Hi, I dont suppose anyone knows if there are any noticeable differences in heatsinks from various models as I have option of 1.8 DP or 2.3 DP to provide donor heatsinks for a conversion. Apparently the 2.0 heatsinks need minimal cutting (see here https://forums.macrumors.com/threads/g5-quad-air-conversion-help.2346823/)
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
QUAD G5 AIR COOLED (dxf included)
- Thread starter MrVitalic
- Start date
- Sort by reaction score
with the older heatsinks there is only a difference with the cold pads where the CPU meets the heatsink. some are magnesium/nickel while others are made of copper. the copper ones being better of course.
copper example:
Magnesium/nickel example:
these heatsinks will work but the fit and finish is not as ideal as cutting down dual core heat sinks. during the G5 line apple did not seem to have a rhyme or reason for doing anything in particular. ideally you would want to see the copper cold pads on the higher clocked CPU's but i have seen copper cold pads on single 1.8ghz units stock and nickel on 2ghz. I have also seen the close quarters heat pipes (2.0ghz dual core supposedly) heat sink on 2.3ghz Dual core chips stock. So unfortunately it’s a real gamble which heat sink you are going to get until you take it apart. Apple must have had alot of parts lying around and just used what they could when they could.
Last edited:
This is my kind of thread.
I too have a thing for twins.
I recently started a thread about waterless coolant to be used in liquid cooled Macs. IMO, keeping everything stock and simply switching to waterless coolant would resolve any and all non-mechanical issues. Read along to see what I mean.
Otherwise this is an interesting project. If the temperatures are the same then why did Apple resort to liquid? I read the part about the paste but in 2005 I assume the paste in the air cooled Mac was as good as in the liquid cooled Mac.
To our machinist: invariably in car cooling components it is taken for granted that tubular radiators are vastly inferior to flattened ovoid pipes. Flat pipes allow more air flow while exposing more surface of the pipes to air. The benefits are substantial. Your cross-sectional obstruction to flow is significantly less and thus allows a greater volume of air at sustained velocity and laminar adhesion (less turbulence). I can imagine that by flattening your copper through the fin stack you would need fewer RPM's to maintain the same or even better temps. Use of a rolling mill would be preferred if the copper is merely solid. You could use thinner copper cores and have equal surface area to your current tubes once flattened.
Photo here shows hollow pipe but the principle is the same for solid pipes.
Regarding SSD's: I just bought a bundle of SSD's from a rando on CL and chucked all three into the G5. They work exactly as expected and one is now my boot drive. Here is the specs and test results;
PowerMac 7,2 2.0GHz DP
OCZ Vertex 3 (boot drive) SATA 3: SR 90, SW 90, RR 86, RW 90
Samsung MZ-5PA2560/0D1 SATA 2: SR 90, SW 90, RR 70, RW 70
PNY CS1311 SATA 3: SR 100, SW 93, RR 84, RW 77
I too have a thing for twins.
I recently started a thread about waterless coolant to be used in liquid cooled Macs. IMO, keeping everything stock and simply switching to waterless coolant would resolve any and all non-mechanical issues. Read along to see what I mean.
Otherwise this is an interesting project. If the temperatures are the same then why did Apple resort to liquid? I read the part about the paste but in 2005 I assume the paste in the air cooled Mac was as good as in the liquid cooled Mac.
To our machinist: invariably in car cooling components it is taken for granted that tubular radiators are vastly inferior to flattened ovoid pipes. Flat pipes allow more air flow while exposing more surface of the pipes to air. The benefits are substantial. Your cross-sectional obstruction to flow is significantly less and thus allows a greater volume of air at sustained velocity and laminar adhesion (less turbulence). I can imagine that by flattening your copper through the fin stack you would need fewer RPM's to maintain the same or even better temps. Use of a rolling mill would be preferred if the copper is merely solid. You could use thinner copper cores and have equal surface area to your current tubes once flattened.
Photo here shows hollow pipe but the principle is the same for solid pipes.
Regarding SSD's: I just bought a bundle of SSD's from a rando on CL and chucked all three into the G5. They work exactly as expected and one is now my boot drive. Here is the specs and test results;
PowerMac 7,2 2.0GHz DP
OCZ Vertex 3 (boot drive) SATA 3: SR 90, SW 90, RR 86, RW 90
Samsung MZ-5PA2560/0D1 SATA 2: SR 90, SW 90, RR 70, RW 70
PNY CS1311 SATA 3: SR 100, SW 93, RR 84, RW 77
That Samsung is a 470. The sequential results in particular seem a little low even for SATA I, but we know benchmarks aren’t the be-all-end-all.
Interesting. What should we expect from a G5?
My only concern with this drive is junk collection. Not sure how that would be handled or how long before performance takes a dive. It’s performing sweetly enough and as far as PPC stuff I would not ask for better than this.
My only concern with this drive is junk collection. Not sure how that would be handled or how long before performance takes a dive. It’s performing sweetly enough and as far as PPC stuff I would not ask for better than this.
#26
The Samsung 470 (2010) is known to handle that quite well by itself. Apple used it in the 2010 MBA, before OS X TRIM’ed, for a reason and I also used it in my Hackintosh back in the day, without TRIM.
I’d probably make this one my boot drive.
No idea about the other drives. The OCZ Vertex 3 (2011) uses a SandForce SF-2281 controller, the PNY CS1311 (2016) uses a Phison.
