The Ultimate Mac Overclocking Thread

[rabidz7]

I have compiled a sticky that shows how to overclock your PowerPC mac CPUs to eliminate the large amount of questions. I want every PowerPC mac to be overclockable, and that is only possible with your help; I cannot purchase every G5 iMac, 17" PowerBook, 12" PowerBook and PowerMac G5.

G3:

iMac
Multiplier
Bus

PowerMac B&W


iBook


PowerBook:

Kanga
No overclock available

All PowerBook G3s after Kanga



G4:


PowerMac:

Yikes


Sawtooth

Gigabit Ethernet
Multiplier
Bus

Digital Audio
Multiplier
Bus

Quick silver
Multiplier
Dual Multiplier
Bus

MDD Dual
Multiplier and vCore
Bus
MDD Single

Multiplier
Bus

iMac:

100MHz bus
167MHz bus


PowerBook:

Titanium


12" Aluminum

15" Aluminum low resolution

15" Aluminum high resolution:



Multiplier
All multiplier settings appear to be 10K-Ohm

..................0...........1
PLL0 -> R3721+R3720

PLL1 -> R3723+R3722

PLL2 -> R3725+R3724

PLL3 -> R3727+R3726

PLL4 -> R3729+R3728

PLL5 -> R3731+R3730

NOTE: PLL5 is only for diagnostic use. Do not set it to anything other than 0. The PowerBook will not function properly with PLL5 set to 1.



Bus
All bus settings appear to be 10K-Ohm

....................0...........1
BOM0 -> R2305+R2304

BOM1 -> R2307+R2306

BOM2 -> R2309+R2308

BOM3 -> R2311+R2310

When the resistor on the 0 side of a BOM jumped, the BOM is set to 0. When the resistor on the 1 side of a BOM jumped, the BOM is set to 1. If both the 0 side and 1 side is jumped the PowerBook will not function properly.

Voltage
All voltage pins are 470K-Ohm.

...............0..........1
D0 -> R3988+R3989

D1 -> R3986+R3987

D2 -> R3984+R3985

D3 -> R3982+R3983

D4 -> R3980+R3981

When only the resistor on the left of the chart is jumped, the pin is set to 0. When the only resistor on the right side of the chart is jumped, the pin is set to 1. Example: D4 is set to 0 if only R3980 is jumped. D4 is set to 1 if only R398 is jumped.


17" Aluminum:
I will be attempting an overclock when my 17" arrives.

iBook:

Late 2003 and Early 2004


Late 2004 and Mid 2005
None yet, you can help by posting close up pictures of every part of your motherboard.

Mac Mini


eMac


G5

 

[Swampus]

MDD Dual:
Multiplier and voltage:
http://xlr8yourmac.com/systems/g4_mirrored_drive_doors/G4_MDD_CPU_Module.html

MDD Single:
None yet: You can help by posting pictures of both sides of your CPU daughter card.

The same table can be used for single and dual CPU MDD modules. Blacksheep (from Poland, maybe same as 666sheep on this site?) wrote about it at Aquamac.

These machines aren't speed demons any more. I'm not sure that it makes a lot of sense to be overclocking these aging machines. Some processors are better candidates than others. 7455B, but not 7455A, for example. Often, you risk stability (or worse) for only a marginal speed increase. I've got one running at 1.58GHz that I'm planning to drop down to 1.25 when I get a chance. That's plenty for what the machine is currently used for and it will run cooler.

Downclocking can also be useful at times, so these tables are handy. It might also be wise to include a disclaimer about reasonable expectations. I saw a post not long ago on Apple's forums where someone tried to get 2.1GHz out of his MDD and was surprised that he fried his CPU.

 

[rabidz7]

The same table can be used for single and dual CPU MDD modules. Blacksheep (from Poland, maybe same as 666sheep on this site?) wrote about it at Aquamac.

These machines aren't speed demons any more. I'm not sure that it makes a lot of sense to be overclocking these aging machines. Some processors are better candidates than others. 7455B, but not 7455A, for example. Often, you risk stability (or worse) for only a marginal speed increase. I've got one running at 1.58GHz that I'm planning to drop down to 1.25 when I get a chance. That's plenty for what the machine is currently used for an it will run cooler.

Downclocking can also be useful at times, so these tables are handy. It might also be wise to include a disclaimer about reasonable expectations. I saw a post not long ago on Apple's forums where someone tried to get 2.1GHz out of his MDD and was surprised that he fried his CPU.

