MSI's Latest BIOS Updates Working Well For Ryzen

[phoronix]

Phoronix: MSI's Latest BIOS Updates Working Well For Ryzen

It's been a few weeks since last trying out the latest BIOS updates on my two AMD Ryzen Linux systems, but the latest releases on these MSI motherboards are indeed an improvement...

http://www.phoronix.com/scan.php?page=news_item&px=MSI-Ryzen-BIOS-Memory-Working

[wizard69]

Not unexpected. The real question is how fast can Ryzen ultimately go. We likely will have a good handle on that question 6 months from now. Micheal it woild be interesting to see the resukts of AMDs optimizing compiler along side updated benchmarks.

Ryzen is surprisingly good for such a new archetecture. It is very interesting to watch as support firms up for the chip.

[Chewi]

Interesting benchmarks. I was under the impression this wouldn't make that much difference but it is certainly measurable. I ordered my Ryzen system at the weekend but I settled on 2666MHz sticks in the end. I think that's good enough for me.

[Michael]

Not unexpected. The real question is how fast can Ryzen ultimately go. We likely will have a good handle on that question 6 months from now. Micheal it woild be interesting to see the resukts of AMDs optimizing compiler along side updated benchmarks.

Ryzen is surprisingly good for such a new archetecture. It is very interesting to watch as support firms up for the chip.

AOCC benchmarks going up ~tomorrow.

[nadro]

Michael
You mean AMP profile or A-XMP? I have G.Skill FlareX 3200MHz modules and MSI B350M Mortar and just A-XMP is available on my mobo.

[M@GOid]

Michael
You mean AMP profile or A-XMP? I have G.Skill FlareX 3200MHz modules and MSI B350M Mortar and just A-XMP is available on my mobo.

How is your experience with that motherboard? I was keeping a eye on it for a future purchase and was curious about it on a Linux system.

[nadro]

How is your experience with that motherboard? I was keeping a eye on it for a future purchase and was curious about it on a Linux system.

At first I just want to mention that since 1 month I have two platforms with these motherboards:
PC1 (used by my girlfriend). MSI B350M Mortar + Ryzen R5 1500X + 2x 8GB G.Skill Flare X 3200MHz + MSI Radeon RX 470 Gaming X 8GB + 2 SSD
PC2 (used by me). MSI B350M Mortar + Ryzen R5 1600X + 2x 8GB G.Skill Flare X 3200MHz + Sapphire Radeon R9 380 2GB (I'm waiting for RX Vega Core^^) + 2 SSD
- with BIOS v1.20 and RAM set to 3200MHz both platforms was unstable (often BIOS reset or sometimes PC couldn't even run), however with 2933MHz both platforms works fine
- with BIOS v1.30 (PC1) I had a lot of BIOS restarts even with 2933MHz (it's strange that even after a BIOS downgrade to 1.20 situation was the same...) and 3200MHz was unstable, (PC2) 2933MHz works fine, however with 3200MHz I had BIOS restarts (however not that often like in PC1)
- with BIOS v1.40 I cleared CMOS, removed and installed again memory modules on both platforms and now both PC1 and PC2 works fine with RAM set to 3200MHz via XMP profile (without BIOS restarts, boot problems etc; all is stable).
Overall it looks like BIOS v1.30 was bugged, however with BIOS 1.40 all works really good (BIOS v1.50 should be even better - new AGESA microcode), so I'm definitly happy that I choosed MSI B350M Mortar for both PC. It works without problems with Linux, however all my MSI motherboards worked fine with Linux in the past (with ASUS A88XM-E I had some minor problems with SVM) thats why I will probably choose MSI motherboards in future too. If you need M-ATX motherboard I can fully recommend this model.

BTW. After switch from A10 7850K to Ryzen games on linux works much better eg. in Shadow of Mordor I have more than 2x higher FPS than before (on PC2 I use kernel 4.12 rc1 from Ubuntu kernel ppa). I can also confirm that Windows 10 works fine too on both platforms.

[bachchain]

Is this single or dual-channel

[mlau]

How is your experience with that motherboard? I was keeping a eye on it for a future purchase and was curious about it on a Linux system.

Since bios v40 (or 1.4) it's pretty stable, even manages to go beyond the AMD-recommended 1866MHz when the full 64GB of RAM are equipped. As nadro said, previous versions were pretty unstable, I had random segfaults when e.g. compiling stuff with -j32.

The fch xhci controller is a bit finicky (the one routed to the usb3 pinheaders): sometimes it just dies when plugging in a usb3 device, and system needs to be rebooted to revive it, maybe there's a driver quirk missing for it.

There's no direct sensors support in linux, you need to add the "nct6775" module and add parameter "force_id=0xd120" to get it to recognize the sensor chip. Sensor labels are mostly wrong but it's easy to spot what's what.

[Adarion]

Well, Michael can be happy to have received an update for the FW. Even if something is fixed upstream long ago it takes ages until it makes it into BIOS/FW releases (if ever). One is very dependent on the good will of the board manufacturer to release a FW update - and this is why something independent and more user controlled like Libre*/Coreboot would be better.
(E.g. anyone remember the talk of Rudolf Marek on the CCC I guess 2 or 3 years ago? He found a bug/issue in the AGESA, AMD confirmed, updated the code but most boards never received the new AGESA part.)

But good to see that memory handling is improved now.