AMD Ryzen - what's in your build?

[Melcar]

I got my week 33 Ryzen 1700 from an AMD RMA, and happy to report the kill-ryzen script runs error free with the opcode cache on. Here's the rest of my build:
Gigabyte GA-AB350M-D3H has 2xUSB 3.1 Gen2
16x2 G.Skill TridentZ 3200 CL16 running at 2666 CL16 at 1.2V
16x2 Micron 2400 CL17 running at 2666 CL18 at 1.2V

also, 2xWD green HDs, Samsung 960 evo 250GB NVMe (saving money), GTX 1050 (CUDA), mATX rosewill case (saving money), EVGA BV 500W (over specd for this build by 30+%).

The system has been rock solid for ~a week, running kill-ryzen, memtest, mprime, and various gentoo compiles, so my memory config seems to be working. Yay 2 memory controllers. If I try to go any faster than 2666, the Gigabyte DualBIOS fails over (still figuring out how it works), so I think that's the limit for 64GB RAM. I bought both pairs for ~250, and they aren't anywhere near that now.

I would recommend the mATX Gigabyte board if you want USB 3.1 Gen 2 ports, it seems like one of the few boards with this. Would've liked to get ECC with the Asrock, but spending it on 64GB of RAM feels good, and I'm sure it will help my JVM dev work.

You seem to be running mismatched pairs of RAM sticks. The slower pair is likely not able to run at the speed you are setting them to. Still, you are at a good speed. I think the officially supported speed for 4 DIMMS of dual rank RAM is 1866MHz.

[audir8]

Melcar Yep, 2800 might be too much to ask for form the Micron/Crucial pair, but 2666 CL18 seems to be doing fine.

The Gigabyte board took a step backward with the last UEFI BIOS update and doesn't let me boot into Linux with UEFI anymore without copy/pasting the .efi into the Microsoft/Boot directory. Seems like almost all motherboard makers don't seem to officially support Linux, atleast on the uATX boards I looked at. It's been a very stable board, and I don't think I will care too much as long as it supports Zen2 upgrades.

[Melcar]

Got my R7 1700 yesterday. It's week 8. Dammit. Going to put everything together this weekend and see how it goes.

[Melcar]

Well, fails after 40 minutes. An RMA will will me without a machine for who knows how long. So far everything else works fine, so whatever.

[audir8]

Melcar It didn't take too long for me, 1-2 weeks max once it's approved (just show the output of the kill-ryzen script). If you know you're going to upgrade to a Zen2/3 later on AM4, and/or can turn the opcode cache off in the BIOS i guess you can put it off altogether, otherwise, I would do it.

[Melcar]

Melcar It didn't take too long for me, 1-2 weeks max once it's approved (just show the output of the kill-ryzen script). If you know you're going to upgrade to a Zen2/3 later on AM4, and/or can turn the opcode cache off in the BIOS i guess you can put it off altogether, otherwise, I would do it.

No real plans on Ryzen 2 and was hopping to use this chip for a few years. Aside from the script, I have experienced no issues so far.

[audir8]

No real plans on Ryzen 2 and was hopping to use this chip for a few years. Aside from the script, I have experienced no issues so far.

Turning the opcode cache off makes the bug go away, so I would do that in the BIOS if you can, with a ~5% performance penalty. If you plan on using this processor for a while, I would do the RMA. I wouldn't want the issue popping up elsewhere down the road if you left it as is. I plan on upgrading to a Ryzen 2, but still using the RMA'd 1700 elsewhere, or trading it in.