Wait, the FP imac G4 had a 7441 and then 7445 for the entire line. So the only real upgrades for it are 7447 and 7448.
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.
G4 7448 and 7457 swap
- Thread starter JoyBed
- Start date
- Sort by reaction score
I recommend giving my thread a read
first gen iMacs (and eMac's) for that matter use 483 CBGA CPUs as Apple toyed with the idea of including L3 cacheuncooking the sunflower seeds, first generation iMac G4 repair/CPU replacement
some of the long timers here may remember my post here https://forums.macrumors.com/threads/show-us-your-chips.2127981/?post=26273568#post-26273568 about an iMac G4 I had with a very dead CPU, now while this would normally be the end of the line for such a machine or its logic board, I did...
forums.macrumors.com
mmmm, it also looks like there is a unused place for a L3 cache on the motherboard, it would be cool for a imac g4 to have a 7457 with a L3 4mb cache.I recommend giving my thread a read
uncooking the sunflower seeds, first generation iMac G4 repair/CPU replacement
some of the long timers here may remember my post here https://forums.macrumors.com/threads/show-us-your-chips.2127981/?post=26273568#post-26273568 about an iMac G4 I had with a very dead CPU, now while this would normally be the end of the line for such a machine or its logic board, I did...
It was a 7447 but there is essentially no difference, other than the 7457 supports an L3 cache, it think.
They share the same PVR, so that made me think it was the 7457, but Daystar did make some 7457 upgrades. Tho, oddly, the NVRAM script form Daystar included in the iMac upgrade changes the " cpu-version" property of both /cpus/@0 and /cpus/@1, so they must have reused the same NVRAM script for multiple upgrades.
The iMac G4 FP upgrade software for OS 9 also installs the XLR8 CPU Plugin, but I need someone to confirm that a 7447/7457 can enter deep sleep and wake from deep sleep under OS 9.
Also, Sonnet's 7447a, was deep sleep working on desktop systems under OS 9 with Sonnet's CPU Plugin?
It's just really hard to debug why OS 9 won't wake from deep sleep, at this time I'm thinking it has to do with the PLL, which should be disabled for "deep sleep" on the 7447/7457/7447a.
I know the system( iBook G4 ) has not crashed in sleep mode, because I set a wakeup time, and the machine tries to wake at that time. Also if the battery runs critical low, the system will enter deep sleep, and not respond to wakeup events until power is plugged in.
I believe increasing the clock speed in your machine is only possible by playing with physical jumper settings. The patches are, I believe, so that your laptop recognizes the new 7457 chip appropriately. You can apply this patch in advance if you plan to send your laptop to someone (like @JoyBed) to solder the 7457 into your machine.
Also, I was just reading all sorts of 74xx documentation and reports, and I heard that the 7457 (and I believe also 7447 and 7448) introduced 200MHz bus support. Now, we already know all MDDs can have at least 167MHz bus speed, and there were talks of changing the crystal oscillator to allow a sliiiightly higher speed, but has anyone dared to attempt just overclocking the stock MDD to 200MHz bus speed just like that, provided a 7457/7447/7448 is installed? (Not the stock 7455, which I assume wouldn't work with such high bus speeds.)
Regarding the 7455, regardless if Mac OS 9 is stuck or not with those as far as dual CPUs are concerned, I noticed something interesting: the highest base clock speed for the 7457 is 1.267GHz, while for the 7455 it is 1.333GHz (with specification for a 1.4GHz version being published, then rectified as a mistake found in the initial release of the specification, as seen at the end of the PDF document). Now, that's a minuscule difference, and even so it could be that perhaps the 7457 is more overclockable than the 7455 anyway (who knows), but I couldn't help, but notice that.
And speaking of the 7457, as we all know it upgrades the L2 cache from 256kB to 512kB, but I also noticed that it is mentioned the L3 cache "Total SRAM space supported" and "Direct mapped SRAM sizes" can be as high as 4MB, instead of 2MB, but notes that out of the 4MB, "a maximum of 2 MB can be configured as cache memory; the remaining 2 MB may be unused or configured as private memory". Does this translate to some performance gain for us Mac OS and Mac OS X users, or is this just a potential of the chip that is not exploited by Macs?
Edit: Whoa, lots of new posts since I last refreshed. My apologies for some redundancy here and there.
The datasheet of the 7457 is explaining all of that actually. The L3 cache and its capacity and its layout config. I cant tell how much will the OS X benefit from the extra 2mb but the “private space” means that the CPU is using it for storing some variables or some code that isnt needed for now but will be needed in a while so it doesnt need to fetch it thru 167mhz bus from slow HDD but rather from fast L3 cache.
EDIT: Yes, the 7457 has bigger internal cached and better pipelines and can execute and fetch more instructions at once and also its manufactured with lower nm process so less power, less heat, higher clocks.
Last edited:
The original iMac G4 FP had a 7450( 700/800Mhz ) and Daystar did in fact make 1.3Ghz 7457 upgrades for those.
The issue with this is - time and expense for just a minuscule increase in performance - I am sure there has to be someone in the USA(now UASS) who knows how to do this, Slovakia is far from the country and I am sure to ship plus all the customs fees will be even more than having it modded locally.
