What gcc -march are you using

What kernel major version are you using

Any stability concerns (segfaults, reboots etc)

Mesa compilation almost always segfaults when using many jobs >4
My system sometimes freezes completely, it used to happen more often but has gotten better. I'm guessing becuase of bios updates.

mesa compile always segfaults if -j16 is used. If -j1 is used the compile completes just fine

X370 Gaming Pro Carbon

-march=bdver4 -mno-fma4 -mno-tbm -mno-xop -mno-lwp

Tried with multiple different RAM settings (2133 1.2V, 2933 1.35V, 2133 1.35V, 2x8GB in dual channel, single 1x8GB stick - memory is on the QVL for the MB and rated for 3200); tried disabling SMT; don't have bios option for OP codes. In every case I get infrequent, inconsistent segfaults in /bin/sh when compiling.

No issue whatsoever, can't reproduce the bug. CPU is 1700x

-march=bdver4 -mno-fma4 -mno-tbm -mno-xop -mno-lwp

Fatal1ty X370 Gaming K4

Fatal1ty X370 Professional Gaming

rare sigfaults when compiling like sh[28239]: segfault at 4c ip 000000000041bebe sp 00007ffd3ab758b0 error 6 in bash[400000+b1000]

UEFI 2.40, cannot find AGESA versoin

Very rare segfaults on compilation (in particular webkit-gtk). chromium compilation runs into OOM killer if it starts to swap (when it starts to swap, memory usage starts to explode until OOM killer kicks in). Killing it and resuming works just fine without bad memory consumption. Kde Lock screen tends to freeze the entire machine (not sure whether this is Ryzen related)

Segfaults with big stuff (chromium & mesa) only if i overclock my ram. it doesn't care if -j8 or -j16 - latest bios in use "PRIME B350-PLUS BIOS 0613"

Rare crashes when building system images with yocto, very rare when using emerge (Memory is underclocked 2133Mhz instead of 3200Mhz)

Segfaults during compilation are seldom, but they happen.

MSI X370 GAMING PRO CARBON (MS-7A32)

-march=znver1 -mtune=haswell

Random lockups requiring a hard reboot are not uncommon. So far, unable to determine cause or extract debugging information.

-march=bdver4 -mno-fma4 -mno-tbm -mno-xop -mno-lwp

Fixed freezes with CONFIG_RCU_NOCB_CPU_ALL=y and segfaults with norandmaps (disables ASLR).

GA-AB350-Gaming 3 (rev. 1.0)

I encounter these occasional segfaults when compiling, also sometimes freezes of may machine (only reset helps) immediately after booting. Happened with gcc 5.x and without -march=znver1, too. I’m not overclocking or making other obviously stability-affecting settings.

In one week while idling: 2 reboots, 1 kernel panic, 1 rcu_sched stall in CPUs/tasks, 1 hard lockup. first run at bios default little under 3 days before reboot. Upping memory to 2133 resulted kernel panic inside 8 hours. mem to 1.25V 1.5 days->reboots, mem to 1.3V 1.5 days->rcu_sched stall, configuring kernel with CONFIG_RCU_NOCB_CPU and CONFIG_RCU_NOCB_CPU_ALL-> hard lockup in just over 2 days.

Seldom SEGFAULTs, recompile the same package will work

CFLAGS="-O2 -march=bdver4 -mno-fma4 -mno-tbm -mno-xop -mno-lwp"

A few segfaults happened while emerging software. Emerge restart usually solves the problem.

Segfaults when MAKEOPTS="-jx" higher then 6(on Ryzen 5 1600x) sometimes hangouts/frezee need reset

Arch Linux. Segfault in /bin/sh (libtool) when building mesa.

mostly stable gentoo hardened. segfaults and random hard resets. also happened with SMT off, CPU underclocked to 2.0 and RAM at 1866 (3 SR ECC 2400 sticks). turning off ASLR seems to fix it. catpig

Freezes when idle for too long. Happens under Slackware, Win7 too

B35AG413 - AGESA update - 2017-04-13

Using Funtoo - makeopts -j17 causes problems for some packages, I had to set -j3 in package.env for some packages (mesa, gtk 2.4 etc)

Extensive testing on Gentoo (mostly kernel 4.11 and some 4.12) and Debian (kernels 4.10, 4.11) and OpenBSD -current under kvm, no stability issues with any compilation workload tried (C and C++ in various versions of gcc and clang). CPU is a Ryzen 7 1700, GPU a factory-overclocked RX460. Power supply is a SeaSonic S12II 520W, believed to be substantially underloaded given the low power consumption ratings of other components.

both segfaults, especially while compiling certain packages such as media-libs/mesa, and the occasional reboot had occurred until turning off ASLR using the randomize_va_space sysctl.

Update on idling stability. Tried Fedora 25, managed 7 day uptime easily. After that updated Gentoo, and with kernel 4.11.7 managed easily 7 day uptimes, twice. Rebooted to older kernel (4.11.4) and machine rebooted after 2 days of uptime. Strongly looks like kernel bug that got fixed. I had issues with 4.11.3 and 4.11.4.

Got segfault updating to mesa-17.1.4. Usually segfaults in bash or libc, failure rate 2-4/10 of compiling trials. Booting with norandmaps failure rate was 0/20, so it seems to solve it.

tested with -march znver1, native, and bdver4 with suggested and experimental option https://wiki.gentoo.org/wiki/Ryzen

seems Intel compiled code, also with deep optimization goes well, all compression and hashing, crypto ebuilds have compilation problems with amd compilation flags not with intel compilation flags.

