-
Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.
G5 2 GHz dual in the early 2005 dual 2.7 GHz
- Thread starter bengi3
- Start date
beachycove
Well-known member
As an experiment, I once successfully swapped 1.8 processors from (as I recall) a four-RAM-slot machine from 2004 into what had been a 2.0 eight-RAM-slot machine from 2003 (one of its processors was dead). The machine needed to be calibrated, which any processor swap whatsoever entails in a G5, but it worked just fine afterwards. I don't have it anymore so can't check the details, but I posted on here about it somewhere, probably about two years ago. If you search, it will come.
What I don't think possible is use of the final G5 processor in the earlier machines. The earlier processors may well all be swappable -- despite what people say. You will, however, need the relevant Apple Service Diagnostics disc to do the calibration (there are two versions, one of which is specific to the very last of the G5s). Refreshed heatsink compound will work wonders, too.
What I don't think possible is use of the final G5 processor in the earlier machines. The earlier processors may well all be swappable -- despite what people say. You will, however, need the relevant Apple Service Diagnostics disc to do the calibration (there are two versions, one of which is specific to the very last of the G5s). Refreshed heatsink compound will work wonders, too.
Last edited by a moderator:
yes, I am well aware of the thermal calibration having performed it several times. I have ASD for 2003, 2004 and 2005 PM G5.
This time I would try to make one working G5 out of a bunch of dual 2.7 carcasses but without knowing if its the logic board or the processor that are busted I am groping in the dark.
I have a working 2004 G5 2x2 GHz so I would like to do some experiments.
This time I would try to make one working G5 out of a bunch of dual 2.7 carcasses but without knowing if its the logic board or the processor that are busted I am groping in the dark.
I have a working 2004 G5 2x2 GHz so I would like to do some experiments.
Franklinstein
Well-known member
I hate the Delphi LCS because it wasn't a great design in the first place and wasn't intended to be serviced, so I'd prefer to get rid of it. Some of the dual 2.5s had a Panasonic LCS and they were much better, but I didn't have any of those to swap over. Even if I did I doubt I would have done so because that system will still ultimately need maintenance that may not go well.
Here's what I did with my dual 2.7:
I disassembled the heatsink assemblies from a couple bad 2.0 processor modules and attached them to the 2.7 modules. I reinstalled the new air-cooled 2.7 modules and also a bodged LCS cooling pump. Why the cooling pump? Diagnostics and thermal recalibration fail on these if the LCS pump isn't running. So what I did was create a loop with the old hose, fill the pump with water, and set it out of the way as much as possible during testing. Once the pump's electrical connector was attached and the air deflector installed, I booted the Apple diagnostic disc and ran the basic suite of logic board tests to make sure everything relevant was properly recognized. Then I ran thermal calibration, which passed. After that, I shut it down, disconnected and removed the LCS pump, and closed up the case. It rebooted in OS X and has been running fine since. The fans rev up more frequently and a bit louder than they used to, but CPU temperatures are kept reasonable and there are no system errors or crashes like it used to have with the old LCS. Under normal use the computer does not miss the LCS pump.
I had considered doing the same with the Quad G5, but unfortunately the 970MP runs much hotter than the 970FX so it's not a great idea. Also the air heatsink for the single dual-core 2.0 and 2.3 modules is much larger than that of the old single-core modules, so it's not possible to fit two of them in the same case. So eventually I'll have to rebuild the Quad's LCS.
Here's what I did with my dual 2.7:
I disassembled the heatsink assemblies from a couple bad 2.0 processor modules and attached them to the 2.7 modules. I reinstalled the new air-cooled 2.7 modules and also a bodged LCS cooling pump. Why the cooling pump? Diagnostics and thermal recalibration fail on these if the LCS pump isn't running. So what I did was create a loop with the old hose, fill the pump with water, and set it out of the way as much as possible during testing. Once the pump's electrical connector was attached and the air deflector installed, I booted the Apple diagnostic disc and ran the basic suite of logic board tests to make sure everything relevant was properly recognized. Then I ran thermal calibration, which passed. After that, I shut it down, disconnected and removed the LCS pump, and closed up the case. It rebooted in OS X and has been running fine since. The fans rev up more frequently and a bit louder than they used to, but CPU temperatures are kept reasonable and there are no system errors or crashes like it used to have with the old LCS. Under normal use the computer does not miss the LCS pump.
I had considered doing the same with the Quad G5, but unfortunately the 970MP runs much hotter than the 970FX so it's not a great idea. Also the air heatsink for the single dual-core 2.0 and 2.3 modules is much larger than that of the old single-core modules, so it's not possible to fit two of them in the same case. So eventually I'll have to rebuild the Quad's LCS.
Thank you Frankl,
I did some experiments with a known good 2.7 logic board.
The 2.0 GHz processors does work on 2.7 logic, unfortunately only one was good but the Mac needs both processors in place to power on. So I booted into open firmware and luckily I managed, before the system eventually crashed, to reboot with only one processor (for the less experts: setenv boot-arg cpus=1), see picture.
