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.
Well... I screwed up my PBG4 DLSD
- Thread starter JoyBed
- Start date
- Sort by reaction score
I ran a few tests again to confirm and I had mislabeled my REV/STD slots - so flip the slot labels on my earlier post.
I can confirm that my 15-inch DLSD will boot and recognize full memory with a single 2GB PC2-5300 in either slot. No matter what I try, having 2x 2GB will not post (tried different brands of the same speed).
I then put 1GB in the lower slot (STD) and the 2GB in the upper (REV) and the result was 2GB of available RAM, with nothing registered in the other slot.
I followed the OF technique and was able to successfully have both sticks recognized as 1GB each (40000000 and 400000000) and booted OS X OK with 2GB total RAM from the 3GB combo.
In a final test I mapped 40000000 and 80000000 for 1GB +2GB and then setenv boot-args "-v" before mac-boot to see the verbose output...
I get a kernel panic at:
Which correlates with the xnu 1228.15.4 source code at line 96;
The xnu source is available to download from opensource.apple.com. Also install darwinbuild which sets up the tools to simplify the build (port install darwinbuild) - perhaps the rtclock.c code can be patched? I don't have any understanding of how this works though, but it might just be an unsigned int range issue?
An interesting tidbit - version 1504.15.3 of xnu (Darwin bootable image, mach_kernel, IOKit, etc) which shipped with 10.6.8 includes source code and build instructions for ppc, i386 and arm as does 10.5.8's 1228 version - which could mean that a bootable Darwin 10.x (Snow Leopard) for PowerPC might be possible - but for reasons beyond my understanding, the knowledgeable people behind darwinbuild restricted ppc builds using their automated system to version 9.x (Leopard), so I figure they must have already tried and failed.
Another fun find is the Panther xnu (v517.x) can build for i386.. Did osx86 ever have a Panther build or did it start at Tiger? I should pickup a P4 box to play with just for the fun of it
I can confirm that my 15-inch DLSD will boot and recognize full memory with a single 2GB PC2-5300 in either slot. No matter what I try, having 2x 2GB will not post (tried different brands of the same speed).
I then put 1GB in the lower slot (STD) and the 2GB in the upper (REV) and the result was 2GB of available RAM, with nothing registered in the other slot.
I followed the OF technique and was able to successfully have both sticks recognized as 1GB each (40000000 and 400000000) and booted OS X OK with 2GB total RAM from the 3GB combo.
In a final test I mapped 40000000 and 80000000 for 1GB +2GB and then setenv boot-args "-v" before mac-boot to see the verbose output...
I get a kernel panic at:
Code:
Warning: AppleMacIO self test fails
panic(cpu 0 caller 0x000BBA3C): "rtclock" timebase callback: invalid constant 0 / 1"@/SourceCache/xnu/xnu-1228.15.4/osfmk/ppc/rtclock.c:96
...
Backtrace:
...
com.apple.driver.AppleKeyLargo(1.7.2f1)@...
...(snip)Which correlates with the xnu 1228.15.4 source code at line 96;
Code:
static void
timebase_callback(
struct timebase_freq_t *freq)
{
uint32_t numer, denom;
spl_t s;
if ( freq->timebase_den < 1 || freq->timebase_den > 8 ||
freq->timebase_num < freq->timebase_den )
-> panic("rtclock timebase_callback: invalid constant %lu / %lu",
freq->timebase_num, freq->timebase_den);
denom = freq->timebase_num;
numer = freq->timebase_den * NSEC_PER_SEC;
...(snip)The xnu source is available to download from opensource.apple.com. Also install darwinbuild which sets up the tools to simplify the build (port install darwinbuild) - perhaps the rtclock.c code can be patched? I don't have any understanding of how this works though, but it might just be an unsigned int range issue?
An interesting tidbit - version 1504.15.3 of xnu (Darwin bootable image, mach_kernel, IOKit, etc) which shipped with 10.6.8 includes source code and build instructions for ppc, i386 and arm as does 10.5.8's 1228 version - which could mean that a bootable Darwin 10.x (Snow Leopard) for PowerPC might be possible - but for reasons beyond my understanding, the knowledgeable people behind darwinbuild restricted ppc builds using their automated system to version 9.x (Leopard), so I figure they must have already tried and failed.
Another fun find is the Panther xnu (v517.x) can build for i386.. Did osx86 ever have a Panther build or did it start at Tiger? I should pickup a P4 box to play with just for the fun of it
Last edited:
The first public OSx86 was 10.4.1 from the DTK. But... (a) Steve said there's been x86 builds since Cheetah and (b) there's a Darwin/x86 7.x.x binary and source release corresponding to Panther. There's also OpenDarwin 7.2.1 for x86.