Gigabyte AX-370 gaming K7

haswell, znver1, native, x86-64

Ubuntu 16.04. Test compiling GCC-7.1 in loop - 5 success from 10 loops. Compiling Linux kernel 4.12 - 9 success from 10 loops. Sometime system randomly hang without any load. RAM 1866MHz 64GB, Ryzen 1800X

Gigabyte Aorus GA-AX370-Gaming 5

-march=znver1 -O2 -pipe -mtune=broadwell -mprefer-avx128

CPU_FLAGS_X86="aes avx avx2 f16c fma3 mmx mmxext popcnt sse sse2 sse3 sse4_1 sse4_2 sse4a ssse3"

Parallel compilation causes reproducible segfaults. RAM timings and RAM and CPU voltages have no effect.

PRIME B350M-A (w/ custom BIOS)

Dedicated server. Ubuntu 16.04, w/ custom kernel (4.12.4, built w/o KASLR). KASLR causes random freezes.

ubuntu 16.04, 1700@3.75@1.3v, 4*8GB dual rank DDR4 @ 2667CL12 @ 1.35v, memtest86 ok, systester ok, mprime ok, compile wine ok (-j16), no stability issue in day to day usage. kill-ryzen has 5 fails in 2mins but doesn't seem to fail anymore after (very high concurrency required)

Segfault Issue are gone since I have added some watercooling (480mm of rads) and put all voltages to the roof. Ryzen7 1700 OC @ 3.9.

Arch Linux (all updated). ) 2666Mhz(@2400MHz) Corsair 4xDimms(Single Rank) 1T Non B-die RAM, 15-16-16-16 with tight timings at 1.3V Load Line Calibration Level 3 = Segfaults guaranteed, MCE's, also leads to kernel Panics.

But at 16-17-17-17 with 2400Mhz at 1.27V Stock Clocks, Auto LLC, All Auto Voltages 2T timings = stable across days.

Also ran sandsifter, did not lead to kernel panics or lockups in any configuration.

Gigabyte GA-AB350 Gaming 3

no issues whatsoever. running 1700x evga 16G ram at 3000

CrossHair VI Hero WI-FI AC

I will try, contrary to your recommendation, to upgrade to UEFI v1402 and see if it helps.

I merged 197 packages yesterday; including: GNOME 2.24, GCC-6.4 and glibc-2.25-r4 and all went fine with MAKEOPTS="-j 17 -l 15.0".

The only packages I am struggling with are:

* media-libs/liblo-0.29
* media-libs/mesa-17.2.0_rc4
* net-libs/webkit-gtk-2.4.11-r200

A conversation documenting this can be found at: https://bugs.gentoo.org/show_bug.cgi?id=628130

ASUS ROG CROSSHAIR VI HERO

MSI X370 Gaming Pro Carbon

http://www.asrock.com/mb/AMD/Fatal1ty%20AB350%20Gaming%20K4/

A few segfaults per hour on a ryzen 1600x with parallel compiling (yes it's gentoo)

Gigabyte Technology A320MA-M.2

Fatal1ty X370 Professional Gaming

Segfaults were fixed by CPU RMA. I'm experiencing random reboots daily, MCE Errors in logs. Every once in a while hard lock.

I'm running Ubuntu 17.04. Ryzen date code is UA1726PGT.

Disabling the micro opcache in the BIOS (called just "Opcache" there) makes the segfaults either completely gone or at least orders of magnitude less likely. With opcache enabled, segfaults occur within minutes on a parallel build.

PRIME X370-PRO, BIOS 0805 (2017/06/26)

Booted from install-amd64-minimal-20170907.iso (CSM aka BIOS mode) using chroot (following the Gentoo handbook). I get frequent segfaults.
I didn't manage to install and boot Gentoo from the internal NVMe flash storage yet. (Windows 10 booted fine when I tested a short and quick install.)

uname -a
Linux livecd 4.12.5-gentoo #1 SMP Thu Sep 7 07:57:16 UTC 2017 x86_64 AMD Ryzen 7 1800X Eight-Core Processor AuthenticAMD GNU/Linux

cat /proc/cpuinfo | grep microcode | tail -n 1
microcode : 0x8001126

gcc -v
gcc version 6.3.0 (Gentoo 6.3.0 p1.0)

NOTE: There is a newer AGESA version available from BIOS 0902 (2017/09/18), AGESA 1.0.0.6B. I will try that out as well.

I purchased the CPU as soon as it was available, so I expect it was a very early production lot...

Every day I have experienced random reboots. Since I compiled the kernel with recommendations from http://blog.programster.org/ubuntu-16-04-compile-custom-kernel-for-ryzen , I have no random system reboots.

Ryzen 7 1700X, 1733SUS (RMA exchange for segfaulting 1707SUT)

SUS 33 Week Model.. no segfaults.. had a idle lockup once.. since using rcu_nocbs=0-15 on kernel boot line it didnt happen again.. so far I'm happy

ga-abb350-gaming (rev 1.0)

I RMA'ed a 1707SUT for a 1733SUS which fixed it up. Also have a new R5 1600 retail 1737PGS on ASrock B350 pro4 build/gentoo, with no problems so far.