Processor Scheduling and Memory Management Flashcards
What is push migration?
A form of load balancing. The load of each processor is periodically checked by a separate process. If the workload is considered imbalanced then threads are moved from busy processors to less busy processors.
What is pull migration?
A form of load balancing. Idle processors pull awaiting threads from busy processors.
Why is load balancing only required when each processor has it’s own private queue and not when there is a common ready queue?
When there is a common ready queue idle processors will automatically pull a process from the queue and so the load is automatically balanced.
What is processor affinity?
Keeping a given thread running on the same processor instead of moving it so that the thread can make use of the cache memory on the given processor.
What is a warm cache?
When the thread can find data it needs in the cache instead of having to go back and find it in the main memory.
How does the Linux scheduler efficiently utilise both load balancing and processor affinity?
It groups the cores into domains. Each core, while still having its own L1 cache, will share a L2 cache with other cores in the same domain. Cores in the same domain can then perform load balancing while still benefiting from processor affinity and being able to utilise cache.
