Overclocking an Apple Macintosh Powerbook G3 Series (Wallstreet) Laptop

My goal was to take the 250MHz G3 and get 333MHz out of it with 66x5.0
But upon upping the clock to 333MHz , the cpu appeared to lock up due to the accumulated heat in the heatsink.
I am now successfully running the cpu at 300MHz with 66x4.5 and it appears to be running nicely.
Now, it's been running for about 2 hours total, so this isn't final. But this statement is more a proof of concept.

One thing to note is that the FSB is running at 66MHz, thus the memory bus is running at 66MHz and not 83MHz which
is a drop of 17MHz. When running GaugePRO it shows that the 'memory performance' benchmark test it performs is slower
on the slower bus speed. But I have yet to find any noticible performance loss with the slower memory speed.

I have now gotten the 250MHz cpu to run at the following speeds.

3.0 x 66MHz = 200MHz (successfully ran the 1MB L2 cache at 1:1 200MHz)
3.5 x 66MHz = 233MHz
4.0 x 66MHz = 266MHz
4.5 x 66MHz = 300MHz
5.0 x 66MHz = 333MHz (boots, but eventually overheats)
3.0 x 83MHz = 250MHz (stock configuration)
3.5 x 83MHz = 292MHz

The most annoying part of this project is the 0 ohm resistors being so small. For those of you with a
magnifying glass, it will be easier to see if you've made successful solder joints, but for me,
I ended up not using them and bridging the pads together with wire.


Below is a graphical representation of the various multiplier configurations on the CPU.


3.0x Multiplier.
R118, R119,
R115, R114
o (R120) o (R118) o
 
o (R119) o (R117) o
 
o (R115) o (R113) o
 
o (R114) o (R112) o
3.5x Multiplier.
R118, R117,
R113, R114
o (R120) o (R118) o
 
o (R119) o (R117) o
 
o (R115) o (R113) o
 
o (R114) o (R112) o
4.0x Multiplier.
R118, R119,
R113, R114
o (R120) o (R118) o
 
o (R119) o (R117) o
 
o (R115) o (R113) o
 
o (R114) o (R112) o
4.5x Multiplier.
R120, R117,
R113, R112
o (R120) o (R118) o
 
o (R119) o (R117) o
 
o (R115) o (R113) o
 
o (R114) o (R112) o
5.0x Multiplier.
R119, R118,
R113, R112
o (R120) o (R118) o
 
o (R119) o (R117) o
 
o (R115) o (R113) o
 
o (R114) o (R112) o
5.5x Multiplier.
R119, R118,
R115, R112
o (R120) o (R118) o
 
o (R119) o (R117) o
 
o (R115) o (R113) o
 
o (R114) o (R112) o
6.0x Multiplier.
R117, R118,
R115, R112
o (R120) o (R118) o
 
o (R119) o (R117) o
 
o (R115) o (R113) o
 
o (R114) o (R112) o


The memory bus frequency is decided by a Motorola MPC106 Bridge/Memory Controller.
The MPC106 is configured by 8-registers R6,R7,R8,R9,R10,R11,R12,R13 on the top side of the CPU card.
66 MHz FSB
o (R6) o (R7) o
 
o (R8) o (R9) o
 
o (R10) o (R11) o
 
o (R12) o (R13) o
 
83 MHz FSB
o (R6) o (R7) o
 
o (R8) o (R9) o
 
o (R10) o (R11) o
 
o (R12) o (R13) o
The bus clock frequency is decided by a Motorola MPC932 Low Voltage PLL Clock Driver. [datasheet]
The MPC932 configured by 4-registers R78,R79,R80,R81 on the under side of the CPU card.
66 MHz FSB
o (R77) o (R76) o
 
o (R79) o (R78) o
 
o (R81) o (R80) o
 
o (R85) o (R84) o
 
83 MHz FSB
o (R77) o (R76) o
 
o (R79) o (R78) o
 
o (R81) o (R80) o
 
o (R85) o (R84) o

I've been posting emails about this project to the LEM G-Books list.
Subject: Successful overclock 250MHz -> 292MHz
    Sep 22, 2004 (Nils)     Oct 13, 2004 (Nils)     Nov 15, 2004 (Nils)
Subject: How to control fan on Wallstreet in OS X ?
    Oct 29, 2004 (Nils)     +-> Oct 29, 2004 (Bill)     | +-> Oct 29, 2004 (Nils)     +-> Oct 29, 2004 (Anne)     | +-> Oct 29, 2004 (Nils)     +-> Oct 29, 2004 (Laurent)     | +-> Oct 29, 2004 (Nils)     | +-> Oct 30, 2004 (Mikael)     | | +-> Oct 30, 2004 (Nils)     | +-> Oct 29, 2004 (Anne)     | | +-> Oct 29, 2004 (Nils)     | +-> Oct 29, 2004 (Laurent)     | +->Oct 29, 2004 (Nils)     +-> Oct 29, 2004 (Bob)     +-> Oct 29, 2004 (Nils)     +-> Oct 30, 2004 (Mikael)     | +-> Oct 30, 2004 (Nils)     | +-> Oct 30, 2004 (Mikael)     | +-> Oct 30, 2004 (Nils)     | +-> Oct 31, 2004 (Mikael)     | +-> Oct 31, 2004 (Nils)     +-> Oct 31, 2004 (Nils)     +-> Nov 01, 2004 (Byron)     | +-> Nov 01, 2004 (Anne)     | | +->Nov 01, 2004 (Byron)     | +-> Nov 01, 2004 (Bob)     +-> Nov 01, 2004 (Nils)

I armed myself with information from the following sites..
http://www.powerbook-fr.com/pbg3/bricolage/overclock/pbg3serie.html
http://www.bekkoame.ne.jp/~t-imai/maine.html
http://www.powerlogix.com/products/cpudirector/index.html

Update : April 26, 2005

Had the machine running at 300MHz (66 x 4.5) for a while and found that the fan problem
looks to be solved with a 10.3.9 update. WOOT! So I've just clocked the machine
up to 333 (83 x 4) to see how things go. Hopefully there will be no problems.

On another note, I found a mistake in my page! The 66MHz/83MHz jumpers for R80/81 and
R78/79 were wrong. I've updated them. They now show the correct way to set
the jumpers.

-nils