This only confirms my suspicions about 'performance improvement' seen with 10.2.x updates. Something is going wrong with optimizing various bits of Jaguar.
with a couple of exceptions, and obvious caveats about boot time differing due to more kexts and components in newer systems logically slowing load time vs. slimmer, earlier builds, what's the big deal?
cosmetic speed differences for a more robust and complete system (10.2.6)
and no mention of the extra time 10.1.x spent blowing up and freezing
unless you're a quake nut, or spend all day ripping, doubt you'd notice.
One of the touted performance increases for 10.2.3 was the Aqua interface. The benchmarks show this. Launching and other other areas differ as we have 12 of the 800mhz G4's and I cannot replicate the application launch times nor the boot times on 10.1.5 or 10.2.3 and 10.2.6.
The times are similar though.
I don't have a copy of Quake so if someone could supply a copy, I could do more Benchmarking (Unreal 2003 would also suffice)
The tested 10.1.5 installation was run from a really fragmented harddisk partition (...) OS X 10.2 was installed on an empty partition, 10.2.1, 10.2.2 and so on are updates to this installation, each of them (together with the daily work on the computer) increasing disk fragmentation again.
(...)
The framerates were observed over the period of about two minutes, they are not properly measurable because the visualizer relies to some degree on effects chosen by random, and these effects are using the CPU and graphics accelerator to different extents.
Great, exactly how useful are benchmarks if they are not done under the same conditions?
I recently jumped directly from 10.1.5 to 10.2.6 and have noticed definite performance improvements (e.g., booting, Finder, MS Office). This is true even though I can't take advantage of Quartz Extreme (I have a IMac G3 500).
I do find the tests noted at the beginning of this thread interesting, however, because it seems fairly clear that the speed improvements in Jag are not across the board and that greater optimization is needed. Whether this optimization is a matter for the OS or for individual applications, or both, is something I don't know.
Comments
cosmetic speed differences for a more robust and complete system (10.2.6)
and no mention of the extra time 10.1.x spent blowing up and freezing
unless you're a quake nut, or spend all day ripping, doubt you'd notice.
The times are similar though.
I don't have a copy of Quake so if someone could supply a copy, I could do more Benchmarking (Unreal 2003 would also suffice)
Dobby.
The tested 10.1.5 installation was run from a really fragmented harddisk partition (...) OS X 10.2 was installed on an empty partition, 10.2.1, 10.2.2 and so on are updates to this installation, each of them (together with the daily work on the computer) increasing disk fragmentation again.
(...)
The framerates were observed over the period of about two minutes, they are not properly measurable because the visualizer relies to some degree on effects chosen by random, and these effects are using the CPU and graphics accelerator to different extents.
Great, exactly how useful are benchmarks if they are not done under the same conditions?
The only things I've changed are : Radeon Mac Edition to Radeon 8500, and 640mbs of Ram to 1024mbs.
10.2.6 is faster than the older versions.
I do find the tests noted at the beginning of this thread interesting, however, because it seems fairly clear that the speed improvements in Jag are not across the board and that greater optimization is needed. Whether this optimization is a matter for the OS or for individual applications, or both, is something I don't know.