzler (The Rasterman) >> <ras...@rasterman.com> wrote: >> >> > On Tue, 19 Jun 2012 18:20:38 -0300 "Eduardo Lima (Etrunko)" >> > <ebl...@gmail.com> said: >> > >> > > One thing I noticed when running some instances of terminology it >> > > that it is very processor hungry. For instance when running 'top' >> > > you may notice it figures among the first ones. This doesn't happen >> > > when you run top from gnome-terminal for instance. >> > >> > really? wow... are you sure u are using the same terminology i am? >> > the same efl? i have 3 terms (gnome, xterm, terminology) running, top >> > updates 1 time per second, sometimes xterm gets to 1%, sometim4es >> > gnome-terminal does, sometimes terminology - they are about the same. >> > just beware - gnome-terminal and xterm have the XSERVER do the text >> > rendering and draw, terminology is don't client-side in terminology. >> > (x just copies the pixels to the fb). a better comparison - have each >> > run top separately and measure system idle. i need to make it do >> > something otherwise i get 99.9% idle for all. so: >> > >> > top -d 0.01 >> > >> > terminology: 93.3% idle >> > gnome-terminal: 94% idle >> > xterm: 92.2% idle >> > >> > really... they are all the same. :) >> >> Opening up terminology and running it on a smaller window over the top >> of much larger roxterm window, my usual terminal these days, I get 9 to >> 11% cpu usage for terminology, and 1% for roxterm. Running top on both >> at the same time. Getting the same when I run top on each individually, >> with the idle terminal not even registering enough CPU to show up on the >> screen. This is with terminology using DejaVu Sans Mono font. >> >> With Terminus-12 font terminology runs a little faster, steady at 8%. >> >> For some odd reason, Monospace, the font I use in roxterm, does not >> show up in the terminology font selector. On the other hand >> Terminus-12 does not show up on roxterm, but that's a terminology >> built in font? So I changed roxterm to use Dejavu Sans Mono for a more >> direct comparison. Roxterm then started using between 1 and 4%, but >> mostly the 1% end of that range. >> >> Note that for this test the roxterm window is over four times the area >> of the terminology window, and showing twice the number of lines for >> top. >> >> This is terminology as compiled late last night. Only the font was >> changed. >> >> System idle is not a good test here, I run servers with random loads. >> Someone from the USA walking around a virtual world here in Australia is >> likely to ruin any system idle tests. lol >> >> I likely have gnome-terminal installed, as this is Ubuntu and it tends >> to break stuff when you try to remove the crap it installs by default. >> I also got xterm installed. >> >> xterm is consistently clocking in faster than roxterm at 1% to >> "0%" (more precision needed, but does not matter), even when I make it >> the same size. Changing xterms font looks complicated, so I did not >> bother. >> >> Gnome-terminal seems to be between xterm and roxterm, though sometimes >> peaking a percentage point above roxterm, and sometimes getting as fast >> as xterm. Note, Gnome-terminal also shows the Monospace font that >> Terminology does not, but I did the test in DejaVu Sans Mono. >> >> Really not a lot to pick speed wise between gnome-terminal, roxterm, >> and xterm. Terminology is much slower than all three, even in a much >> smaller window. Increasing the terminology window to the same size as >> the others and I'm getting 12 to 14% cpu usage. >> >> The difference in how they render text might be responsible, but I >> can't bring down my servers just to do that sort of test. > > without doing a system-wide comparison you are comparing apples and buffalo. > :) > xterm, rxvt and gnome-terminal push rendering into x. terminology pulls it > into > client. measuring JUST the client usage is not that helpful when at least 1 > other process is involved. :) > > as for speed-comparisons. gnome-terminal is taking 20x as long to cat large > files as terminology... here. in fact on every machine i have its that kind of > order of magnitude. :) >
Indeed, no question about the speed. Just run find in some directory (e.g. e svn checkout) and this is already a nice comparison. :) > as such given that terminology is using the cpu to do all the rendering, > nothing is offloaded (unless u choose the gl engine - then it is all offloaded > - your choice), is overlaying some nice lighting effects and pulsing cursor > (at > the expense of some cpu for niceness) i am absolutely STOKED with its current > performance and footprint, considering. Yep, using GL engine here. But the pulsing cursor is still using a fair amount of CPU, see my previous email. -- Eduardo de Barros Lima ebl...@gmail.com ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
