System and method of asymmetric system description for optimized scheduling

Claim 1

An asymmetric information handling system comprising: a system board having a plurality of sockets; a plurality of processors, each processor disposed in a respective socket; a plurality of interconnect links configured to provide point-to-point links between at least some of the sockets; and a plurality of memories corresponding to the processors; wherein one of the processors is operable to: determine an arrangement of the processors, the memories, and the interconnect links wherein the arrangement includes a plurality of system localities, and wherein each system locality includes one of the processors and one of the memories; determine a first value for the each of the processors, a second value for each of the memories, and a third value for each of the interconnect links, wherein the first value of the each processor is based on a weighted processor bandwidth capacity of the each processor, wherein the second value of the each memory is based on a weighted memory bandwidth capacity of the each memory, and wherein the third value of the each interconnect link is based on a weighted interconnect link bandwidth capacity of the each interconnect link; calculate interconnect link bandwidth values for the each of the interconnect links based at least in part on the first value of the each processor, the second value of the each memory, and the third value of the each interconnect link; and populate an interconnect bandwidth table using the interconnect link bandwidth values, wherein an operating system executing on the asymmetric information handling system prioritize resources of the asymmetric information handling system by selecting an interconnect link having maximum available bandwidth from the interconnect bandwidth table, and wherein the operating system allocates processes among a subset of the system localities in the asymmetric information handling system based on the prioritization.

Claim 8

A method to augment a scheduler of an asymmetric information handling system, the method comprising: determining an arrangement of a plurality of processors, a plurality of memories and a plurality of interconnect links, wherein the arrangement includes a plurality of system localities, and wherein each system locality includes one of the processors and one of the memories; determining a first value for each processor of the processors, a second value for each memory of the memories, and a third value for each interconnect link of the interconnect links, wherein the first value of the each processor is based on a weighted processor bandwidth capacity of the processor, wherein the second value of the each memory is based on a weighted memory bandwidth capacity of the each memory, and wherein the third value of the interconnect link is based on a weighted interconnect link bandwidth capacity of the each interconnect link; calculating an interconnect link bandwidth value for each of the interconnect links providing point-to-point links between at least some of the system localities, based at least in part on the first value of the each processor, the second value of the each memory, and the third value of the each interconnect link; and populating a hardware memory with an interconnect bandwidth table using the interconnect link bandwidth values, wherein an operating system executing on the asymmetric information handling system prioritizes resources of the asymmetric information handling system, by selecting an interconnect link having maximum available bandwidth from the interconnect bandwidth table and wherein the operating system processes among a subset of the system localities in the asymmetric information handling system based on the prioritization.

Claim 9

A method to augment a scheduler of an asymmetric information handling system, the method comprising: determining an arrangement of a plurality of processors, a plurality of memories and a plurality of interconnect links, wherein the arrangement includes a plurality of system localities, and wherein each system locality includes one of the processors and one of the memories; determining a first value for each processor of the processors, a second value for each memory of the memories, and a third value for each interconnect link of the interconnect links, wherein the first value of the each processor is based on a weighted processor bandwidth capacity of the processor, wherein the second value of the each memory is based on a weighted memory bandwidth capacity of the each memory, and wherein the third value of the interconnect link is based on a weighted interconnect link bandwidth capacity of the each interconnect link; calculating an interconnect link bandwidth value for each of the interconnect links providing point-to-point links between at least some of the system localities, based at least in part on the first value of the each processor, the second value of the each memory, and the third value of the each interconnect link; and populating a hardware memory with an interconnect bandwidth table using the interconnect link bandwidth values, wherein an operating system executing on the asymmetric information handling system prioritizes resources of the asymmetric information handling system, by selecting an interconnect link having maximum available bandwidth from the interconnect bandwidth table and wherein the operating system processes among a subset of the system localities in the asymmetric information handling system based on the prioritization.

Claim 12

A method for optimizing an asymmetric information handling system, the method comprising: determining system localities in the asymmetric information handling system, wherein at least one of the system localities includes a processor and a memory; calculating a first value of the processor and a second value of the memory for each of the system localities, wherein the first value of each processor is based on a weighted processor bandwidth capacity of the each processor and wherein the second value of each memory is based on a weighted memory bandwidth capacity of the memory; determining possible pairings of the system localities; calculating a bandwidth value for each of a plurality of interconnect links providing point to point link between each of the possible pairings based on the first value of each processor and the second value of each memory; and populating an interconnect bandwidth table stored in a hardware memory using the calculated bandwidth values, wherein an operating system executing on the asymmetric information handling system prioritizes resources of the asymmetric information handling system by selecting an interconnect link having maximum available bandwidth from the interconnect bandwidth table, and wherein the operating system allocates processes among a subset of the system localities in the asymmetric information handling system based on the prioritization.

Claim 13

A method for optimizing an asymmetric information handling system, the method comprising: determining system localities in the asymmetric information handling system, wherein at least one of the system localities includes a processor and a memory; calculating a first value of the processor and a second value of the memory for each of the system localities, wherein the first value of each processor is based on a weighted processor bandwidth capacity of the each processor and wherein the second value of each memory is based on a weighted memory bandwidth capacity of the memory; determining possible pairings of the system localities; calculating a bandwidth value for each of a plurality of interconnect links providing point to point link between each of the possible pairings based on the first value of each processor and the second value of each memory; and populating an interconnect bandwidth table stored in a hardware memory using the calculated bandwidth values, wherein an operating system executing on the asymmetric information handling system prioritizes resources of the asymmetric information handling system by selecting an interconnect link having maximum available bandwidth from the interconnect bandwidth table, and wherein the operating system allocates processes among a subset of the system localities in the asymmetric information handling system based on the prioritization.