Description of problem: Sometimes, it's better to ignore HT cores, for performance reasons. Currently even with cpu_allocation_ratio set correctly, a node with 2 cores per sockets and 2 threads per core will be reported with 4pCPUs where some would prefer to ignore the HT count. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:[ykaul@ykaul ~]$ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 4 On-line CPU(s) list: 0-3 Thread(s) per core: 2 Core(s) per socket: 2 Socket(s): 1 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 42 Stepping: 7 CPU MHz: 2801.000 BogoMIPS: 5581.43 Virtualization: VT-x L1d cache: 32K L1i cache: 32K L2 cache: 256K L3 cache: 4096K NUMA node0 CPU(s): 0-3 2012-10-14 17:39:31 INFO nova.compute.resource_tracker [-] Compute_service record updated for ykaul.tlv.redhat.com 2012-10-14 17:40:32 AUDIT nova.compute.resource_tracker [-] Free ram (MB): 7358 2012-10-14 17:40:32 AUDIT nova.compute.resource_tracker [-] Free disk (GB): 49 2012-10-14 17:40:32 AUDIT nova.compute.resource_tracker [-] Free VCPUS: 4
There's not enough data here to make a case for upstream inclusion. If anybody has the data to justify changes in this area, please reopen.