Sorry if that was a rhetorical question.
[automerge]1587482893[/automerge]
Meh - I'd rather get a Pentium M box. Way cooler.
Last edited:
Did osx86 ever have a Panther build or did it start at Tiger? I should pickup a P4 box to play with just for the fun of it
The final release bundled with the late 2005 PPC Macs was 10.4.2 although a later refresh with 10.4.4 can be found. So 10.4.3 onwards will boot any G4/5. Intel Macs therefore started just after, so no Panther x86.
Meh - I'd rather get a Pentium M box. Way cooler.
A Dell D600 laptop with a Banias processor was my first Hackintosh running Tiger. It did run hot and we could never get full acceleration for the Radeon 9000 mobility chipset.
Another fun find is the Panther xnu (v517.x) can build for i386.. Did osx86 ever have a Panther build or did it start at Tiger? I should pickup a P4 box to play with just for the fun of it
apple does/did provide bootable Darwin installer ISOs for both PPC and x86 from the OS X equivalent of 10.0-10.4
the page is no longer on apples severs (at least not at that address anymore)
but the direct links do still work
the fun one is the early x86 darwin builds stay entirely in more or less low rez text mode rather then switching to a higher rez "graphical" mode
Meh - I'd rather get a Pentium M box. Way cooler.
Anything Intel could spit out can't get much cooler than:
Perhaps something could be appended to Open Firmware so that it will at least POST? Maybe a modified RAM ceiling, or an edited maximum for each slot at a single time?
At this point, I feel like we would get higher yields from a (locally installed) Linux distro than anything OS X was artificially hardcoded with.
@Amethyst1 At a basic level, it's an enhanced Tualatin with SSE2 added.
The difference is, I've never heard of a Pentium M workstation. Let alone a dual CPU-capable (Pentium M) mainboard.
Pardon my glaring ignorance, but is there even any Pentium M system that supports a dedicated graphics card? It's a cool chip for sure, but until it further evolved into the Core 2 Duo (more often than not accompanied by i965+), it felt most at home in a notebook.
The difference is, I've never heard of a Pentium M workstation. Let alone a dual CPU-capable (Pentium M) mainboard.
Pardon my glaring ignorance, but is there even any Pentium M system that supports a dedicated graphics card? It's a cool chip for sure, but until it further evolved into the Core 2 Duo (more often than not accompanied by i965+), it felt most at home in a notebook.
There were a couple of (quite expensive, admittedly) desktop mainboards for the Pentium M which were equipped with a full-size PCI-Express x16 slot, yes. I've had one of those a couple of years ago, with a 2 GHz "Dothan". There was also an adapter made by Asus that allowed using a P-M on some of their P4 mainboards, enabling serious overclocking etc.
And if you fancied a dual-CPU workstation, the "Sossaman" Xeons were based on Core Duos, i.e. basically dual-core Pentium Ms.
I'll concede that Pentium M outside of a laptop was anything but mainstream. But who cares about mainstream?
Dronecatcher, I guess I just want to browse the internet like I did in 2000 when I originally got into Mac, via a G4 sawtooth 500. Never owned a PowerBook Pismo though. I tried classila and with websites like Macintosh garden for instance I kept getting security warnings - how do I disable those ? Can you also remind me what web browsing even looked like in 2000 ? I remember I had some form of DSL then.
snap!
Intel Pentium M 730 @ 1596.19 MHz - CPU-Z VALIDATOR
[bthsh9] Validated Dump by LightBulbFun (2017-03-30 19:36:00) - MB: AOpen i915GMm-HFS - RAM: 2048 MB
valid.x86.fr
(well same mobo but diffrent speed of CPU, sadly the BIOS got corrupted and the BIOS chip aint socketed, its in recovery mode asking for a floppy disk with the BIOS on it, I need to get a USB floppy and hope to hell that works because it uses a proprietary FDD header that i never got the cable for sadly)
I did love to pinmod Pentium Ms , so easy to overclock them!
Back to the topic, I have installed Lubuntu 12.04; yet no luck.. maybe the configuration is not correct.