Then I have a spare 2.0 GHz processor: this one has the plastic that holds the connector in black rather than the usual whitish (see picture) and while it works with the logic board of the June 2003 dual 2 GHz, is not accepted by the 2.7. I believe this is a later 2.0 model and the incompatibility can depend on a different multiplier (bus ratio).
First picture:
PowerMac G5 2.7 with 2.0 processors booted off only one CPU:
Second and third pictures:
Black connector and processor:
I did some experiments with a known good 2.7 logic board.
The 2.0 GHz processors does work on 2.7 logic, unfortunately only one was good but the Mac needs both processors in place to power on. So I booted into open firmware and luckily I managed, before the system eventually crashed, to reboot with only one processor (for the less experts: setenv boot-arg cpus=1), see picture.
Then I have a spare 2.0 GHz processor: this one has the plastic that holds the connector in black rather than the usual whitish (see picture) and while it works with the logic board of the June 2003 dual 2 GHz, is not accepted by the 2.7. I believe this is a later 2.0 model and the incompatibility can depend on a different multiplier (bus ratio).
First picture:
PowerMac G5 2.7 with 2.0 processors booted off only one CPU:
Second and third pictures:
Black connector and processor:
Last edited by a moderator:
demik
Well-known member
Good job !
2.0 GHz G5 have a tendency to dies. I got a pile of them back in my lab. There is a few different 2.0 GHz models :
- PowerMac7,2 (original G5) are 970 processors
- PowerMac7,3 (really 2005) are in fact 970fx Processors
2.7 GHz were 970fx, If I remember correctly an early motherboard won't work with fx processors (don't quote me on this though)
2.0 GHz G5 have a tendency to dies. I got a pile of them back in my lab. There is a few different 2.0 GHz models :
- PowerMac7,2 (original G5) are 970 processors
- PowerMac7,3 (really 2005) are in fact 970fx Processors
2.7 GHz were 970fx, If I remember correctly an early motherboard won't work with fx processors (don't quote me on this though)
I did check, my 2.0 GHz are all 970, none of them is FX. So far I have 5 couples of stone dead 2.7 and another bunch of dead 2.0. What a disaster has been this G5 project!
demik
Well-known member
I heard you. Out of my 8 G5 collection, 3 died because of CPU / Logic board.
- One of them went through TWO logic boards and a two pairs of CPUs, dunno if that make 4 dead.
- 2 Nvidia GPUs blown
- Coutless of DDR sticks burned
Lately one of them is becoming flaky, but seems a recap will save it. Reliability for PowerMac G5 is really bad. If you don't know, here is a good article here: https://lowendmac.com/2014/which-power-supply-works-in-my-power-mac-g5/
Most reliables here ones here are the 2005 2.3 DP and single 1.8s (HDD died, but this is ok)
G5 CPUs and Logic Boards are dying left and right :/
- One of them went through TWO logic boards and a two pairs of CPUs, dunno if that make 4 dead.
- 2 Nvidia GPUs blown
- Coutless of DDR sticks burned
Lately one of them is becoming flaky, but seems a recap will save it. Reliability for PowerMac G5 is really bad. If you don't know, here is a good article here: https://lowendmac.com/2014/which-power-supply-works-in-my-power-mac-g5/
Most reliables here ones here are the 2005 2.3 DP and single 1.8s (HDD died, but this is ok)
G5 CPUs and Logic Boards are dying left and right :/
demik
Well-known member
Hmmm... Something fishy there. Guessing you did try with minimum RAM installed. Did you try a Linux CD ? I would guess it will fail too, but where is the interesting part. Where does Mac OS fail in verbose mode ?
You can find ISO online, for example ubuntu Mate provided a Linux CD for PowerMac computers :
http://cdimage.ubuntu.com/ubuntu-mate/releases/16.04/release/
You can find ISO online, for example ubuntu Mate provided a Linux CD for PowerMac computers :
http://cdimage.ubuntu.com/ubuntu-mate/releases/16.04/release/
I am running right now the ASD tests. RAM yes, two sticks 256 for a total of 512.
I was thinking of trying Morphos, but will also try Linux and verbose.
Will keep you posted.
Some results:
ASD: all tests passed
Booting in verbose mode says: overtemp signal
Connceted the pump (this one in the picture is one of the LCSs that I have fixed): the boot process arrives at a blue screen but cannot get to the desktop and the fans become quiter, so I immediately powered it off.
Tried again the thermal calibration but the software says that CPU 0 and CPU 1 have wrong register: I believe that the hybrid thing (7,3 logic and 7,2 processors) is not a good idea.
ASD: all tests passed
Booting in verbose mode says: overtemp signal
Connceted the pump (this one in the picture is one of the LCSs that I have fixed): the boot process arrives at a blue screen but cannot get to the desktop and the fans become quiter, so I immediately powered it off.
Tried again the thermal calibration but the software says that CPU 0 and CPU 1 have wrong register: I believe that the hybrid thing (7,3 logic and 7,2 processors) is not a good idea.
Last edited by a moderator:
demik
Well-known member
Similar threads
