Anyone got Rome/Milan BIOS for H11SSL-i, or how do I hack the BIOS to add support for?

I have this H11SSL-i for $149 and I kinda regret it, because I actually have the S8030 option for $229 while not only offering official PCIE4.0 support but also official support for Rome and Milan series. I do plan on upgrading the CPU for my H11SLL-i in the upcoming future (assuming the homo sapiens would have one in the near term, but considering that we are getting close to a terminally ill-fate very likely, which the ongoing war will turn full-scale on the planet I believe), but since Milan is release in 2021…trying to find decommissioned servers that uses Milan would be nothing other than a pipe dream since they are still serving the DC contracts, just probably one-fourth or at most half way through.

I know that H11SSL-i does not even claimed to support Rome entirely in the first iteration of their motherboard because it is designated for Naples but I won’t use Naples due to its terrible performance from the mouths and ears of my friends who works with infra and optimization, but I see that their BIOS does seems to indicate that they added support for both Naples and Rome since BIOS version 2.0, even for the Board Revision 1.01 (which is the one I have)…? So I would beg to question, what if we can mod the BIOS so that we entirely remove Naples and add Milan (and Milan-X) from the microcode?

More specifically speaking, we need to mod the entire AGESA suite, but if I read it correctly from my previous experience in writing UEFI firmware, AGESA comes with ComboPI, while AGESA most of the common parts are open-sourced, ComboPI is actually closed-sourced and vendor-dependent, designated by AMD to provide trusted computing root keys, Platform Initialization routines like CPU bootstrap code (the so-called “microcode” which is encrypted by the way) and UEFI initialization. This would mean modding for Milan would be a pretty hard problem to solve, unless I make up something wrong…

Manufacturer’s website for official BIOS.
If that isn’t an option, you could try finding a damn near identical board and basically copy pasting that BIOS. That would likely require pulling the BIOS from the board that does what you want, with a tool like a CH341A USB programmer or similar; and flashing it to your board with that same tool.
You might be able download a BIOS from some random EPYC board and figure out which part is what you need to replace on yours. There’s a tool called HxD to look at that. Sometimes you’ll notice lumps of data padded with 0s that end at addresses ending in FFF0 or something, then more lumps of data. Sometimes things need to end at those F addresses; that’ why there’s 0 padding.
That’s what little I’m aware of.
Official BIOS updates are sometimes/usually compressed data bundled with a program that runs without an OS (in the UEFI/BIOS) which decompresses it and organizes it as needed.

Sometimes just because you can, doesn’t mean you should

Using Milan on a Naples board is just asking for the angry pixies to let out the magic smoke

Isn’t it all hardware compatible, you just don’t always get the firmware… like shitty OEMs have been doing forever.

It’s the vrms that’s the concern

We once popped in Rome in a Naples board, the vrm couldn’t keep up and throttle heavily

Even at same or near same power consumption it was acting super weird

1 Like

Were both rated for same power? I could see power limit fucking things up on a part that would otherwise draw more. Why I didn’t even look into the workstation specific Xeon over my 2667 v2.

Can’t remember the specifics but the way he put it was consumer boards play fast and loose with the vrms where as epyc adheres to standards strictly and the vrms couldn’t keep up while doing so

Maybe misremembering so don’t quote me on it, my memory isn’t so good lately

1 Like

Probably higher power, even if just transients or similar. Or maybe even a specifc rail like for the IO die or something.

Too bad about those PSB fuses.