That should not have fried the CPU if he didn't do a crazy voltage increase.

 

[666sheep]

iMac G3 info from OP does not cover all models. 600 and 700 MHz ones have different PLL resistors location and setup. IIRC their voltage regulators also do differ from DV ones. Useful info about these here:
https://forums.macrumors.com/threads/1545868/

MDD daughtercards have 2 different voltage regulators. Each has slightly different Vid resistors setup. Famous bitsandpieces.info had more complete info than xlr8yourmac.
http://web.archive.org/web/20090301000742/http://bitsandpieces.info/Multipliers.htm

Single CPU MDD resistors location:
http://aquamac.proboards.com/index.cgi?board=talk&action=display&thread=767

MDD Bus overclock (better to use french version, there is an error in english one):
http://www.macbidouille.com/articles/89/page2

iMac G4 link from OP also does not cover all models. It's for USB 2.0 only.
100MHz bus Lampshade overclock info is here:
http://thundernerd.com/iMacG4Chart.html

Quicksilver info is incomplete as well. No info about dual CPU daughtercards.
It's not much of a loss, because DP cards don't overclock well and they're running very hot.
I've OC'ed dual 800 to 933 and it was PITA to keep it cool enough to stay stable. Certainly not worth the effort.
Edit: found link, I knew I had it somewhere:

Quicksilver DP multipliers:
http://www.applefool.com/clockchipping/pmg4_1000_1200.html

I saw a post not long ago on Apple's forums where someone tried to get 2.1GHz out of his MDD and was surprised that he fried his CPU.

Any chance for link? It's interesting what CPU he was trying. 7455 won't go that high.

 

[Swampus]

Any chance for link? It's interesting what CPU he was trying. 7455 won't go that high.

His thread is here: https://discussions.apple.com/thread/3641272

But it wasn't until the same person posted in this thread a month later that he admitted that he fried the CPU (3rd post down):

https://discussions.apple.com/thread/3757186?start=0&tstart=0

He appeared to have been using an OEM dual 1.42 module. He said that he adjusted voltage, but didn't specify how. There is no polite way to say this, but I think he was being somewhat less than truthful. I can believe that a novice might try for 2.3GHz, but I don't believe that it would boot, much less be stable "for an hour or two." I've never heard of anything greater than 1.67 making it to the desktop. I do believe that he fried his CPU in this experiment.

 

[666sheep]

Thanks for links. I don't believe this guy. L3 chips of 1.42 daughtercard are rated for maximum 370 MHz. This particular CPU has 1:6 core-to-L3 ratio.
370 MHz means 2.2 GHz theoretical maximum CPU clock.
But the problem is, that 1.42 module refuses even to POST with 1.75 GHz @1.80V. Maximum possible voltage that LTC3732CG is able to supply is 1.85V. So I'd say BS

 

[skinniezinho]

thanks for all the info, if you don't mind I will post a link to this thread on the FAQ

 

[rabidz7]

thanks for all the info, if you don't mind I will post a link to this thread on the FAQ

Thanks!

 

[Nova77]

Thanks for links. I don't believe this guy. L3 chips of 1.42 daughtercard are rated for maximum 370 MHz. This particular CPU has 1:6 core-to-L3 ratio.
370 MHz means 2.2 GHz theoretical maximum CPU clock.
But the problem is, that 1.42 module refuses even to POST with 1.75 GHz @1.80V. Maximum possible voltage that LTC3732CG is able to supply is 1.85V. So I'd say BS

My MDD is overclocked to 1.58 (see sig) and it was unstable at high CPU load until I changed my PSU to a bigger and better one. (Well at least it hasn't crashed yet and it feels more responsive)

Is it possible that crashes are caused by the PSU not being able to give sufficient stable power to the CPUs when you increase voltage (rather than CPU cache crash)?

Lets not forget the MDD has a lowly 360 watt PSU while it is quite comparable in performance and specs to early G5s that have at least 600 watt PSU...

Give me your thoughts!

 

[Andropov]

My MDD is overclocked to 1.58 (see sig) and it was unstable at high CPU load until I changed my PSU to a bigger and better one. (Well at least it hasn't crashed yet and it feels more responsive)

Is it possible that crashes are caused by the PSU not being able to give sufficient stable power to the CPUs when you increase voltage (rather than CPU cache crash)?

