Which RAM do you have now? I have two Crucuial 16GB 2400 sticks. I picked them up on Amazon when they were on discount and to be honest, I’ve gotten used to 32GB RAM 
QVL 32GB RAM seems to be pretty expensive though…
I am running 32GB (4x8) G.Skill Ripjaws V 3200 16, 16, 16, 16, 36 @2933MHz.
Got that kit on a deal for 240,- Euro incl. tax and shipping. 
[EDIT] Ah, I knew I had that somewhere. This is the exact name of the kit:
F4-3200C16Q-32GVK
1 Like
I did VMware Workstation + latest virtual hardware version + Ubuntu 17.04. I can’t confirm VirtualBox will work, but it might. Worth a shot at least if Linux doesn’t work natively.
Edit:
Just shipped my CPU to AMD as an RMA using the provided FedEx label. Thursday Night…
Florida has mandatory evacuations going on due to Hurricane Irma (which is wider than the entire state).
Somehow… I don’t think Miami is a good place to ship things to right now…
FYI: USA RMA’s are being sent To Miami but replacements are being shipped From other places, Ontario, Netherlands.
Today I was able to boot into Fedora and there seems to be a default Fedora support, so we don’t need these additional changes above anymore.
That being said, my USB thumb drive was read-only, apparently Fedora’s media creator doesn’t create a writeable drive, which is why I wasn’t able to run this script. When I re-created the a Live USB stick using Rufus I had the mce boot error again.
Hey,
I was able to confirm the issue using Ubuntu 16.04.3 LTS in VirtualBox on Windows 10 Pro x64 as a host. Exactly the same behaviour as described with a “natively” booted Linux.
You can get around the ramdisk limitations by using “2 8” as the parameters for kill-ryzen.
I’m in contact with AMD for a replacement as we speak.
1 Like
I’m assuming that what you are saying is that you tested with a Ubuntu 16.04 Guest OS in a Virtual Machine on a Windows Host?
Exactly
Got another 1600X and will test it now. It is a 1723 though…
Crap, I have that error too. I don’t really want top wait long for RMA, I hope the shop does it itself, not directly AMD.
Then you can’t be sure to get a post-week-25-Ryzen. Their RMA process seems very well done so far. They’ll cover the full DHL expense upfront and send you the replacement once you give them a tracking number.
I’m very pleased with their RMA department currently.
I am running the blender build right now and it just passed 50%. I might be overly optimistic but … well, I’ll hammer it a while more before I finish that sentence.
[update] Blender passed. Now running mprime to heat the chip a bit and start a new blender build.
Just realized I did not take a picture before installing the chip…
I know it is a week 23 1600X from Malaysia, microcode is 0x8001126.
[update] Second blender build successful. Looking good.
Third build done. This seems to be a winner. Yeehaw!
taichi. should be here by the end of the week (I wasn’t expecting the CPU to come so soon).
1 Like
OK, I ran a lot of blender builds, with and without mprime -t running right next to it. After that I built mesa-git with all dependencies. Took half a day, completed without any problems. Is there anything worse to throw at it or is it a reasonable assumption that my chip is fine?
chip is probably fine.
1 Like
1 Like
Don’t be too sure. Blender always worked for me and I have a week 21 Malaysia one. I had to run kill-ryzen for it to trigger.
1 Like
Good to know.
Couldn’t get that ryzen kill thing to run on manjaro. So I’ll try this, that seems to be written for stupid people like me.
I think (U know nothing Jon Snow) I am ok as well. Long blender renders, handbrake encodes and the Kill-ryzen code for about 4 hours only then mined etherium for a day with the CPU and no problems.
I would have an aneurysm after my MB issues if the CPU was bad as well. I hope it get sorted for people affected quick.
OK, i’ve got my replacement CPU. I’ve gone from a 1721PGT, Malaysia, to a 1730SUS, China. No bugs yet and crazy binning.
4.1 GHz @ 1.3375V prime95 and AVX stable.
Weird thing is, I’m experiencing opposite issues with offset/voltage/pstate compared to my old Ryzen. With the old one, manual voltage override would result in a stuck multiplier and offset + pstate worked. With the new one only manual voltage override works and offset/pstate doesn’t.
As for the whole RMA process: Their response time per e-mail is about one working day. After providing them with all the info, referencing to this thread here, I was given a DHL Express pre-paid account number and they picked the CPU up the day after.
6 days later the replacement part arrived. Pretty good.
1 Like