All very good information..thank you...hmmm ok you say you don't really know about the spi way....I would actually be a guinea pig now if I just decided to pull the card and give it a go? It may work? 😆😆😆 what's the worse that could happen? 😆😆😆😆😆😆 or! You are saying that now it is locked...external has no chance?
I don't know the scenario of "SPI fixing SMU/SME lock"
SPI will force "Flash-ROM" to be in full accessible state, and for example software for CH341 can put "expected protection" back
But current EVC could only read and remove WP
While current CH341 Neoprogrammer, (v2.2.0.10) can only do
Up to 8C but not 9C like it's expecting
Soo if Card has removedWP, soo 0x00
Turns to 0x40 once receiving power
But SME expects 0x9C - it will refuse to run full initialization and update // even if bios is fully compatible
Yet,
If the same thing happens on same firmware, same vbios and same memory training strap
It will just lock it down again to 9C, verify firmware, bootup and everything is fine ~ but you accomplished nothing
The problem,
Replacing the brain , without the body expecting a change, only works if the replacement is compatible and doesn't force an internal update
If you updated tho, and it expects a change it doesn't get ~ of course it won't boot up with then new "different" firmware :
It may work? 😆😆😆 what's the worse that could happen?
It maybe fixes itself, as you will force remove write protect ~ before even being able to flash
But it also may as well do nothing, when initialization refuses
From my side, it makes more sense to protect you from such event vs having to dual GPU flash unlock and flash it back ~ when the card hangs in an init loop
^ which people at elmorslabs had ~ hence i'm aware of the resolve, but not about the "SPI will fix everything" topic. It can not fix everything as ROM is not fully used.
Not much different on Navi10/11
Not much different on RTX 2000/3000/4000
The worst would be, having to solder wires and prevent the rom from being locked ~ then still do vbflash recovery (IF you mess up)
Option given before, was how to be aware of what happens (or has happened) while being in a functional state. To prevent non init boot-loop
Soldering is option #3,
You can do whatever you please, but a warning has been given
It was bothersome to recover and i've sit through it.
I wouldn't like to solder much on a potentially given back card, as the state of vbflash is not unlocked. You can put yourself in a situation having to RMA the card.
I know this, as if you check the #support thread in ElmorLabs discord ~ i've been assisting the 3rd person that tries to recover from an rebrand or generally stupid attempted idea.
Soo my recommendation stays,
Try to follow the instructions given, or "yolo" it and it might even resolve // we might learn something new
First before doing anything ~ confirm current state of the GPU , if locked or not
As again, using 3.31 or 4.71 vbflash "with a successful operation" ~ will indeed lock it down. That's the certainty i can give you. And a certainty how you recover from this.
The rest ~ your expected methods ~ misses data.
)