Last edited:
According to that post, I should *max* out at 130Mb/s. I'll take the consistent 90s I'm getting on my OCZ, then. Can't imagine the performance increase is worth any modding etc.
Since you pointed out the controller on the OCZ, it revealed another benefit. The specific SandForce controller has garbage collection below the file system level so they are preferred where TRIM is not enabled! Got a keeper.
Matches my result 135 MB/s
SATA I max is 150 MB/s without accounting for overhead (1.5 Gbps * 8b/10b) so 135 MB/s is reasonable.
Perhaps there is some loss when switching to different interfaces in these drives. All I can do is speculate and show test results. Here they are in my Mac Pro with in-built SATA II.
OCZ Vertex 3 (boot drive) SATA 3: SR 150, SW 120
Samsung MZ-5PA2560/0D1 SATA 2: SR 160, SW 153
PNY CS1311 SATA 3: SR 162, SW 113
OCZ Vertex 3 (boot drive) SATA 3: SR 150, SW 120
Samsung MZ-5PA2560/0D1 SATA 2: SR 160, SW 153
PNY CS1311 SATA 3: SR 162, SW 113
Just to go back to the OP's initial thought, is it possible at all to put a DC/Quad mobo in a DP case? Likely the PSUs/PSU connectors don't match, right, along with other connectors/case supports (e.g. Airport/BT etc) as well as fans? Or using DC CPUs in a DP, is this a firmware/mobo or physical/electrical limitation?
What about upgrading a PCI DP to PCI-X? Would that work without any issues?
What about upgrading a PCI DP to PCI-X? Would that work without any issues?
Last edited:
from what i understand, the dualcore machines are pci-e, and have different mounts and different rear connection openings. so unless you want to hack&slash a dual processor case to fit a late 2005 logicboard...
as for your second part question, and pci to pci-x boards should be cross compatible. i was able to take a case from a 1.8 and with a logicboard from a dual 2.3 ghz and put in dual 2.5 ghz cpu's
Th
So apparently a 2.5Gz CPU would also work on a 2.0Ghz mobo, air-cooled:
I am wondering about 2.7Ghz for my DP 2.0GHz as I have an LCS unit with 2x2.7Ghz G5s that I had bought as LCS spare for my G5 quad (even though it turns out that the LCS unit is not directly compatible with the Quad, at least not w/o modification as the connectors have different numbers of pins). But 2.7GHz might be pushing it a bit running on heatsinks only, is there any way to hook up an LCS to a DP 2.0GHz mobo/case? Or do I need a 2.5GHz mobo, or do they all have connectors for LCS units? Otherwise I would get a 2.7GHz DP mobo (or 2.5GHz) as those are LCS-enabled, but I am not sure about fitting the LCS in a 2.0 DP case?
Thanks!
So apparently a 2.5Gz CPU would also work on a 2.0Ghz mobo, air-cooled:
Successful CPU swapping G5
After much research I just want to let everyone know that it is possible to upgrade your PowerMac G5 tower's CPU(s) The only requirements are that the processors have to be in the same family 970,970fx etc, as well as the model Identifier 7,3 7,2 etc. Ive gotten my Dual 2GHz G5 to successfully...
www.mac-forums.com
I am wondering about 2.7Ghz for my DP 2.0GHz as I have an LCS unit with 2x2.7Ghz G5s that I had bought as LCS spare for my G5 quad (even though it turns out that the LCS unit is not directly compatible with the Quad, at least not w/o modification as the connectors have different numbers of pins). But 2.7GHz might be pushing it a bit running on heatsinks only, is there any way to hook up an LCS to a DP 2.0GHz mobo/case? Or do I need a 2.5GHz mobo, or do they all have connectors for LCS units? Otherwise I would get a 2.7GHz DP mobo (or 2.5GHz) as those are LCS-enabled, but I am not sure about fitting the LCS in a 2.0 DP case?
I need some advice! 
I've done the same conversion on my Quad G5 with 2 of the better air-cooled heatsinks. However, the machine is not passing thermal calibration (gives me a Max temp exceeded on both CPUs). In OSX it seems to run perfectly fine and stable. When pushed the temps are a little on the high side; Idle hovers around 37-42c (with automatic mode). When I push it with a couple of Geekbench tests or games the performance is great but the fans are full speed (3000-3200 RPM) around 65c with peaks of up to 76-77c. I've used fresh Noctua thermal paste and generally the air cooled setup seems fine (I went the hacksaw route so the heatsinks aren't as clean as a laser cut... but they generally fit well and look ok'ish). Any ideas to make my Quad run a bit better to pass the thermals? What temp do I need to get rid of the warning and is my Quad fine for use without the calibration?
I've done the same conversion on my Quad G5 with 2 of the better air-cooled heatsinks. However, the machine is not passing thermal calibration (gives me a Max temp exceeded on both CPUs). In OSX it seems to run perfectly fine and stable. When pushed the temps are a little on the high side; Idle hovers around 37-42c (with automatic mode). When I push it with a couple of Geekbench tests or games the performance is great but the fans are full speed (3000-3200 RPM) around 65c with peaks of up to 76-77c. I've used fresh Noctua thermal paste and generally the air cooled setup seems fine (I went the hacksaw route so the heatsinks aren't as clean as a laser cut... but they generally fit well and look ok'ish). Any ideas to make my Quad run a bit better to pass the thermals? What temp do I need to get rid of the warning and is my Quad fine for use without the calibration?
Last edited: