2D Also Bad For Ubuntu Unity Against KDE, GNOME, LXDE, Xfce
Phoronix: 2D Also Bad For Ubuntu Unity Against KDE, GNOME, LXDE, Xfce
In continuation of last week's OpenGL benchmarks under Unity, GNOME, KDE, Xfce, and LXDE desktops from Ubuntu 12.10, here are benchmarks looking at the 2D performance of these different Linux desktop environment choices when testing both the Intel UXA and SNA acceleration back-ends.
How about including standalone compiz in these benchmarks, so we can see just how much of this is Unity itself causing issues? It would be useful for developers to see where they need to target their optimizations, of course.
Was composition enabled for XFCE in the test? It is off by default.
What about a test with an actual graphics card? :/
So there was a GPU hang (cause unknown as the i915_error_state was not logged) which affected the gnome-shell results. Not sure why the disparity for lxde, but I did observe that pts fails to inhibit the xscreensaver used by lxsession which plays havoc with the results. There's a minor performance disparity between our systems (apart from the large delta caused by Ubuntu's broken patches ;-) which I guess will be done to GPU clock frequences and cache size.
For comparison see http://openbenchmarking.org/result/1...SU-1209080RA58, which happens to include results from one of those immensely noisy, hungry discrete beasts.
Well, according to the posts here, linux is only used for real work which doesn't require "actual" graphics cards. A couple of NAND gates and some latches for a frame buffer would be all 90% of linux users really need.
Originally Posted by Rigaldo
What is surprising is how poor the KDE performance is
While we are hammering on Unity, perhaps we should be hammering on the KDE guys as well.
what do you expect. you have a buggy window manager that hasn't hit 1.0 yet and a crappy graphics stack that was designed in the 80s. the big names have absolutely no interest in making desktop features better because the money from linux is in running database clusters or a web server. oh did i mention an office suite ten years behind and running on java.
Originally Posted by garegin
I'm not sure how to take your post(maybe it has a bit of "irony" in it too, or serious?).
Originally Posted by boast
And what exactly is this real work if I may ask? I'm not sure what posts you're reffering to.
(And I wouldn't take phoronix readers, even more so registered phoronix readers or any tech site's readers as a reliable sample of linux users, maybe as a sample of more advanced users mostly )
Besides, a gpu of any sort can help for rendering more than games etc, like webpages, documents, videos and more, if I am not mistaken that is. So all users would welcome it ..
(I assume real work would also require some internet research now and then and possibly writing some document. I doubt it's all done in a virtual terminal. And what about graphics design? Ain't that real work? :@ )
Oh, back on topic, since this are benchmarks regarding graphics, I think we should see a (more powerful) graphics card too. Sorry intel, you're doing good work, but it's not as powerful as what I consider a graphics card(and I believe it isn't meant to be). I mean, isn't that technically right? It's a gpu, not a "card".
(intel gpu and a graphics card , see the difference? xP)
Also we see some intel benchmarks and they even contain game framerates(sometimes only intel). While newer intel graphics are viable for at least some of those games, I believe it's more realistic to see the performance on discrete cards as well(I believe it's more important there actually ..). Or gamers prefer integrated graphics these days?
(here it wasn't games, but still graphics ..!)