Single Processor Cache and Context Switching

Context Switching in Operating Systems

• Illustration of Context Switching

  • Operating system swaps execution between processes P1 and P2 to share CPU resources.

  • Process Control Blocks (PCBs) for P1 and P2 reside in memory.

  • CPU values change depending on the currently executing process.

• Definition of Context Switch

  • A context switch is a mechanism used by the operating system to switch execution from one process to another.

  • An example of this occurs when the OS switches from executing process P1 to process P2, and vice versa.

• Cost of Context Switching

  • Context switching can be expensive due to:

  • Direct Costs

    • These involve the number of CPU cycles needed to load and store values of the PCBs to and from memory.

  • Indirect Costs

    • While process P1 runs, its data is stored in the processor cache.

    • Modern CPUs utilize a multi-level cache hierarchy (L1, L2, etc.), increasing in size but potentially slower at each level.

    • Access to cache is significantly faster compared to memory.

    • Example: Access time for cache is measured in a few cycles, while memory access can take hundreds of cycles.

• Cache Behavior During Context Switching

  • When P1 executes, its data is likely to reside in the cache, resulting in a "hot" cache situation.

  • Upon switching to P2, P1's data may be replaced in the cache to accommodate P2’s data.

  • When P1 is scheduled to execute again, its data will not be in the cache, leading to longer access times from memory known as a "cold cache" situation.

• Implications of Cold Cache

  • Running with a cold cache results in increased latency for data accesses due to slower memory access times.

  • Increased latency negatively impacts the execution speed of processes.

  • It is essential to limit the frequency of context switches to reduce the overhead associated with cold caches.