Hi Folks,
as you see I`m opening another thread on Quarz 2D Extreme, and yes its supposed to be a thread about the "unsupported" feature of OS X 10.4.(0-3) Quarz 2D Extreme and not about the "supported" feature Quarz Extreme.
But why open another thread on on Quarz 2D Extreme?
I read the article on arstechnica (1.) about OS X Tiger and was facinated by the Q2DE technologie, but as mentiont in this article and on many other documents on the net it is disabled in the standard configuration.
It is not that difficuled to enable it(2.) and there is a bunch of people that report of drastic speed increase after aktivating it.
To the point, I activated it and there was no speed increase, my PB PB 12" late 2004 model even got slower.
So what I like to see and missed in other threads is a clear fact based documentation of working and not working Q2DE in the currend developement state and a investgation on the background.
So here is what I did:
At first I activated Q2DE by the Quarz Debuger, the first thing I noticed is that Safari (2.0.2) got very slow, that means when I, for example, try to scroll on a webpage it does not scroll smooth and there is a very high load on my CPU.
I looked at the Quarz Debuger Window List and found that the Safari-Window is marked green, green for accelerated, so it realy was using Q2DE.
Same thing goes for the Finder, when 'accelerated' it went damm slow, especially when any kind of sliding is done.
The only thing it seems to work for is the dock, when it is shown as accelearated it seems smoother and seems to have a lower CPU load when moving the mouse over it.
Many other apps and theire Windows where not markted as accelerated, even after restarting them a couple of times, so I activated Q2DE via the .plist file and restarted the book. The restart took longer as normal, after restart still not all windows where marked accelerated. The performance did not increase.
I took a XBench (Version 1.2) run with only the graphics test activated, the results read like this:
Qartz 2D Extreme disabelt:
Quartz Graphics Test 54.40
Line 47.88 3.19 Klines/sec [50% alpha]
Rectangle 53.26 15.90 Krects/sec [50% alpha]
Circle 55.23 4.50 Kcircles/sec [50% alpha]
Bezier 62.18 1.57 Kbeziers/sec [50% alpha]
Text 55.37 3.46 Kchars/sec
OpenGL Graphics Test 58.32
Spinning Squares 58.32 73.98 frames/sec
User Interface Test 32.53
Elements 32.53 149.31 refresh/sec
Qartz 2D Extreme enabelt:
Quartz Graphics Test 67.01
Line 96.62 6.43 Klines/sec [50% alpha]
Rectangle 73.86 22.05 Krects/sec [50% alpha]
Circle 89.01 7.26 Kcircles/sec [50% alpha]
Bezier 71.81 1.81 Kbeziers/sec [50% alpha]
Text 39.11 2.45 Kchars/sec
OpenGL Graphics Test 58.23
Spinning Squares 58.23 73.87 frames/sec
User Interface Test 2.31
Elements 2.31 10.61 refresh/sec
As one can see the all Quarz tests, but the text test, were noticable faster with activated Q2DE, but therefore the User Interface Test is realy messed up.
So I now konw on my book Q2DE rather messes things up then making my workflow faster.
But one question remains, why.
After crawling around in the devtools I found the OpenGL Driver Monitor, I realy dont know if this is the rigth tool but I gave it a try, and monitored some parameters.
I could not see any significant diverence between the enabled and disabeld mode.
Please send your experiences with Q2DE and some Benchmarks.
Any suggestions about the monitoring?
Is the performace increase an urbanmyth?
Is it a driver issue?
Is it about the poor performance of the Geforce 5200?
1.
http://arstechnica.com/reviews/os/macosx-10.4.ars/1
2.
http://www.macosxhints.com/article.p...ry=performance
My System:
System Version 10.4.3 (8F46)
Physical RAM 1280 MB
Model PowerBook6,4
Processor PowerPC G4 @ 1.33 GHz
L1 Cache 32K (instruction), 32K (data)
L2 Cache 512K @ 1.33 GHz No NAP
Bus Frequency 167 MHz
Video Card GeForce FX Go5200 64MB VRAM
Drive Type FUJITSU MHT2060AT
Top