Linux 4.15 Is A Huge Update For Both AMD CPU & Radeon GPU Owners

[pjssilva]

xpander , to be sure that your CPU is not affected by the segfault bug you should try the kill-ryzen script for at least 24hous. I have seem my old processor segfault after 4 hours, for example. I have a new CPU (I did the RAM) now that is able to sustain many days of kill-ryzen without problems. However I am facing the random freezes... I have just recompiled the kernel to add the RCU_NOCB_CPU=y and started passing the correct option to the kernel at boot time. Now, I'll see if this solves the freezes during the week.

uhh.. reading all that here. I have been extremely lucky then. i bought my ryzen 1700X at launch in march. i havent got the segfault bug at least ryzen-kill script didnt throw up anything with 2 times running it, both times about 2 hours, maybe i need some more tests. that freeze/crash bug, i have had it like 3 times only and its always been when im away from computer when it sits in idle, so i havent lost anything usefull. with kernels prior to 4.13 i didnt have issues at all. RAM speeds were a issue way way back in march and april, but now i have been running RAMs quite high speeds, not really on the full speed my ram can do but 4 sticks is harder to get into the correct frequencies. Anyway i have been super happy with my Ryzen for the most part. Compiling something with it is just a breeze, it will chew through everything so fast.

But i agree that AMD has been slow with the supporting and fixing bugs. Those things should have been done before Launching the CPU's, like the temp and volt monitoring and stuff like that. And nasty bugs should have been fixed within few months of the release. but its been taking too long. Sad to say but i say the same about AMD GPU support. New GPUs that launch will not be supported very well at first and it takes way too much time for them to fix things. Nvidia and Intel seem to be faster in those areas sadly.


Still i love my Ryzen CPU. For the price i got it, its amazing in multithreaded workloads as well as performs good enough in singlethreaded workloads.

[xpander]

xpander , to be sure that your CPU is not affected by the segfault bug you should try the kill-ryzen script for at least 24hous. I have seem my old processor segfault after 4 hours, for example. I have a new CPU (I did the RAM) now that is able to sustain many days of kill-ryzen without problems.

Running 24/7 with kill ryzen is not an option for me, i want to use my PC also I guess it doesn't affect me with my workloads anyway, i don't do much parallel compiles anyway.

[johanb]

Oh, so that's the reason why my R5 1600 has had two random freezes the past week. Have been running arch and didn't do a upgrade for a while until early last week, since I've had two random freezes where I had to restart the computer. I really hope that they solve this issue as fast as possible, it's quite annoying!

[pal666]

Please add weight, it should be available for distribution to enable by default.

set regression: yes in bugreport

[theriddick]

It should be pointed out that my gigabyte board with anything above AGESA 1.0.0.1 would cause memory configuration errors, XMP values NOT being set regardless of selection. I run 1.0.0.5 or so now and I had to put in quite a few values manually based off the XMP table, BUT with a slight tweak, one of the settings needed to be set to 15 and not 12 or crashing would occur (or instability). I documented this in the Gigabyte forum.

Anyway, wouldn't be surprised if MANY people get bit by this type of bug or similar and incorrectly blame the CPU.