It Looks Like Raptor Is Gearing Up To Release A New Open-Source POWER System

[freespirit]

Basically, petitboot shows you a boot menu with a countdown when you start up the system. You just need to start interacting with that menu via keyboard and the countdown will stop. You can navigate to a settings page right from that main menu, change the settings, save, and go.

PSU selection depends on your intended CPUs and accessories. A single 4-core machine with NVMe and just the integrated on-board VGA uses less than 150W at the wall, which should give you some idea of the size PSU required. A full dual 22 core system with GPUs and spinning rust alongside leading edge NVMe might use more like 1kW under full load.

Your RAM selection looks fine. Rank is not important, just the fact that it is DDR4 ECC registered and not LR-DIMM.

sounds good
i'm following your twitter waiting other news

[freespirit]

madscientist159
I have a doubt as the pcie m.2 storage. My intention as i wrote is to buy an m.2 nvme ssd and a pcie adapter.and use it as primary disk for the os the system it will boot the os over this interface?

[jaesharp]

madscientist159
I have a doubt as the pcie m.2 storage. My intention as i wrote is to buy an m.2 nvme ssd and a pcie adapter.and use it as primary disk for the os the system it will boot the os over this interface?

The bootloader, petitboot, is a small Linux distribution which kexecs into the OS installation of your choice. So, if your hardware is supported by Linux and shows up as a storage device, it'll almost certainly work. If your device needs drivers or some updated version of the kernel - you can always compile the changes in - Raptor has also added support for storing firmware without needing a rebuild of the flash image (depending on the device). Of course, the source and build toolchain for all the firmware is fully available.

[chithanh]

they want to avoid expensive CPU recalls due to bugs,

Eh, it's not like AMD issued a recall for the Ryzen segfault bug. Despite that a large percentage, possibly the majority of Ryzens, were affected. And AMD continued to manufacture and sell broken CPUs (here is a segfault report from week 46 Summit Ridge).

[ssokolow]

Eh, it's not like AMD issued a recall for the Ryzen segfault bug. Despite that a large percentage, possibly the majority of Ryzens, were affected. And AMD continued to manufacture and sell broken CPUs (here is a segfault report from week 46 Summit Ridge).

There's another reason to to stick to either pre-PSP Opterons or Intel chips that I forgot to mention. I value uptime on my Linux systems extremely highly and I'd completely forgotten that not everyone might care about that bug as much as I do.

[madscientist159]

Eh, it's not like AMD issued a recall for the Ryzen segfault bug. Despite that a large percentage, possibly the majority of Ryzens, were affected. And AMD continued to manufacture and sell broken CPUs (here is a segfault report from week 46 Summit Ridge).

I thought those CPUs had been recalled / replaced already? If they are actually broken, and AMD is not replacing them free of charge, at least in the EU there are basic rights to return for full refund. Probably in the US too but may require class action.

[jaesharp]

I thought those CPUs had been recalled / replaced already? If they are actually broken, and AMD is not replacing them free of charge, at least in the EU there are basic rights to return for full refund. Probably in the US too but may require class action.

Same in Australia. If an item sold is found to be unfit for the purpose for which it was sold there is a right to return which cannot be waived by either party.

[chithanh]

I thought those CPUs had been recalled / replaced already? If they are actually broken, and AMD is not replacing them free of charge, at least in the EU there are basic rights to return for full refund. Probably in the US too but may require class action.

Same in Australia. If an item sold is found to be unfit for the purpose for which it was sold there is a right to return which cannot be waived by either party.

There was no recall. AMD replaces affected CPUs on request if you contact customer support. They didn't even recall the CPUs from the channel, which means unsuspecting customers continued to buy broken ones.
You can of course go to your retailer and return the CPU, but chances are that you only get another defective Ryzen in exchange.

I think AMD got lucky that there is no easy and well-published way to reproduce the problem with a Windows program or in a web browser (only on WSL or in a Linux VM). Though in the thread on AMD community forums there was a comment about XML processing causing segfaults too, so I would not discount the possibility of a kill-ryzen.js if someone were sufficiently motivated to create it.

[tajjada]

You are aware that ME Cleaner leaves the ME kernel and part of its userspace intact and running at all times, correct? In a "cleaned" state, it's probably similar in active functionality to the PSP.

No, that is incorrect. `me_cleaner` removes the entire ME operating system, including drivers and the kernel (https://github.com/corna/me_cleaner/...oes-it-work%3F). The only thing it leaves behind is the "BUP" (BringUP) module, which is responsible for initializing the CPU and turning off the watchdog, so that your computer can boot without the watchdog resetting it after 30 minutes.

I also remember that there was an article that was published by some security researchers (and quickly taken down by request from Intel, so I can't find it to link it), where they showed the disassembled/decompiled source code for the ME BUP module (in Skylake+) and analyzed it to confirm that it does nothing shady, just simple initialization. They could do this, because Intel switched from using a customized ARC core to a regular x86 atom core for the ME starting from Skylake, so regular x86 tools can be used to reverse-engineer the ME binary blob. This is the only thing that me_cleaner leaves behind.

After applying `me_cleaner`, the Intel ME is not running any kind of operating system and does not have drivers to access your hardware, so it is effectively disabled.

[starshipeleven]

No, that is incorrect. `me_cleaner` removes the entire ME operating system,

On Skylake or newer no, as stated in that page

Unfortunately it seems that the hashes of the modules rbe, bup, kernel and syslib are checked together, increasing the number of the fundamental modules to four.

Although yeah, it's still inaccessible on its known APIs in this state.