AMD Ryzen 7 2700X Linux Performance Boosted By Updated BIOS/AGESA

[shmerl]

That was RavenPI, and the current one is PinnaclePI. They reset the version number to 1.0.0.0 when support for the new gen was added.

Ah, thanks for clarifying.

[Wielkie G]

Sounds like this enabled Turbo, or something like that.

Is there any change in how the kernel reports Spectre status with the new BIOS?

Would like to see gaming tests now.

[Marc Driftmeyer]

That's a pretty significant free performance boost. Great work AMD!

Those PostgreSQL numbers, PHP, Web Server and more really stand out, never mind all the multimedia jumps. In short, very beneficial workloads that are done by millions of systems on a daily basis.

[shmerl]

Only 10? for me it reports 90-110C mostly idle, in bios its shows ok under 40C

What motherboard do you have? Some noted, that they get the weird 60°C offset. I get only 10°C.

[edwaleni]

What exactly is AGESA? Sorry for my ignorance.

AMD Generic Encapsulated Software Architecture (AGESA), is a bootstrap protocol by which system devices on AMD64-architecture mainboards are initialized. The AGESA software in the BIOS of such mainboards is responsible for the initialization of the processor cores, memory, and the HyperTransport controller.

AGESA documentation was previously available only to AMD partners that had signed a non-disclosure agreement (NDA). A form of AGESA source code scrubbed of "proprietary code, identifiers and concepts" was open-sourced in early 2011 to gain track in coreboot,^[1] but these releases were stopped in 2014.^[2][3]

AGESA is linked to AMD PowerTune

[xorbe]

Michael would you mind doing some testing for the soft lockup bug on the new Ryzen chips?

https://bugzilla.kernel.org/show_bug.cgi?id=196683

That's from Aug 2017 (original Ryzen). edit: Ah, I see a 2xxx APU mentioned, but pretty sure I read about Linux teething issues elsewhere already. But I don't see 2700, 2600 mentioned.

[cbxbiker61]

It's very frustrating that the motherboard manufacturers are afraid to tell us what is supposedly fixed in their firmware listings. Some of that probably stems from the fact that they would actually have to acknowledge/own firmware bugs.

The newer Asus firmware has the Spectre V2 microcode fix in it (0x8001137). When you see IBPB listed in the vulnerability fixes, it means it has that fix.

It's possible that the Spectre V2 microcode fix eliminates some of the kernel code mitigation and therefore runs faster.

journalctl -a | grep microcode | head -n1
microcode: CPU0: patch_level=0x08001137

cat /sys/devices/system/cpu/vulnerabilities/*
Not affected (meltdown)
Mitigation: __user pointer sanitization (spectre_v1)
Mitigation: Full AMD retpoline, IBPB (spectre_v2)

[rene]

That was RavenPI, and the current one is PinnaclePI. They reset the version number to 1.0.0.0 when support for the new gen was added.

if that is true, I find it pretty unprofessional and confusing. Which sane company would "reset" a platform's firmware version number? This makes no sense to me and only loads to unnecessary confusion. And I'm saying this as AMD fan, … :-/

[Stoatally]

That's from Aug 2017 (original Ryzen). edit: Ah, I see a 2xxx APU mentioned, but pretty sure I read about Linux teething issues elsewhere already. But I don't see 2700, 2600 mentioned.

He's asking if Michael would be able to help them out by checking to see if the 2700/2600 are affected.

[Sloth]

What motherboard do you have? Some noted, that they get the weird 60°C offset. I get only 10°C.

I have a Crosshair VI with BIOS 6101. Looks like ~60C to me.