Many of you still seem to use Linpack to evaluate the stability.
Linpack is a pretty poor stability test, unless you're able use extremely large problems sizes.
The CPU basically sits idle around 20% of the total time spent for the computation of each loop (memory alloc & dealloc).
If you want to test the true stability of the CPU, use Prime95 28.10 for Ryzen: http://www.mersenne.org/ftp_root/gim...2810.win64.zip
with custom 128/128 in-place settings.
Newer P95 versions are less stressful to the CPU, so use the 28.10 instead.
Linpack no doubt "tests" (stresses) the memory better (more) than Prime95 does, however to determine the memory stability you should be using a dedicated utility from the get-go (Ram Test or HCI).
Personally, I wouldn't use a system which is just Linpack stable for anything productive.