Skip to content

Memory hierarchy performance

Memory hierarchy performance is a fundamental concept in computer science concerning the speed and cost differences between various memory technologies.^[600-developer__principle__cache.md]

Performance disparities

The performance gap between different levels of the memory hierarchy is substantial, often quantified in orders of magnitude.^[600-developer__principle__cache.md]

  • Memory vs. Network: Accessing data over a network is approximately 100 times slower than accessing it from system memory (RAM).^[600-developer__principle__cache.md]
  • Disk Seek vs. Read: There is a significant latency penalty associated with mechanical movement in traditional hard drives; the time required to perform a single hard disk seek is approximately half the time required to read 1 MB of data.^[600-developer__principle__cache.md]

Implications

These discrepancies illustrate the "Memory Wall" problem, where processor speeds far outpace the ability to supply data from main memory or storage, necessitating complex [[caching]] strategies to hide latency and maintain system throughput.^[600-developer__principle__cache.md]

  • [[Cache]]
  • [[Latency]]

Sources

^[600-developer__principle__cache.md]