I really wish Phoronix would stop making articles about stuff they have no idea about

Signed FSP introduced with PantherLake (announced 2 days ago) does indeed introduce a new component called "FSP-O" which moves reset vector to the FSP.

It's controlled by the same mechanism as BootGuard - if PCH is fused with vendor's keys (EOM), then only FSP binaries signed with that key will be allowed to execute.

This however doesn't change anything for API mode (like what coreboot is using), only dispatch (UDK, likely AMI/Insyde).

I'm sure support for signing in API mode will eventually happen, but the point is that as long as PCH isn't fused with vendor's keys (or BootGuard is disabled), FSP signing can be skipped (and you can even build your own if you have appropriate NDAs in place to get sources).

Then, they claim "ACPI support is also added to the FSP" which is just factually false.
Intel simply bumped PPI (Platform Initialization Specification), which extends topology information with Module, Tile and Die:
- https://uefi.org/sites/default/files/resources/PI_Spec_1_7_A_final_May1.pdf (page 2-187)
- https://github.com/tianocore/edk2/blob/master/MdePkg/Include/Protocol/MpService.h#L104-L132
- https://review.coreboot.org/c/coreboot/+/89462

Like... come on, do some research beforehand.

@elly they'd have nothing left to write about!

(j/k, but maybe that's a good thing?)

@elly Ha ha, there is just no way he could be understanding any of that without a couple of talks on it.

No offense to him, he used to be a carpenter, but I see the value of phoronix as being an RSS feed with the occasional benchmarks. This is not where you get your hard facts from.

When I was more involved in Xorg communication, I was basically writing press releases for him to quote, and everyone was happy. Expecting him to know everything will just make you pissed.

@mupuf @elly I don't care much about the regular inaccuracies or the weird things he publishes (at a certain point he had an article for every boring pull request and pach series I sent out to LKML).

The much bigger problem is the comment section, it's essentially unmoderated and full of vile people. I've blocked his website on all my machines now.

@sven @elly Yeah, the comment section not being moderated is terrible...

@sven @mupuf @elly I used to have a subscription to Phoronix, but I cancelled it after the extreme harassment I saw, particularly a lot of transphobia around the Asahi project.

I emailed Micheal to explain my motivations, and he shared his frustration and told me that there would be a cleanup, and to tell him about any violations. To his credit, he did clean up some.

I offered to volunteer to moderate, but never got a reply to that. The fact still remains, that it is a mess over there.

@jonkoops @sven @mupuf @elly the best to deal with that is to put the entire comment sections into a moderation mode where _every_ comment needs approved. Could flag certain accounts as safe if you want.

But that means often finding people willing to do the work or even paying people...

@jonkoops @sven @mupuf @elly Yeah it doesn't seem like Michael is very much for all these vile comments, but it seems he doesn't actually notice most of the time or something?

@mupuf
speaking of benchmarks, is Phoronix good at them, in terms of controlling variables and such?

@wolf480pl No, it isn't good. His automation doesn't check that the settings he wants are applied.

That being said, this is a common problem in the industry.

@mupuf
so nobody makes good benchmarks?

@wolf480pl Not nobody, it just takes a lot of thinking and analysing results to increase the SNR and phoronix just doesn't have the bandwidth for that.

@fun @jonkoops @mupuf @elly that doesn’t matter imho, he has a large audience which he directly benefits from (through ad revenue) and that comes with responsibility. “He doesn’t know about it” is not an acceptable excuse.

@sven @jonkoops @mupuf @elly fully agree