As I understand it, the only thing missing with regard to nvidia dGPUs on
Coreboot are the ACPI calls to turn them on and off at run-time, which is
what Optimus is. Things like DRI_PRIME and similar work just fine as long
as the dGPU is running, even under nouveau.
What is the current status of the work to get nvidia Optimus working? From
what I gather there have been a couple of attempts and a recent restart,
but I don't know what the current status of this work is.
I have heard that some laptops currently have the option to switch the dGPU
on or off at boot-time. For which ones does this work?
If the work on Optimus is completed, what laptops are most likely to
support it? My expectation is that turning the GPU on or off may be
different for different laptops, even though the ACPI calls are the same,
and so may not work everywhere.
A week ago I wrote here about my problems trying to port coreboot to
my board. Unfortunately I am still no closer to booting.
In the meantime I flashed my new chip with my OEM firmware backup. It
boots; then I flashed my patched IFD (for chip ID and flash unlock)
and it still boots. So it's not chip compatibility or corrupted
The only sign of life I got is the bootblock banner left in the SPI
console. My PCI POST card is showing nothing, but knowing that it sits
on a PCIe-PCI bridge (ASM1063 that P8Z77M-PRO does not have) and not
knowing if it needs software init to work, I am now trying to pull
POST codes off the LPC bus over the TPM header, using an Arduino Due.
Do I have to add some early init to have port 80 accesses sent to LPC
bus for this to work?
Thanks for your help
albeit the board has been deprecated in coreboot 4.11, I wanted to contribute a working RAM setup for all who still use it. I managed to run a KGPE-D16 with a single CPU (Opteron 6380) and 96 GB of RAM and the following LRDIMMs:
orange slots: 4x Samsung 16 GB DDR3L reg ECC (M393B2G70BH0-YK0)
black slots: 4x Samsung 8 GB DDR3L reg ECC (M393B1K70DH0-YK0)
Using full 128 GB with the 16 GB modules resulted in ECC errors during memtester runs (and under heavy load). Also, I was not able to get a configuration with 96 GB using only 16 GB modules running - neither by the configuration described as problematic in the Wiki (leaving the slots next to the CPU empty) nor using the configuration desribed in the user manual (leaving the slots B1 and D1 empty.
I tested the "working so far" configuration with memtester (running aprox 10 minutes) and by compiling AOSP for about one hour with 80 GB allocated to the VM where the compilation is running. No entries about ECC-recovered hardware errors and such in syslog so far (unline the previous run when I had 128 GB in the machine).
Btw.: For me, it would make sense to have a section in the documentation about boards that were previously supported in coreboot such as the KGPE-D16 that is currently missing there.