Thanks for reaching out to us.
Please try out the below parameter and let me know the observation
Set OpenMP* environment variables (OMP_) and extensions (KMP_).
os.environ["OMP_NUM_THREADS"] = "64"
os.environ["KMP_BLOCKTIME"] = "30"
os.environ["KMP_SETTINGS"] = "1"
Please share the below information if it still not working
1. Framework and topology details
2. Please share the code to try out from our end.
Performance optimizations are code dependent. There can be other factors influencing the performance in addition to the compute power of machines. It would be great if you could share the details of the code,framework/topology/environment setup, so that we could check it here.