There is, actually. I just don't know if he's open to do it for everyone (it's done more as a hobby rather than for profit), but I will try asking him about it once he's more available (maybe in a month or two from now).
"Minuscule" is not quite right. That would be, like, bumping 1.42Ghz to 1.5Ghz on some given processor, rather than up it by 50% in clock speed, increase L2 cache and L3 cache. "Monstrous" is more accurate! But of course, it being worth it or not purely depends on your own personal aspirations. A 1GHz PowerBook is a good 'Book.
One thing I would like to see is the G4 accepting 4GB of memory. Yes, the 1Ghz PB Titanium I have is nice with the M2 sata SSD I have and now I know drives higher than 128GB are supported, I may get a higher drive/OS 9 still uses 128GB. Would that person be Dosdude ?
If someone can get me the files of a MDD about the schematics and board layouts, than I will remake it to ise PCI-E and DDR2 and mire than 2Gb of RAM.
The hunt for MDD schematics starts. Now.
To be honest, it's about time we have schematics for boards onto which we can flash ROMs for Macs we already own for our own personal use. I too have been thinking how great it would be to be able to "fix" damaged MDD boards by ordering new compatible boards from a PCB manufacturer. No shortage of MDD-like boards anymore.
Do you think upgrading the Ultra ATA/100 (ATA-6) connectors to Ultra ATA/133 (ATA-7) would also be possible? That's what 1.33GHz Xserve G4s use. And the crystal oscillator, for higher system bus speeds (which would also allow faster RAM)?
Well, I guess those are questions for later. Schematics first. We shall preserve (and upgrade?) the architecture.
Yeah, I suggested ATA-7 purely because of OS 9. Compatible SATA cards just "trick" it to think it's a SCSI interface, because of lack of proper SATA drivers. But we can always have multiple variations of the board design: OS-X-centric design, and OS-9-centric design.
Also, I think the following is appropriate for our quest:
Also, I think the following is appropriate for our quest:
Yes, why not, its just about using a different I/O chip so two designs with different connectors for HDDs are not a problem. Just those schematics are needed. I have the DLSD schematics but U dont want those, I want MDD.
I'm still searching for the schematics. By the way, speaking of the DLSD, do we know why its bus can be overclocked to 189MHz, but the MDD's bus can't? Is it the crystal oscillator? Is it other chips in the motherboard?
Its because the MDD is supposed to have a 7448 in it, not a 7447. The schematics and the writing and descriptions are refering to the 7448 not the 7447. Its interesting that it was sold with 7446. Even after checking the firmware the part that checks CPU id’s have a 7448 id in it.
I think you meant to say "DLSD" instead of "MDD", and "7447" instead of "7446". But I see, thank you, that answer makes sense.
Incidentally, I believe I looked everywhere, and while I found PowerBook and iMac schematics, and 68k Mac and even earlier Apple hardware schematics, I have found no PowerMac G4 schematics or board view files. I suspect they sadly are not available, which leaves us with reverse-engineering new schematics as the only choice...
Ah, yes. Sorry i meant DLSD and 7447 as you said. Reverse engineering the MDD would be real pain in the rear tho.
Incidentally, does anyone know if there are any technical differences between the 7448 chips labelled as "SC7448" and "MC7448"?
I had someone try 3 of the former for me, all @1.7GHz base clock speed (SC7448HX1700LD, as opposed to MC7448HX1700LD). Thing is, none of the chips could reach even 2.0GHz (tested at 1.35v core voltage, as was used by @dosdude1 by the end of his iMac video). The highest possible was 1.87GHz (1.3v core coltage) on a single chip.
This is a pretty sad outcome, but the biggest question is: why? Just bad luck, getting bad chips? Or is there something more to it?
I had someone try 3 of the former for me, all @1.7GHz base clock speed (SC7448HX1700LD, as opposed to MC7448HX1700LD). Thing is, none of the chips could reach even 2.0GHz (tested at 1.35v core voltage, as was used by @dosdude1 by the end of his iMac video). The highest possible was 1.87GHz (1.3v core coltage) on a single chip.
This is a pretty sad outcome, but the biggest question is: why? Just bad luck, getting bad chips? Or is there something more to it?
Probably just down to binning, but from what I understand, the SC7448 is more designed for embedded systems application, while MC7448 is for general purpose computing. I don't think there are any major actual differences between the chips.
Hey, @JoyBed, earlier in the thread, you mentioned the L3 cache chips in the MDD are clocked at 300MHz, and that you found faster, 370MHz chips that are 100% compatible instead. Do those chips have a name or product ID, so I can search for them? Do they come in sizes of both 2MB and 4MB?
My friend doing all the hard work for me somehow managed to pull this off with the SC7448 chips today:
The beast is out!
it would be really neat to try such an interposer and 7448 with an MDD, imagine a Dual 7448 2Ghz MDD that would be pretty epic
so in theory 7448's should work once soldered to an MDD CPU card with a suitable interposer, or onto a Sonnet MDX card directly
a Dual 2Ghz 7448 MDD would be quite something
ohh a Dual 7448 MDD that sounds like fun
My friend doing all the hard work for me somehow managed to pull this off with the SC7448 chips today:
The beast is out!