so can some explain to me what's going on with ryzen and

>>61860862
http://phoronix.com/scan.php?page=news_item&px=Ryzen-Segv-Response

File: alexjones.png (95KB, 261x238px) Image search: [Google]

95KB, 261x238px

>>61860925
that still doesn't explain if its hardware or linux though. it makes it sound like its linux, since it says linux exclusive, but also says "early ryzen processors." which makes it sound like a hardware problem.
what is it?

>>61860862
they will probably fix it by a microcode update

heavy multi-core compiling works just fine in windows,

>>61860862
Wait a bit, updates will come.

>>61861428
It can be a hardware problem, but those type of errata can usually be fixed by workaround in software. As I understand from the response they haven't actually debugged why it is happening yet, it took a year for Intel to fix their skylake hyperthreading bug

>>61862284
>just wait™

>>61861428
Segmentation faults are the BSODs of linux. If AMD will fix this it would be through a commit.

When I get ryzen 7 in a few days I'm going to get Gentoo. A lot of compiling is involved so I'll see if i get segfaults. I really want to see how fast I can compile chromium on 8c/16t

>>61860862
In real life use cases where you do'nt pin the processor 100% all the time problems like this are nonexistent. Event on my batch workloads that do this dont have any issues. That said i'd recommend a newer kernel because proper multithreading is an issue on the older ones.

>>61860862
could easily be compiler bug, which probably is, because back in the day compilers were being made in languages that worked, like C. today compilers are being made in video game languages like C++, so it's no wonder they suck so much like the whole concept of object oriented programming

>>61865137
I went from ~90 minutes on a Xeon E3-1241v3 to ~60 minutes with a 1700x.
Both CPUs at stock clock.

>>61865861
Both LLVM and GCC are bugged? I don't think so.
More likely some CPUs are faulty.
My 1700x for example compiles just fine, I emerged world a couple of times on gentoo and regularly compile android.

>>61865913
>Both LLVM and GCC are bugged?

given the fact that they both share a lot of C++ code, yes that could easily be the issue

File: m6.jpg (9KB, 225x225px) Image search: [Google]

9KB, 225x225px

>>61865986
Stop.

If you end up having segfault issues, contact AMD. People got their CPUs replaced, and it seems the newer silicon no longer had the problem.