Dynamic optimization of thread assignments for application workloads in parallel computing
View Patent ↗A method for dynamic optimization of thread assignments for application workloads in an simultaneous multi-threading (SMT) computing environment includes monitoring and periodically recording an operational status of different processor cores each supporting a number of threads of the thread pool of the SMT computing environment and also operational characteristics of different workloads of a computing application executing in the SMT computing environment. The method further can include identifying by way of the recorded operational characteristics a particular one of the workloads demonstrating a threshold level of activity. Finally, the method can include matching a recorded operational characteristic of the particular one of the workloads to a recorded status of a processor core best able amongst the different processor cores to host execution in one or more threads of the particular one of the workloads and directing the matched processor core to host execution of the particular one of the workloads.
1. A method for dynamic optimization of thread assignments for application workloads in a simultaneous multi-threading (SMT) computing environment, the method comprising:
monitoring and periodically recording coefficients associated with an operational status of different processor cores hosting threads accessible in a thread pool of the SMT computing environment and also recording operational characteristics of different workloads of a computing application executing in the SMT computing environment, wherein the operational characteristics include a number of clock ticks consumed by a corresponding one of the workloads, a higher number of clock ticks indicating a higher degree of activity, and a measurement of cycles per instruction (CPI) consumed by a corresponding one of the workloads and a variance in measurements of CPI consumed by the corresponding one of the workloads over time, a lower variance indicating greater operational stability;
identifying by way of the recorded operational characteristics of the different workloads a particular one of the workloads demonstrating both a threshold level of variance indicating operational stability and a threshold degree of activity,
matching a recorded operational characteristic of the particular one of the workloads to recorded coefficients associated with the operational status of a processor core best able amongst the different processor cores to host in one or more threads execution of the particular one of the workloads; and,
directing the matched processor core to host execution of the particular one of the workloads.
2. The method of claim 1 , further comprising:
determining that no one of the processor cores is available to host the execution of the particular one of the workloads; and,
re-arranging an existing assignment of ones of the different workloads to corresponding ones of the processor cores to optimize matching of the recorded operational characteristics of the ones of the different workloads to the recorded coefficients associated with the operational status of the processor cores.