Comparisons
Two things routinely conflated, put side by side. Neither column is the winner — what decides is the workload and the machine.
ISA vs MicroarchitectureArray traversal vs Linked-list traversalLatency vs BandwidthCPU vs GPUCPU cache vs OS page cacheHardware thread vs OS threadPolling vs InterruptsBranch vs Branchless
Polling vs Interrupts
Ask repeatedly, or be told. Interrupts free the CPU between events but cost a disruption each time; polling burns cycles but responds immediately and predictably. At very high event rates, polling wins — which is why high-performance networking often abandons interrupts.
PollingOpen lesson →
Strengths
Lowest and most predictable latency; no interrupt overhead
Costs
Burns a core doing nothing when events are rare
Use when
Event rates are high and latency matters more than efficiency
InterruptsOpen lesson →
Strengths
CPU is free to do other work between events
Costs
Per-event cost: pipeline disruption, plus cache and TLB pollution
Use when
Events are rare or unpredictable and the CPU has other work
| Dimension | Polling | Interrupts |
|---|---|---|
| Cost when idle | A whole core | Nothing |
| Cost per event | Nearly nothing | Disruption plus handler |
| Best at | Millions of events per second | Occasional events |