Thanks!
I didn't mean that it doesn't have L3 cache at all, but if the 2003 MDD Dual 1,25GHz reissue BTO version had the 2MB L3 per CPU, i.e. 2x2MB L3.
...
Users that are OS9 only may choose a single CPU; depending on the DAW host application, a single will run both cooler and more stable (with some OS 9 based applications), so the optimal CPU (without overclocking) is the fastest single with the most cache...
http://www.everymac.com/systems/apple/xserve/specs/xserve_g4_1.33.html
Rip one out of an Xserve G4 1.33 and BAM... single cool CPU, runs at 1.33 Ghz. and has 2 MB Cache
Or overclock a stock single CPU to 1.5 Ghz
...While users that use both OS X and OS 9 will opt to rip a Dual 1.42 Processor board out of a FW 800 and put it in their 2003 MDD
and BAM, Dual 1.42 Ghz monster with 2 MB cache per CPU
Let us see, if I understand, what you are saying:
We have (all ever built):
August 20021. Power Mac G4 867 DP MDD, 133Mhz FSB, 1MB L3 per Core
2. Power Mac G4 1.0 DP MDD , 167MHz FSB, "
3. Power Mac G4 1.25 DP MDD, 167MHz FSB, 2MB L3 per Core
January 20034. Power Mac G4 1.0 FW800, 133Mhz FSB, 1MB L3 per Core
5. Power Mac G4 1.25 DP FW800, 167MHz FSB, "
6. Power Mac G4 1.42 DP FW800, 167MHz FSB, 2MB L3 per Core
June 20037. Power Mac G4 1.25 MDD ("reissue"), 167MHz FSB, 1MB L3 per Core
8. Power Mac G4 1.25 DP MDD BTO ("reissue"), 167MHz FSB, 1MB L3 per Core
Xserve9. Xserve G4 lower than 1.33GHz, 133MHz FSB, 2MB L3 per core
10. Xserve G4 1.33GHz Dual and Single, 167MHz FSB, 2MB L3 per core
So, if I want to have the best OS 9 machine:
a) get "7" = 1,25GHz G4 single, 167MHz cache, 1MB L3 cache and overclock to 1,5GHz (to get highest CPU clock + highest FSB)
b) get "7" and put in "10" = Single 1,33GHz G4 Xserve CPU, 167MHz FSB, 2MB L3 per core (to get highest CPU clock + highest FSB + biggest L3 cache)
So, if I want the best OS X + OS 9 bootable machine:
a) take CPU from "6" and put in 2, 3, 7 or 8
b) second best, take CPU 1,33GHz Dual out of Xserve and put in 2, 3, 7 or 8 (but will run cooler.
The Xserve CPU runs cooler at a similar clock speed, is it not slower? if we can trust Geekbench, I mean a 2x1,8GHz 7447 is just as fast as a 1.42GHz DP 7455...:
Looking at geekbench (all Dual CPU):
Xserve 1.33: 1200
PM 1.25: 1074
PM 1.42: 1224
-> Hm, seems not, if we can trust geekbench.