I added those arguments to my xorg.conf successfully and still no dice - e compositing and my 3d games run like a slide show. I have an nvidia 260m gtx that handles all other compositing managers without issue. I've got the latest beta nvidia drivers installed (270.xx). and I tried other versions with the same results. Are there certain flags I can compile evas/ecore with that might yield me better performance?
Also, I am compiling everything on a separate system and then bringing the packages over to my nvidia system - but that should change anything right? ~Jeff Hoogland On Sat, Mar 19, 2011 at 6:15 AM, Carsten Haitzler <ras...@rasterman.com>wrote: > On Sat, 19 Mar 2011 17:31:04 +0800 P Purkayastha <ppu...@gmail.com> said: > > > It seems to depend on the type of nvidia card you have. I had a T61 with > > nvidia card NVS140m with 128M video ram. Composite would be decently fast > > for a while, but gradually get very slow after a couple of hours. The > > slow-down could be hastened if I dared to run any video via vdpau, or > > restart e. ecomorph used to run fine for a while, but eventually get slow > > too (in about a day). > > > > I now have a different laptop with nvidia 310M with 1G video ram. > composite > > is really fast in this. The card seems able to run without any slowness, > > tested for over a week. Restarts of e or playing videos via vdpau has no > > slow-down effect. > > > > The difference between the above two systems was so stark that I believe > my > > earlier graphics card (or the driver) was just not up to the mark. Also, > > probably composite is more taxing and unforgiving on your gpu than > > ecomorph/compiz (raster can confirm perhaps?). Eventually, the nvs140m > > graphics card died suddenly in the middle of playing a video (the > infamous > > nvidia hardware problem). This also leads me to believe that the hardware > > was not up to the mark. > > > > raster has always claimed that composite was smooth on his system (that > too > > with a dual screen setup at high resolution). I presume he has a powerful > > enough and recent graphics card. > > actually have a whole range off them. my desktop has the lowest end nvidia > - > 8600GTS with only 256m ram. laptop actually is the highest end (gt-330m, > 1gb > vid ram). > > as such this slowdown issue is almost certainly some nvidia driver resource > leak. you will find that no matter how many times you restart the > enlightenment > process, it will remain slow until an Xorg restart. > > evas is the rendering engine for e17's comp module. that means it needs > what > evas needs. evas needs a GLSL capable GPU. old GPU's just don't cut it and > drivers that don't support GLSL (properly and efficiently) will be poor. As > such a good driver that supports GLSL will perform equally well as compiz. > just > the baseline support requirement for evas's rendering is higher than compiz > (needs newer card and decent drivers). As such all nvidia cards have done > full > hardware shaders that are GLSL capable since the GF6xxx series. there will > be > no difference between GLSL and fixed function on these level of cards and > up as > all the fixed pipeline is implemented as shaders anyway. apparently the > open > radeon drivers don't (properly) support GLSL shaders, so with ati you're > screwed unless you use the gallium drivers i understand. they can do > shaders > right. fglrx (closed drivers) do do shaders right, but fglrx just cant > properly > handle direct rendering + compositing, so texture-from-pixmap exhibits > bugs. > nvidia handles this just fine. the intel drivers do shaders just fine on > the > 945GM i have and work just fine with e17's comp and speed is good. no > resource > leak issues. > > you may want to try the newest nvidia drivers and combinations to see if > issues went away. jeffdammeth also suspects loose binding of textures may > trigger it, though compiz also can use loose binding (loose binding is a > good > speedup for nvidia - or was). it might be interesting to disable loose > bindings > on nvidia (see evas_x_main.c in the gl_x11 engine where it does > gw->detected.loose_binding = 1 when detecting nvidia). > > > On Sat, Mar 19, 2011 at 7:44 AM, Jeff Hoogland <jeffhoogl...@linux.com > >wrote: > > > > > Howdy There, > > > > > > So I finally got Evas/Ecore to build with OpenGL support - so my > > > compositing > > > is now running in OpenGL mode (instead of software) and much to my > dismay > > > everything still runs horridly slow on my nvidia graphics card! > > > Ecomorph/other three-d run just fine, but E's built in compositing is > just > > > a > > > dog. It is lessthan smooth when changing desktops and it cuts the FPS I > see > > > when gaming down to 1/3 of what it normally is. Is this normal or is > > > something wrong with my setup? > > > > > > ~Jeff Hoogland > > > > > > > ------------------------------------------------------------------------------ > > > Colocation vs. Managed Hosting > > > A question and answer guide to determining the best fit > > > for your organization - today and in the future. > > > http://p.sf.net/sfu/internap-sfd2d > > > _______________________________________________ > > > enlightenment-users mailing list > > > enlightenment-users@lists.sourceforge.net > > > https://lists.sourceforge.net/lists/listinfo/enlightenment-users > > > > > > ------------------------------------------------------------------------------ > > Colocation vs. Managed Hosting > > A question and answer guide to determining the best fit > > for your organization - today and in the future. > > http://p.sf.net/sfu/internap-sfd2d > > _______________________________________________ > > enlightenment-users mailing list > > enlightenment-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/enlightenment-users > > > > > -- > ------------- Codito, ergo sum - "I code, therefore I am" -------------- > The Rasterman (Carsten Haitzler) ras...@rasterman.com > > ------------------------------------------------------------------------------ Colocation vs. Managed Hosting A question and answer guide to determining the best fit for your organization - today and in the future. http://p.sf.net/sfu/internap-sfd2d _______________________________________________ enlightenment-users mailing list enlightenment-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-users
