First up was the LAMMPS molecular dynamics workload. The Ryzen 7 1800X here was running between the speed of a Core i5 6600K and Core i5 7600K, both of which processors retail for less than the $500 USD of the 1800X. But this performance is much better than we saw out of the older AMD FX CPUs.
When seeing this second result, FFTW, instilled more confidence in Ryzen's potential... At least here this single-threaded test was now running just behind the Core i7 7700K, albeit that is Intel's $340 Kabylake CPU.
But the third test run was John The Ripper and here we finally have some multi-threaded action. With John The Ripper, which makes use of OpenMP, the Ryzen 7 1800X was the fastest CPU tested and came out ahead of all the other CPUs tested, including the Core i7 6800K and Core i7 5960X. Unfortunately I have no Intel Core i7 6950X for comparison.
The TTSIOD renderer results were also great news for Ryzen on Linux. Here the Ryzen 7 1800X was 14% faster than the i7-6800K but just behind the i7-5960X Haswell-E.
When running GraphicsMagick with its OpenMP image processing workload, it was the fastest CPU of the hardware tested.
Or with another GraphicsMagick test that's less thread happy, the Ryzen 7 1800X was coming in just behind the Intel Core i7 7700K.
All of these benchmarks were built from source for each of the CPUs tested, as is usual for the Phoronix Test Suite. Though with Himeno there is a reproducible oddity: the Ryzen 7 performance for this pressure solver workload is very poor. The Ryzen 7 1800X came in behind all of the tested Intel CPUs, strangely. Perhaps a GCC compiler issue? Anyhow, I'll have some Clang vs. GCC tests on the Ryzen 7 soon enough.