Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Glad to hear that everything worked out! I think the issue with the fans is mostly the fact that they run at 12V continuously, but there are fan voltage adapter cables that basically have a resistor built in and drop the voltage from 12V to something like 7V (they used to come with Noctua fans, but I'm sure cheaper non-branded alternatives exist as well). One would need to try how this affects the temperatures under sustained load though.

I guess the ideal solution would be to create some sort of adapter cable from the actual LCS cable harness, but that would be quite the project 😅
Thank you! I've considered installing some of those resistor cables, but I've kind of got acclimated to the fan noise now. My systems sit on my desk about 2 feet off to the side of my face. I'm sure it would be much quieter under the desk. I have a set of Arctic fans I may swap in to test out eventually. I'm just thrilled that the system is 100% stable and I can run it on full load if need be with no worries and stay well below thermal limits. It has been rock solid since the switch to air cooling.

As much as I love my dual core 2.3, (late 2005) that has been doing backup duty, the Quad is noticeably a bit faster when it isn't constantly throttling itself. A few more cores, more cache and a few more mhz matter it seems. I am enjoying the peace of mind not worrying this thing may spring a leak at any moment! On mine, the actual pump motor was slowly leaking out just enough to evaporate before ever dripping. It had a light coating of dried coolant crystals on the bottom of the pump. Over the years, enough evaporated away to cause problems.
 
Your logic board likely does not have the connector for a second processor, and your 710w power supply (unless it was repaired under warranty and replaced with a 1kW one) won't have power connectors for a second processor module.

At best you could install a single G5 Quad processor module to build a 2.5 GHz DC.
 
I guess the ideal solution would be to create some sort of adapter cable from the actual LCS cable harness, but that would be quite the project 😅

The best solution would probably be to have a tiny board which grabs the PWM signal from the original G5 front fans (or back fans, whichever works best) and adapts it to drive regular PC fans, so that the fan speed actually depends on the CPU temp. Depending on what that voltage is and how it works logically, you might even get away with a simple two-resistor voltage divider.

Though it would be most optimal to use a microcontroller which would interpret the PowerMac's PWM signal and then generate its own, since the PC fans obviously have different min and max speeds and noise characteristics, so simply using the PWM signal will probably result in some weirdness.
 
I'm no expert on the G5 but I currently have a 2.3GHz DC (late 05) and EveryMac lists the CPU as being on a daughter card? Is possible to swap it to a Quad by simply changing the daughter card or adding in a second G5 DC?

Take it from a man who spent some considerable time trying to work this out, the best TL;DR answer at this moment is that it is simply not feasible.

The full non-TL;DR answer is as follows:
  1. The heatsink for the DC board would need to be reworked to a considerable extent, which people have done to convert their Quads to air cooling, but it's quite difficult.
  2. Even if you reworked two heatsinks for the two DC boards you would need, there is no second processor socket on the DC board.
  3. You could order the socket from somewhere like Mouser or DigiKey, though it is usually either out of stock or unreasonably expensive for just a little plastic doodad.
  4. Even if you ordered one, you would have a lot of trouble soldering it onto the motherboard, because those motherboards are the worst to solder to. After the fiasco with the self-destructing RAM slot joints, Apple made Late 2005 PowerMac boards out of what feels like solid copper, so that they soak up as much heat as possible, which means that you would need to preheat the heck out of the board.
  5. Even if you manage to solder the socket to the board, I am told that boards which have experienced a lot of repeated thermal stress over the years (which the PowerMac boards 100% have) experience degradation in a way where soldering anything to or from them effectively makes them self-destruct, so the likelihood of success is very low.
  6. Even if you go through all of that, and still have a working PowerMac G5 motherboard, which is a big if, whether or not the second CPU gets detected and initialized is probably determined by the SMU, and that thing probably knows that it's a DC PowerMac and that it isn't supposed to have a second CPU at all. So most likely you'll have to somehow modify the SMU firmware, which I'm not even entirely sure is possible to do, since the microcontroller that is the SMU has firmware encryption capabilities. Granted, it uses 2005 encryption standards with a less-than-optimal encryption key length. But still, it is enough to make you waste centuries trying to brute-force it.
 
At best you could install a single G5 Quad processor module to build a 2.5 GHz DC.

Edit: my original post (remains below) has been disproven, see Matias_'s message below.

Not even that, I'm afraid. I have obtained with a non-working 2.0 GHz DC machine what I'm pretty sure is a 2.3 GHz processor module that was seemingly swapped into it at some point, at least judging by its heatsink. And it ain't running at 2.3 GHz when I install it into a working 2.0 GHz DC machine. The processor speed is determined by the bus speed, and the bus speed seems to be hard-coded into the motherboard firmware (which one, the SMU or OpenFirmware ROM, I'm not sure exactly, probably the SMU since it has to initialize the northbridge).
 
Last edited:
Not even that, I'm afraid. I have obtained with a non-working 2.0 GHz DC machine what I'm pretty sure is a 2.3 GHz processor module that was seemingly swapped into it at some point, at least judging by its heatsink. And it ain't running at 2.3 GHz when I install it into a working 2.0 GHz DC machine.
I have seen people do this, it appears to be possible 🤷‍♂️



The processor speed is determined by the bus speed, and the bus speed seems to be hard-coded into the motherboard firmware (which one, the SMU or OpenFirmware ROM, I'm not sure exactly, probably the SMU since it has to initialize the northbridge).
I do not believe this to be the case for the PCIe models. Again, there are several successful reports of performing this type of upgrade on the internet. And I believe I have seen at least one person convert a liquid cooled Quad 2.5 GHz to an air cooled FrankenQuad 2.3 GHz.
 
Last edited:
  • Like
Reactions: Nullcaller
I have seen people do this, it appears to be possible 🤷‍♂️

I do not believe this to be the case for the PCIe models.

Huh. Thanks a lot for linking all of this! All I can say is that it seems like I've indeed made a mistake. I apologize. Assumptions, assumptions. It seems like I am physically incapable of being careful when making assumptions.

But anyway, that's good news! (I mean, really, that's only news to me, but bear with me) If the Late 2005 PowerMacs can all dynamically recognize the CPU boards connected to them and adjust the initialization accordingly, maybe you can just solder an additional socket onto a DC board and turn it essentially into a quad board.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.