dell-poweredgehp-prolianthyperthreading
I have a system running a financial trading application at a remote facility. I do not have access to the ILO/DRAC, but need to disable hyperthreading. The system runs Intel Westmere 3.33GHz X5680 hex-core CPUs. I can reboot, but want to make sure that the system does not enable hyperthreading due to performance problems. Is there a clean way to do this from within Linux?
Edit: The noht directive added to the kernel boot command line did not work. Same for RHEL.
This link describes how to match physical id, core id, and sibling count to identify which entries in /proc/cpuinfo are on the same core, but I can make no claims as to its accuracy. It also looks similar to this question.
So, after bringing a third system into the mix, and experiencing the same issue, we began to question the environment. I dug up a copy of the HP ProLiant Servers Troubleshooting Guide and found the POST problems flowchart shown below.
Carefully running through the steps in the chart, we realized that the one constant across all of the servers was a KVM switch attached to the data center crash cart. This was a consumer-class USB-enabled KVM. As per the highlighted node in the flowchart, Do you have known good KVM?, I could not answer conclusively.
So, we unplugged the servers from the KVM switch and ran an automated boot, sleep 300; reboot sequence in rc.local. The servers had no issues with this, regardless of the normal DIMM, low-voltage DIMMs, PSU wattage, etc.
This was all the result of a poor interaction with a USB KVM switch. By virtue that this was the console, it ensured we'd see the failure if we were looking for it. Self-fulfilling...
Best Answer
Newer Kernels provide a Simultaneous Multithreading (SMT) control.
You can check the state of SMT with;
Change the state with
Options are;
We have tested this with Linux Kernel 4.4.0