SPEC CPU, openmp, benchmarking

[Fin_]

Hi everyone, I'm not sure if I'm posting in the right thread (please let me know if I am) but tell me what I should do.

I am running a bunch of benchmarks from SPEC CPU2017. And I am getting 10-15% lower results on my system than on similar configurations. I started to run the benchmarks one by one and chose 607 (simply because it builds and runs fast enough).
The 607 benchmark shows a result +- 40% worse than the "benchmark". Thanks to "perf record -ag -F99" I was able to find out that this 40% time is spent on "_ZN11kmp_flag_64ILb0ELb1EE4waitEP8kmp_infoiPv". After a bit of googling I found out that it has something to do with openmp, C++ and multithreading. Maybe there are some special optimisation flags I don't know about?

Here are the build details:

icpx compiler: Intel(R) oneAPI DPC++\/C++ Compiler 2025.0.4 (2025.0.4.20241205)

icpx build line: -std=c++14 -std=c11 -Wl,-z,muldefs -xCORE-AVX512 -Ofast -ffast-math -flto -mfpmath=sse -funroll-loops -qopt-mem-layout-trans=4 -fiopenmp -DSPEC_OPENMP -Wno-implicit-int -L/home/user/jemalloc-5.3.0/lib

OS: Red Hat
Kernel: 5.14.0-503.34.1.el9_5.x86_64
I have a 2 CPU in system

[Viet_H_Intel]

We have published SPECspeed* 2017 https://www.intel.com/content/www/us/en/developer/tools/oneapi/dpc-compiler.html#gs.ljly5d
You can check "Configuration Details" and see if that helps your configuration.