I have a XenServer 7.5 environment where I'm running a single Windows VM with 16 total vCPU's. My server has dual Xeon E5-2620v4 CPU's with 8cores/16threads each clocked at 2.1GHz nominal and a Max Turbo at 3GHz.

My question is whether allowing the CPU's to be automatically managed by XenServer will be better performing, or pinning the CPU to 16 threads would be a better? My rational that pinning the 16 threads would be that I would be spreading the load to 8 threads per physical CPU, therefore increasing the speed at which it could Turbo.

My workloads are generally a mix of single and multithreaded so performance on both types is important.

  • 1
    Trying to pin across NUMA nodes will probably dramatically reduce your performance. Let it go, and use the other CPU for something else. – Michael Hampton Jul 29 '18 at 16:27

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.