Sergej Breiter

Hi, and thanks for taking care. Yes it does fix the issue. At least in the simple case that I have tested: ``` $ for i in $(seq 0 95);...

Here is the [benchmark code](https://github.com/user-attachments/files/20186851/triad.zip) (Makfile included) @lplewa . It compares three memory allocation versions: - Plain `malloc` + `numactl` - OpenMP High Bandwidth Allocator (requires `libmemkind`) - UMF High...

This is a little confusing. I don't get different performance if I swap the order of benchmarks. ``` icx -qopenmp -O3 -march=native -I/home/ru37geh/local/include -L/home/ru37geh/local/lib64 triad.c -lnuma -lumf -o triad UMF...