Lets not forget the MDD has a lowly 360 watt PSU while it is quite comparable in performance and specs to early G5s that have at least 600 watt PSU...

Give me your thoughts!

It's possible, but at one point, it doesn't matter how big is the PSU because it's not the limiting factor anymore , and the L3 cache just cant't be clocked that fast. CPU's without L3 cache like 7447A or 7448 should be easier to overclock.

But sometimes, the processor just refuses to boot, it's not always about power or heat. For example, my Yikes G4 works at 450Mhz, but it refuses to boot at 500Mhz even when the heatsink and the processor is cold.

From 1.42Ghz to 2.5Ghz... it's kind of difficult. Almost impossible with stock cooling.

PS: Does anyone know how to over clock a G5? At least, in theory. How to change multiplier or FSB speed from the motherboard. Theoretically, because it seems like anyone has tried.

 

[GermanyChris]

Good call on the thread looks like it's starting to get the correct people talking.

 

[macuser453787]

But sometimes, the processor just refuses to boot, it's not always about power or heat. For example, my Yikes G4 works at 450Mhz, but it refuses to boot at 500Mhz even when the heatsink and the processor is cold.

Dig that. I have 4 fans on my (aftermarket) processor that I tried to OC to 2.0 (though it actually reported only 1.93), but even with all the extra airflow it wasn't having it, so I backed it down to 1.8. Runs rock-solid.

Does anyone know how to over clock a G5? At least, in theory. How to change multiplier or FSB speed from the motherboard. Theoretically, because it seems like anyone has tried.

I'm kinda curious to know about this too...

 

[Andropov]

Dig that. I have 4 fans on my (aftermarket) processor that I tried to OC to 2.0 (though it actually reported only 1.93), but even with all the extra airflow it wasn't having it, so I backed it down to 1.8. Runs rock-solid.

Aftermarket processors are designed to run at higher speeds, actually, one 7448 from PowerLogix run stock at 2.0Ghz. Extreme overclock is theoretically possible, in Intel processors people has been doing it for years. For example, the Pentium 4 was once overclocked to 8GHz using liquid nitrogen. Obviously we can't use liquid nitrogen for everyday cooling, but the "strange" fact is the processor doesn't even boot (in a Mac) when you overclock it too much.

Back to my Yikes G4, it was a 350Mhz G4 model. Moving jumpers I went to 400Mhz. 450Mhz then. The CPU was as cold as always, but the PowerPC chips back in 1999 generated very little heat. Then I tried 500Mhz. Nothing. But 450Mhz was rock solid. Why? I guess it had something to do with VCore, but I've never tried to manually desolder the resistances to overvolt it and confirm it. But someone has already tried it, reaching 650Mhz in the same machine. And here.


If the voltage goes up, the system is more stable, but generates more heat. You can only overclock (highly) a processor if you also increase the voltage, otherwise it'd be unstable or it won't boot (the last thing happens mostly in Macs, PCs usually boot but start doing funny things until they crash). But, increasing the VCore usually let you overclock the CPU just a bit, so you need to continue increasing the voltage... and it starts to generate a LOT of heat. That's why CPU's usually never came overclocked, just at their native speeds. Sometimes is possible to improve the performance a bit with a moderate overclock and a slight increase in the VCore, even with the stock cooling, but it isn't happen usually, because if it's so easy to do, Apple probably would have done it before.

The highest ratio in 7447A/7448 is 2433Mhz. It won't work unless you over volt it a lot. That's why I doubt is possible to go from 1,42Ghz to 2,45Ghz without first putting and insanely high VCore and second, installing a enormous heatsink with four or five fans on it running at 7,000+ rpm.

In the case of the G5, I'm curious because I have some ideas to improve cooling (peltiers, taking space from the PSU...I can explain it all if someone is interested), so I think a moderate overclock may be possible, but I'll probably be too noisy for everyday purposes or too expensive for a 10 year old computer.

DISCLAIMER
Here it's 2:20AM, so it's possible I've written a few incoherences and mistakes. I'll edit it tomorrow when I woke up.

