HP Pavilion zd8000: CPU Upgrade

[Antinomy]

It's a Phoenix BIOS with Cedar Mill support.

But with different chipset.
BTW: now I think the reason isn't BIOS but the old revision of VRD (Voltage Regulator-Down) on your motherboard.

From my experience, it's not the matter with Cedar Mills. Voltage is set correctly, it's the CPU that doesn't want to boot in "unsupported" systems.

[k24a1]

Alright, I've done some testing. Either with or without J2 isolated, the computer won't fully POST. It sounds like it's going to boot (fan spins up and then back down), but the caps/num lock lights don't blink and the display doesn't come up. I'm sure it recognizes the chip, but it doesn't have the sufficient code to fully start the computer.

[Antinomy]

Too bad there's no POST card so you can check where it stops. It's possible, J2 is not the only required pin for modification. Still, I believe that BIOS is not vital for Cedar Mill to boot.

[k24a1]

I have decided to revisit this somewhat as I have recently seen a post where someone got Core 2 era CPUs to work on i915P. I have a spare zd8k board that has an issue where it crashes only if GPU drivers are installed and I had a Pentium E5700 in my CPU pile. I did a pinmod which made it work with VRD 10.1 instead of 11, and I also bridged some pins to lower the vcore. I actually got it to warm up for a bit, but the machine will soft reset every boot when it tries to POST. This is somewhat typical behavior of a zd8000 with a CPU that is missing microcodes if I'm not mistaken, so having those inserted could potentially be a step forward. I'll also try without the vcore mod to see if it will start. There is a thread on VOGONS that goes into more detail about people getting C2Ds and other core-based processors working on 915 and even 925. While I won't get my hopes up too high, this is a gleam of hope that something could be done. While dual-cores don't usually work on 915, the C2Ds have symmetric multiprocessing which the 915 chipset can allegedly detect just fine. The Pentium D does not have this which explains why that doesn't work, along with the VRM configs being exclusively for those chips being outright incompatible with almost everything thats not from 945 to 975X.

[k24a1]

Hopefully this post shouldn't be an issue here.. I usually get nervous with double posting but I feel like this should be shared as its own thing on this thread.
I main linux, and as such there are a lot of community programs for oddly-specific things. I found one that'll swap microcodes in a Phoenix BIOS from an LGA775 platform (possibly others, the included lib.bin file is for core 2 chips only, not the most helpful for the first half of 775 chips)

katheryn@titandweevil:~/LGA775-Phoenix-bios-microcode-swapper$ ./microcode-swapper
Enter bios file path : zd8k.M24
Trying to load from zd8k.M24.
Loaded file : zd8k.M24
No.1 CPUID=F34 Rev=17 2005/04/21 CRC=2CBD6146 datasize=7120 total=7168 offset=5B805  PlatformID=( 29):0,2,3,4
No.2 CPUID=F41 Rev=17 2005/04/22 CRC=326135C1 datasize=5072 total=5120 offset=5D805  PlatformID=(189):0,2,3,4,5,7
No.3 CPUID=F43 Rev=05 2005/04/21 CRC=77812C17 datasize=  0 total=2048* offset=5F005  PlatformID=(157):0,2,3,4,7
No.4 CPUID=F49 Rev=03 2005/04/21 CRC=F85D53B8 datasize=  0 total=2048* offset=5F805  PlatformID=(189):0,2,3,4,5,7

this should give us more of an insight on what the zd8000 has for its microcodes, and CPU support. It appears as though microcodes do appear to be necessary for the zd8000 to post correctly, and there are probably some other things for it that rely on those as well (aside from the OS of course). This isn't a huge game changer or anything considering we still have the initialization sequence to deal with. The microcodes appear to be rather exclusive sadly, so replacing them may be an issue (particularly want to inject F62, F64, and F65 for cedar mills that don't use the Pentium D VRM config, and the necessary microcode for the E5700 Pentium Dual-core as a test. Seemingly this may require some extensive modifications. :/
Aaaaand this is where the hard part of getting unsupported CPUs working on unsupported platforms comes in, it's all in the name of science. I was going to give up at Prescott-2M but ever since people got C2Ds to boot on 915 chipsets there's a bit of hope left.
Here's the link to that VOGONS thread: https://www.vogons.org/viewtopic.php?t=98060
and the link to the program I tried (but couldn't get working since the uCodes aren't the same size): https://github.com/ChipCE/LGA775-Phoenix...de-swapper (should compile on just about anything using C++)