If the addresses are represented in Hex
then:
20000000 = 512mb
40000000 = 1024mb
80000000 = 2048mb
If I insert 2 1gb modules what I get out is:
00000000 40000000 (first stick)
40000000 40000000 (second stick)
following the same logic to have 3gb then should be
00000000 40000000
40000000 80000000
but it doesn't work, always stuck on "do-quiesce finishedturning from prom_init" not even when addressing 2.5gb (not even 2.25).
sometimes when it goes past it, the screen is just scrambled
There was little writing about people trying to get past 2gb with the MDD Powermac G4, yet again no success.
Probably the share the same memory controller?
My knowledge stops here unfortunately
Back to the topic, I have installed Lubuntu 12.04; yet no luck.. maybe the configuration is not correct.
If the addresses are represented in Hex
then:
20000000 = 512mb
40000000 = 1024mb
80000000 = 2048mb
If I insert 2 1gb modules what I get out is:
00000000 40000000 (first stick)
40000000 40000000 (second stick)
following the same logic to have 3gb then should be
00000000 40000000
40000000 80000000
but it doesn't work, always stuck on "do-quiesce finishedturning from prom_init" not even when addressing 2.5gb (not even 2.25).
sometimes when it goes past it, the screen is just scrambled
There was little writing about people trying to get past 2gb with the MDD Powermac G4, yet again no success.
Probably the share the same memory controller?
My knowledge stops here unfortunately
I'm not too sure about that. The AHT on the Intel Macs looks pretty much the same as the ones on the PPC Macs, but if you go through the files they are all .efi files. I didn't decompile them and check anything, so I'm not sure if it's just a really good copy of the PPC AHT interface or actually a version of OS 9.
On PPC Macs however, I think it is actually some form of OS 9. I checked the files on two of my PPC Macs and booted them to compare. On my TiBook 867 (the last model to officially support OS 9) the AHT is shown as Mac OS 9, and boots up as such. The disc has all the files required to start Mac OS 9 and it is listed with a Mac OS 9 icon in the OF.
On my DP G5 choosing the disc in OF shows an AHT icon, but booting up it's almost identical to the one on my TiBook. Looking through the files I found out it seems to be a version of OS 9 (complete with a System Folder, with System and Finder inside it) and some other things. I can't remember much about it because it's been a few years since I looked through them, I might have to dig them up again to check, but I also recall there was a custom AHT icon set inside the disc that was used for the OF boot volume selector.
I tried all combinations of faking more than 2 gigs, but the only one combination that booted is 2 gigs in one slot and 8 bytes in other slot.
I don't want to derail this thread but this is interesting. My understanding is that prior to Intel, the ASD (and by inference) AHT discs either used a cut down MacOS or OF (presumably Forth scripts) as the means to host the test routines. I never paid attention to the G5 discs but if they are using MacOS, then there is the basis for booting at least a rudimentary MacOS something on G5 hardware.
Perhaps a long shot, but maybe parts of OS 9 could even be swapped in favor of parts of ASD to form a bootable Mac OS 9 on G5 hardware?
I wonder if such an idea has been explored before?
The earlier posts in the thread were indicating it was a 2GB limit in OS X. Which it isn't. If it's specific to a subtype of CPU or firmware that's a hardware limit, not the operating system.
What do we take away from this? Be clear with your terminology.
Is it a limitation of OS X? Some things indicate that its not but some that yes it is. From hardware side(the PCB) it has all the data and address wires to the ram slots to take 4gigs. But it seems like it is in the OpenFirmware because the main thing observed is if you plug 3gigs in the PBG4 the OpenFirmware sees both modules and their capacity but uses just the one in the main ram slot. The main ram slot is the one that is hardwired as Bank1 and Bank2. IIRC than its the reversed one. Maybe patching OpenFirmware fixes this issue.
Well my DLSD lower / STANDARD slot now refuses to see any RAM modules. Tried resetting nvram and PRAM, no change.
There goes another Powerbook G4 lower slot failure due to the test of time.
It’s a good thing this model happily takes a single 2GB module, so it’s really no better or worse off...
There goes another Powerbook G4 lower slot failure due to the test of time.
It’s a good thing this model happily takes a single 2GB module, so it’s really no better or worse off...
Do not worry after trying the 2 gig module in the standard slot it happens to me as well. Every time I try it. But I fix it by turning on the PBG4 with 1 gig stick in the reverse slot. After shuting down and not disconnecting the battery and pluging second 1gig stick to the standard slot it works normally again.
Thank you, I noticed that everything was http and not https. Internet browsing then was simple and clean. Well, do you think Outlook express would allow me to at least get my email ?
That and the email client in Classilla - though I do think you have to allow unsecure apps access - certainly do with Gmail.