 

[Lil Chillbil]

You hear me world, I Kull Stalder Stand before you today and say, that I will beat that record and get a pentium 4 up to 10ghz. I have like 15 different boards so I better get cracking on a way to engineer a super cooling system that would be both stable and cooler than liquid nitrogen

 

[Lil Chillbil]

Update: just blew up 6 of my motherboards once I hit around the 5-6.1ghz mark with the motherboard laying on the bottom of our dry outdoor freezer. with stock cooler fan at full blast along with the power supply and gpu. I just ran the keyboard and power cord and the vga cable in-between the rubber seal on the door...




anyone else wants to give this pentium 4 record a shot be my guest

 

[rabidz7]

You hear me world, I Kull Stalder Stand before you today and say, that I will beat that record and get a pentium 4 up to 10ghz. I have like 15 different boards so I better get cracking on a way to engineer a super cooling system that would be both stable and cooler than liquid nitrogen

Phase change, compress the air above supercritical, and let it decompress=air is near absolute zero, the air will frezee into dry ice at normal presssure, however, this is not normal dry ice, dry helium ice is -268.93C, dry carbon ice is -78C, dry carbon ice is the usual dry ice you find at the grocery.

----------

It's possible, but at one point, it doesn't matter how big is the PSU because it's not the limiting factor anymore , and the L3 cache just cant't be clocked that fast. CPU's without L3 cache like 7447A or 7448 should be easier to overclock.

But sometimes, the processor just refuses to boot, it's not always about power or heat. For example, my Yikes G4 works at 450Mhz, but it refuses to boot at 500Mhz even when the heatsink and the processor is cold.

From 1.42Ghz to 2.5Ghz... it's kind of difficult. Almost impossible with stock cooling.

PS: Does anyone know how to over clock a G5? At least, in theory. How to change multiplier or FSB speed from the motherboard. Theoretically, because it seems like anyone has tried.

If you want to help, post a picture of every iMac motherboard and every PowerMac daughtercard you see.

----------

You hear me world, I Kull Stalder Stand before you today and say, that I will beat that record and get a pentium 4 up to 10ghz. I have like 15 different boards so I better get cracking on a way to engineer a super cooling system that would be both stable and cooler than liquid nitrogen

And I will get ML running (sarcastic)... Believe me, I've tried... And failed, I think this project may end the same way.

 

[666sheep]

My MDD is overclocked to 1.58 (see sig) and it was unstable at high CPU load until I changed my PSU to a bigger and better one. (Well at least it hasn't crashed yet and it feels more responsive)

Is it possible that crashes are caused by the PSU not being able to give sufficient stable power to the CPUs when you increase voltage (rather than CPU cache crash)?

Lets not forget the MDD has a lowly 360 watt PSU while it is quite comparable in performance and specs to early G5s that have at least 600 watt PSU...

Give me your thoughts!

Possible. You have 7800GS and bunch of drives AFAIR? Most likely 12V line of MDD PSU didn't have enough amps to properly feed all components on very high load.
OCed CPU didn't help it for sure. 7455 TDP is 22 W per 1GHz on maximum load. Overclocked and overvolted 1.58 Dual daughtercard could suck ~80W on maximum load.

The highest ratio in 7447A/7448 is 2433Mhz. It won't work unless you over volt it a lot. That's why I doubt is possible to go from 1,42Ghz to 2,45Ghz without first putting and insanely high VCore and second, installing a enormous heatsink with four or five fans on it running at 7,000+ rpm.

Impossible even with nitrogen. Not with components of 1.42 daughtercard. Maybe CPU itself would be able to go beyond 2GHz, but all the rest holds it. And 7447/7448 has very little to do with 7455 overclocking capabilities

 

[skinniezinho]

Update: just blew up 6 of my motherboards once I hit around the 5-6.1ghz mark with the motherboard laying on the bottom of our dry outdoor freezer. with stock cooler fan at full blast along with the power supply and gpu. I just ran the keyboard and power cord and the vga cable in-between the rubber seal on the door...




anyone else wants to give this pentium 4 record a shot be my guest

LOOOL, do you know that processors have limits right?and not only processors but motherboards, rams,etc
Years ago AMD64s even had the "cold bug".
Even if in the pc market using Ram with Samsung TCCD chips (low voltage,high MHz,high clocks 3-4-4-8 as far as I can remember), good boards with voltage mods, great psus and extreme cooling (prommys or dry ice), great psus and lots of knowledge things didn't went to 10GHz why do would they do now?
Powermac motherboards don't have 1/10 of the adjusts (chipset voltage, pci lock,agp lock,timing settings for mems,fsb adjust in 1MHz step) of for example a DFI AMD motherboard, it makes the overclocking much much harder.
If someone still wants to overclock a G4 please:

