AMD Ryzen 3000 Series Playing Nicely With Latest Linux Distros Following BIOS Updates

[finalzone]

Because the Fedora 30 installer uses an unpatched systemd? And I never wanted to bother spinning my own USB installer with a fixed systemd?

So when I setup my Ryzen 3900 I used Ubuntu 18.04 because I knew it would work.

Now I can update the BIOS and use the F30 installer. If I want to.

Have you used the respin version: https://dl.fedoraproject.org/pub/alt/live-respins/ ?

[finalzone]

Wait, was the Ryzen 3400G affected by this? I thought it was a unique issue with Zen 2?

Nope. Running updated Fedora 30 on a friend's HP Pavilion Ryzen 3500U without issue.

[hax0r]

Only to find out you rarely (if ever) need to read 5GB/sec. You'd think it helps with games' level loading, but benchmarking has shown otherwise.
Though honestly, I don't know if I gave up the feature just like that if I were in your shoes. On the other hand, knowing I have a broken instruction... not good either.

I mostly agree, PCIe 4.0 NVMe's are overkill for my needs at the moment and they're pricy, I settled on ~$90 Phison E12 based 1TB NVMe drive, good I/O drive with decent random 4KB read/write performance helps immensely when compiling large projects such as Android or Firefox. As of writing this post the AOSP source code weights around 55GB and once compiled it generates 106G of data in "./out" folder. Now that's a lot of I/O.

[theriddick]

I've had to use clearcpuid=514 for my 3600 to not have random freezes in games, however gigabyte wifi itx mobos got the 1003ABB patch for bios a couple weeks back. which I run.

[bug77]

I mostly agree, PCIe 4.0 NVMe's are overkill for my needs at the moment and they're pricy, I settled on ~$90 Phison E12 based 1TB NVMe drive, good I/O drive with decent random 4KB read/write performance helps immensely when compiling large projects such as Android or Firefox. As of writing this post the AOSP source code weights around 55GB and once compiled it generates 106G of data in "./out" folder. Now that's a lot of I/O.

Yeah, I wish at reviewers would stop touting sequential numbers, since random 4k is where it's at. And PCIe (4.0 or any other version) does nothing for that.
It's why I hoped XPoint would take off quickly: that thing flies in random access.

[Guest]

This makes it sound like these distro wouldn't without some sort of extra work. That's not the case. All this does it allow the Ubuntu Live CD and maybe some outdated distros to boot. The live CD uses an older version of systemd that isn't patched. If (like me) you already have Ubuntu installed and then upgrade your CPU to Zen 2 there won't be a problem because the updated ("patched") version of systemd was rolled out a while back and is already installed. I recently booted my system straight into Ubuntu without any worries for this reason. (My BIOS is on 1.0.0.2AB (not ABB) and that was only rolled out yesterday)

The Ubuntu team has said that they will not be producing a new version of the Live CD. So ONLY if you are using the Live CD will there be a problem. If you use the (imo much better) mini installer (which doesn't use systemd) there will be no issues (the latest systemd will be downloaded and installed automatically). If you already have a system installed then there is no problem because your systemd will already be fixed.

Well only a few distros have the patched systemd version - it's still a glaring CPU bug that shouldn't have been present in systems shipped to consumers in the first place. For most people, it is unusable.

[Guest]

Wait, was the Ryzen 3400G affected by this? I thought it was a unique issue with Zen 2?

I believe first-gen Zen CPUs weren't affected, although some older AMD CPUs were affected (including my brother's laptop - sure put a dent in my attempt to get him to use Arch Linux).

[Madgemade]

Well only a few distros have the patched systemd version - it's still a glaring CPU bug that shouldn't have been present in systems shipped to consumers in the first place. For most people, it is unusable.

It certainly is. A BIOS update is just no enough to fix this. I can just see the posts now from users with an OEM system that uses an outdated BIOS that cannot be updated at all. These systems will be crippled forever. Not to mention the people who buy a Zen 2 computer but have no knowledge or interest in updating the BIOS.

[Zan Lynx]

Yeah, I wish at reviewers would stop touting sequential numbers, since random 4k is where it's at. And PCIe (4.0 or any other version) does nothing for that.
It's why I hoped XPoint would take off quickly: that thing flies in random access.

X-Point / Optane is indeed great stuff. That's why I picked a 905P for the main drive for my Ryzen 3900Xbuild. I use a 900P on my older 5960X. I got a "free" StarCitizen spaceship with that one.

There just isn't a better storage device for a desktop system.

[DRanged]

I'm not updating to newest AGESA due to AMD removing PCIe 4.0 support in lower-end boards, I have https://www.asus.com/us/Motherboards...M-PLUS-GAMING/ as one of my burner systems and I'm sticking to BIOS with 1.0.0.3AB, this B450 motherboard from ASUS is awesome because it allows you to run 16x Gen 4.0 PCIe in top slot and also Gen 4.0 on M.2 slot, it is a best deal ever, you can get the board for $50 at microcenter (after discount) when you purchase any AMD Ryzen 3000. No need to purchase toasty and expensive X570 chipset, for $50 you can have PCIe 4.0 and utilize latest Phison E16 based NVMe drive to get 5GB/sec read speeds.

I've read the specs of that mobo but it only supports PCIe 2.0 and 3.0, so where you get the idea it supports PCIe 4.0 is beyond me.

The X570 boards have PCIe 4.