av@tolsun.oulu.fi (Ari Vaulo) (03/28/91)
Hello!
I have some odd problems with my systems performance in some
Speedometer 2.5 tests. With general processors test the results
are followed:
3.87 times of SE's speed with 8.24 GC init on.
12.70 times os SE's speed with 8.24 CG init off.
My system is IIfx with only 8.24 GC card and lots of memory and
inits, but only 8.24 GC init seems to affect system performance
in tests. What's the real problem?
(1) 8,24 GC init is corrupted,
(2) 8.24 GC doesn't accelerate system performance,
(3) Speedometer tests wrong things, or
(4) test results are right, but I'm worrying unnecessary,
because there is not real problem?
I think that 8.24 GC acts as a bus master during Speedometers
QuickDraw functions tests (within processor test) and steals
control of the Nubus from processor, and thus Speedometer gives wrong
results. But with wristwatch timer there is 5 seconds difference
in processor test (2 vs 7 secs). Maybe the test is not good for 8.24 GC
or controversely. It's easy to see that 8.24 GC is working because
screen update is much better especially using 24 bit color depth.
Further tests with MacWrite II 1.0 and MacDraw II 1.1 (and all other
inits than 8.24 GC init was removed) give me following results. All
times are in seconds (averages of 10 tests).
Scrolling 6 page MacDraw II document with several rows and columns
1 square inch rectangles (1848 graphical objects) in four planes it takes:
(Grouped) 8.24GC on 8.24GC off
Scroll down 44.18 22.69
Scroll up 44.73 22.45
(Ungrouped)
Scroll down 42.98 22.47
Scroll up 43.20 22.39
(Ungrouped and all objects selected)
Scroll down 52.93 28.59
Scroll up 52.59 28.67
Copying same same document, with 10 layers (18480 graphical objects):
() 8.24GC on 8.24GC off
Cut 15.55 7.99
Paste 12.02 8.52
Scrolling 10 page MacWrite II 1.1 document with using Chicago, Geneva and
Monaco fonts and different font sizes. 33 test paragraphs in my document
contains following text: "One two three four five six seven eight nine ten
eleven twelve thirteen fourteen fifteen sixteen eighteen nineteen and
twenty"
() 8.24GC on 8.24GC off
Scroll down 4.73 11.57
Scroll up 4.50 11.58
(Whole text selected)
Scroll down 6.09 17.32
Scroll up 5.76 17.31
Why graphics accelerator slow down system so significant amount (50 %).
Does somebody have same kind of problems or otherwise know what's wrong
in tests (Speedometer, MacDraw II) or 8.24 GC, please tell me your opinion.
Ari
---------------------------------------------------------------------
Ari Vaulo eMail: av@tolsun.oulu.fi
University of Oulu voice: +358-81-352647
Dept. of Inf. Proc. Science
Linnanmaa
SF-90570 OULU, Finland
Ari Vaulo ---------------------------------------- eMail: av@tolsun.oulu.fi
University of Oulu --------------------------------------------------------
Department of Information Processing Science ------------------------------
Finland --------------------------------------------- voice: +358-81-352647