[bcaslis]

I have a new 1.5GHz 17" PowerBook that is maxed out. I run some performance number and found something interesting. It looks like the processor when not running full out doesn't "scale" like the old one. I believe if I remember correctly that the old processor had more steps, but the new 7447A is full speed or half speed. This would seem to bear out from the tests I made. For example

17" 1.33GHz running Xbench. On auto power setting it got a CPU score of 103.45, on highest performance it got 161.51 (sorry didn't run reduced when I tried it).

Now, on a 1.5GHz 17" you get the following: reduced 88.78, automatic 91.24, highest 182.42. Interesting no? The highest scale like I thought but the auto isn't very good.

I tried a couple of real world things also:

Photoshop opening 64 files (one at a time), doing some color conversion and saving (an existing script I had). With reduced 4 minutes 47 seconds, automatic 3 minutes 21 seconds, highest 2 minutes 52 seconds. Not as bad as xbench but not as good as I hoped for given that Activity Monitor was pegged most of the time.

Now the final confusing one. I ran Virtual PC and measured two things. First time to launch Windows and get to the login screen, second after hitting the login button time for Windows desktop to load and all the startup activities to finish so CPU usage drops. I did four runs with interesting results:

Run 1, automatic, 1: 34 seconds 2: 2 minutes 23 seconds
Run 2, highest, 1: 29 seconds 2: 1 minute 38 seconds
Run 3, automatic, 1: 30 seconds 2: 2 minutes 23 seconds
Run 4, reduced, 1: 42 seconds 2: 2 minutes 21 seconds

Interesting? Virtual PC sucks on the 1.5 GHz without highest performance on. I'm not sure what it all means, but it's definitely a behaviour change compared to the previous generation of PowerBooks. So, this thing can screen but automatic performance seems pretty hit or miss.

[Lateralus]

An XBench score of over 180 seems very odd for any PowerBook. My Dual G4 only gets around 160. The G5 1.6 gets even less than that.

[striker100]

I find that everytime I run Xbench I get a different result, even at the same settings. On my 17" 1.33Ghz Powerbook with 512mb ram on highest setting one time I got a total reading of 124 with a CPU score of 142.
One minute later I ran it again at the same setting and got a total score of 119 yet the CPU score was higher.

Today I was at an Apple retail store and ran Xbench on a new 15" 1,5Ghz Powermac. I made sure no apps were open, I believe I ran it at the Automatic setting & got a total score of only 108 which doesn't seem that good even at auto.

[bcaslis]

Yes you are right that you get different results each time from Xbench. I ran it a few times and picked a representative number.

You are right that the auto setting seems low. That was what led to my running other things.

That's also why I ran Photoshop and Virtual PC. It seems clear to me that on auto the results from the 1.5GHz are not as good as the previous processors. However on highest it runs in the range I expected.

Originally posted by striker100:
I find that everytime I run Xbench I get a different result, even at the same settings. On my 17" 1.33Ghz Powerbook with 512mb ram on highest setting one time I got a total reading of 124 with a CPU score of 142.
One minute later I ran it again at the same setting and got a total score of 119 yet the CPU score was higher.

Today I was at an Apple retail store and ran Xbench on a new 15" 1,5Ghz Powermac. I made sure no apps were open, I believe I ran it at the Automatic setting & got a total score of only 108 which doesn't seem that good even at auto.

[ibook_steve]

According to the developer notes, all the new Powerbooks use the 7447 processor which has a DFS (dynamic frequency switch) feature. This means that the processor, in the automatic setting, switches its internal frequency only between full speed and half speed (1.5 GHz and 750 MHz in your case). On the previous versions, bus slewing was used to adjust the frequency of the system bus between full speed at 167 MHz and reduced (probably 133 or 100). Thus, reduced speed on the previous revisions would not have been as slow as on the current ones because the speed was not cut in half as it is with the DFS processors.

Steve

[bcaslis]

Yes, that is the reference I mentioned earlier and in that regard it's not too surprising some of the results. What I didn't expect is that some apps that put high load on the PowerBook basically stink when run under the automatic setting. The Photoshop result is disappointing with the difference between high and automatic, but I was shocked by the Virtual PC result. It could be there is some variable I'm not seeing, but I did the four runs right after each other and basically Virtual PC ran at reduced speed unless I set the power setting to highest. This was not my experience when running it on a 15" 1.25GHz machine in the past.

Originally posted by ibook_steve:
According to the developer notes, all the new Powerbooks use the 7447 processor which has a DFS (dynamic frequency switch) feature. This means that the processor, in the automatic setting, switches its internal frequency only between full speed and half speed (1.5 GHz and 750 MHz in your case). On the previous versions, bus slewing was used to adjust the frequency of the system bus between full speed at 167 MHz and reduced (probably 133 or 100). Thus, reduced speed on the previous revisions would not have been as slow as on the current ones because the speed was not cut in half as it is with the DFS processors.

Steve