I was looking to Hackintosh an older Dell desktop I had squirreled away and in the process of gathering all the hardware every piece of documentation I found insisted that it topped out at 4GB. This was a late-era Core 2 so that seemed completely nonsensical to me (I don't think I ever owned a Core 2 that had less than 8) and, wouldn't you know it, 8GB worked perfectly fine.
Funnily enough, like the article's author I also encountered a DSDT-related problem on that same system: if you were to dump the tables and recompile them with the standard Intel utility it straight up wouldn't work. Came to find out that there are apparently two compilers, one from Intel and another from MS and the MS one is super-lenient about accepting garbage. Eventually worked out that the logic in the stock tables was such that several features (HPET and sleep, iirc) just straight up don't work unless the OS identifies itself to ACPI as Vista (and not like Vista+, Vista exclusively). Such a pain.
There used to be people who would sell you memory for MacBooks or Fujistu laptops that nearly doubled the official RAM. My impression was that they didn’t support arbitrary N GB cards but they did support a subset, and these guys just found a module that they did support.
For my Fujitsu there was memory soldiered on the board so you couldn’t quite double it, but you could get around half again as much memory as the manufacturer claimed was the max.
I had some Dell Core 2 laptops that AFAIK were limited to 4G (well, less because it couldn't map anything above 4G so you wouldn't be able to access all the physical memory) by the Intel 945 chipset. If you had a 965 chipset you could do 8G apparently.
I ran into DSDT problems as well on my laptop running Linux. How did you figure it out? How do people reverse engineer this stuff? I managed to get all USB features working but this ACPI stuff eludes me.
Read the ACPI spec. Seriously. It's a commonly given answer especially in the Hackintosh scene, but I don't think there's a quicker or easier way than to actually understand what everything is about.
On the other hand, it's funny to think that Hackintosh probably was responsible for motivating a lot of people to learn about ACPI and other complex low-level hardware details.
I remember having to faff about with DSDT code in order to get the battery indicator to work when running OSX on VMware Player (which you had to patch)
Funnily enough, like the article's author I also encountered a DSDT-related problem on that same system: if you were to dump the tables and recompile them with the standard Intel utility it straight up wouldn't work. Came to find out that there are apparently two compilers, one from Intel and another from MS and the MS one is super-lenient about accepting garbage. Eventually worked out that the logic in the stock tables was such that several features (HPET and sleep, iirc) just straight up don't work unless the OS identifies itself to ACPI as Vista (and not like Vista+, Vista exclusively). Such a pain.