The Impact Of GCC Zen Compiler Tuning On AMD Ryzen Performance

[phoronix]

Phoronix: The Impact Of GCC Zen Compiler Tuning On AMD Ryzen Performance

The latest in our AMD Ryzen Linux benchmarking is looking at the impact of compiled binaries when making use of Zen "znver1" compiler optimizations with the GNU Compiler Collection (GCC) compared to other optimization levels like Bulldozer and K8-SSE3.

http://www.phoronix.com/vr.php?view=24234

[qsmcomp]

Try -march=haswell -mtune=haswell -mno-rdrnd and -march=haswell -mtune=znver1 -mno-rdrnd -mprefer-avx128 -mvzeroupper.
You will see more interesting results.

[k1e0x]

you might get better results with -O2, -O3 can be unpredictable.

[qsmcomp]

you might get better results with -O2, -O3 can be unpredictable.

with
O3 -fno-inline-functions -funroll-loops -fpeel-loops -ftracer
the results might be more 'tricky'.

[k1e0x]

you might get better results with -O2, -O3 can be unpredictable.

yea yea, he knows that for sure, I'd just rather see something I'd actually use like -O2 -mtune=znver1

[qsmcomp]

yea yea, he knows that for sure, I'd just rather see something I'd actually use like -O2 -mtune=znver1

I would like to see the difference between codes optimized for Intel CPUs and codes optimized for AMD with similar instruction sets.

[TumultuousUnicorn]

Why not -march=native?

[qsmcomp]

Why not -march=native?

-march=native is used in previous benchmarks.

[edwaleni]

Himeno benchmark score is affected by the performance of a computer, especially memory band width.

I was curious why Ryzen stinks at Himeno and it appears to be more related to the size of the cache and bandwidth with main memory.

[indepe]

Himeno benchmark score is affected by the performance of a computer, especially memory band width.

I was curious why Ryzen stinks at Himeno and it appears to be more related to the size of the cache and bandwidth with main memory.

The Himeno benchmark seems to be designed to always do the opposite of what you'd expect. (just joking)