Change the psu to something really stable:
Levicom
OCZ
Corsair
etc
In the budget line: LC Power

Grab a powermac with DDR and buy some "overclockable" sticks (I don't know if they work on macs)
-something with Samsung TCCD chips (I don't know if powermacs like Cl3, but at least you won't need voltage mods)
-something with winbond BH5 chips (tight timmings but like volts)

Grab a good cooling system

STUDY about overclock, study the powermac motherboards.
It may be funny but your best be is a quad g5

 

[seveej]

If I try to view this thread as a quest for pure knowledge, I find the OP as well as some of the latter posts highly respectable.

RGDS,

 

[Andropov]

Update: just blew up 6 of my motherboards once I hit around the 5-6.1ghz mark with the motherboard laying on the bottom of our dry outdoor freezer. with stock cooler fan at full blast along with the power supply and gpu. I just ran the keyboard and power cord and the vga cable in-between the rubber seal on the door...




anyone else wants to give this pentium 4 record a shot be my guest

How are you solving the condensation problem? Is the first step to do... otherwise, the cold air will become water when it touch the hot surface of the CPU/Northbridge and it'll make a short.

If you want to help, post a picture of every iMac motherboard and every PowerMac daughtercard you see.

I have one of the Cube... but none of G5s.

Front

Back

Those are 33Mpx photos combining four different pictures.

LOOOL, do you know that processors have limits right?and not only processors but motherboards, rams,etc
Years ago AMD64s even had the "cold bug".
Even if in the pc market using Ram with Samsung TCCD chips (low voltage,high MHz,high clocks 3-4-4-8 as far as I can remember), good boards with voltage mods, great psus and extreme cooling (prommys or dry ice), great psus and lots of knowledge things didn't went to 10GHz why do would they do now?
Powermac motherboards don't have 1/10 of the adjusts (chipset voltage, pci lock,agp lock,timing settings for mems,fsb adjust in 1MHz step) of for example a DFI AMD motherboard, it makes the overclocking much much harder.
If someone still wants to overclock a G4 please:

Change the psu to something really stable:
Levicom
OCZ
Corsair
etc
In the budget line: LC Power

Grab a powermac with DDR and buy some "overclockable" sticks (I don't know if they work on macs)
-something with Samsung TCCD chips (I don't know if powermacs like Cl3, but at least you won't need voltage mods)
-something with winbond BH5 chips (tight timmings but like volts)

Grab a good cooling system

STUDY about overclock, study the powermac motherboards.
It may be funny but your best be is a quad g5

Yes, some "overclockable" sticks work in Macs. I tried three 167Mhz sticks on the Cube and they worked fine (running at 100Mhz).

 

[Lil Chillbil]

I was half-joking about actually making the 10ghz mark. I had to get rid of a few pentium 4 boards had laying in the shop so I figured why not just see how high I could get them before I throw them out. but now i've got the pentium 4 population down to a healthy 8 so I won't be able to continue on my quest for a while

 

[Goftrey]

I was half-joking about actually making the 10ghz mark. I had to get rid of a few pentium 4 boards had laying in the shop so I figured why not just see how high I could get them before I throw them out. but now i've got the pentium 4 population down to a healthy 8 so I won't be able to continue on my quest for a while

I still don't understand how you were able to even get the thing running if you stuck it in a freezer.

Ice melts.

And I'm no expert but electronics tend not to mix well with water.

 

[MisterKeeks]

He would have to rig it like an LCS; without having direct contact between ice and CPU. Of course, that would undermine the effectiveness of the dry ice.

 

[Lil Chillbil]

I still don't understand how you were able to even get the thing running if you stuck it in a freezer.

Ice melts.

And I'm no expert but electronics tend not to mix well with water.

the freezer had no ice or water inside of it. it was a dry cold plastic bottom with cold air flowing thru it.

 

[Goftrey]

the freezer had no ice or water inside of it. it was a dry cold plastic bottom with cold air flowing thru it.

I call BS on that.

In freezers there are funky little things called evaporator coils - gas passes through these coils & due to the sheer coldness of them, causes any humidity in the freezer to freeze onto them, creating ice.