Powerbook Performance

[Dronecatcher]

I've pondered this one for a while....does anyone know if the G4 CPU used in the last Powerbooks was specifically 'hobbled' in some way, maybe to run cooler in the laptop format and therefore be underpowered compared with it's desktop stablemates?
From what I could find the 7447A chip in the Powerbooks was the most advanced Apple put out, with higher performance plus CPU throttling if needed and yet when you compare a similar speed Powerbook to a Powermac, despite usually having the edge in bus speed, memory, L2 cache etc, the Powerbook has a lower benchmark - well if Geekbench can be believed (!)
This is based purely on Everymac scores not real world - I wonder whether anyone with a similar speed and spec Powermac & Powerbook could shed any light?

 

[Zotaccian]

L3 cache is probably the reason. I assume that the 7447 does run cooler since it can be cooled without monstrous heatsink & fan.

 

[Dronecatcher]

L3 cache is probably the reason. I assume that the 7447 does run cooler since it can be cooled without monstrous heatsink & fan.

Yes, that seems likely - the Titanium Powerbooks, which shared the same 7455 processor and L3 cache as their desktop counterparts were nearer each other in performance.

 

[Henriok]

I've pondered this one for a while....does anyone know if the G4 CPU used in the last Powerbooks was specifically 'hobbled' in some way, maybe to run cooler in the laptop format and therefore be underpowered compared with it's desktop stablemates?

You are questioning about laptop G4 (in PowerBook G4s and iBooks) versus desktop/tower G4 (in PowerMacs, iMacs and minis)? …and not let's say vs G5 in PowerMac G5?

OK. The fastest G4's Apple used was the 7447/B in the last PowerBook G4, clocked at 1670 MHz and that processor didn't come with L3 cache, but had a 5120 kB L2 cache on chip. Those machines used a 167 MHz system bus, DDR333 memory and 5400 rpm 2.5" hard drive.

The fastest G4's in Apple's towers was the 7455 in the Mirror Drive Door PowerMac G4, clocked at 1420 MHz and with 256 kB L2 cache on chip and 2 MB L3 cache off die. The system used two of those, on a 167 MHz system bus, DDR333 memory and 7200 rpm 3.5" hard drive.

So, the question is.. is the 7447/B at 1670 MHz and larger L2 cache but without L3 cache faster than a 7455 at 1420 MHz with smaller L2 cache but with L3 cache?

That'd be a close fight but I'd put my bet on the 7447/B, and by the time Apple did use that (October 2005) they'd been using G5s in their towers since a couple of years back (June 2003).

The G4 processors that Apple used in their laptops was lower clock versions (for heat reasons), often with no external cache (or even G3s), compared to the G4s inside desktops at the same time. The last, 7447/B, was manufactured on a 130 nm process, and Apple never used the L3 enabled big brother, the 7457. You could buy them on third party upgraded though, and you could even buy Freescale's last true G4 processor, the 7448 clocked up to 2 GHz. In dual configurations. That'd be a beast, even compared to G5s.

Pitting the fastest PowerMac G4 version against the fastest PowerBook G4 would favor the tower. Why? Because it had two CPUs, a much faster hard drive and CPU, and probably more RAM. It'd win even if the CPU in it self might have been slower.

 

[Hrududu]

It is really hard to guess, but I have a DLSD 17" PowerBook G4 and it is remarkably slow. G4's have great performance in pairs because of OS X's ability to use multiple CPU's, but in single CPU configurations they just don't have much muscle. Like others have said, the L3 cache on the earlier G4's was HUGE in real life feel and performance. I don't think they were intentionally slowed down by Apple, but had simply hit the ceiling as far as what the G4 could do in what was rapidly becoming a multicore world. I would LOVE to see what Freescale e600 chips would have done side by side with the Core Duo.

 

[gavinstubbs09]

What I think is lame is the fact my 1.42GHz iBook (I know, geekbench) scored 823 while my dual 1GHz MDD scored 825. I mean WTH??? A single 1.42 is equal to a dual 1GHz setup? I think that is lame. My 1GHz QuickSilver scores 926.

 

[Altemose]

What I think is lame is the fact my 1.42GHz iBook (I know, geekbench) scored 823 while my dual 1GHz MDD scored 825. I mean WTH??? A single 1.42 is equal to a dual 1GHz setup? I think that is lame. My 1GHz QuickSilver scores 926.

It has to do with cache I believe.

----------

L3 cache is probably the reason. I assume that the 7447 does run cooler since it can be cooled without monstrous heatsink & fan.

Aluminum PowerBooks had really large heatsink for a laptop. Not as big as a desktop obviously but pretty big!

 

[Dronecatcher]

It's very interesting the apparent role the L3 cache plays in performance. I was drawing a comparison to Al Powerbooks to similar CPU speed, single processor Powermacs only and using geekbench for reference - I've always found benchmarking to be deceptive but it's your only option for referencing machines you don't have.
The Titanium Powerbooks that had more in common spec wise with their desktop counterparts on the whole benchmarked higher - it seems L3 is the key.
Looking back..I had a 667 Titanium PB that I'd fitted with a 7200 rpm drive, running Tiger - it didn't seem any slower in use than the two 1.67 and my current 1.33 Powerbooks I've used since.

Totally agree with the 'laptop as heatsink' - very shrewd on Apple's part!