Memory Hierarchy

Which technique, when fetching a missed line from memory, sends the word the CPU needs first and the rest afterward?

  1. ACritical word first
  2. BWrite-allocate
  3. CVictim cache
Show answer and explanation

Answer: A. Critical word first

Waiting for the whole 64-byte line before handing over data lengthens the CPU's stall. Critical word first transfers the requested word first and passes it to the CPU immediately, filling in the rest of the line afterward, which shortens the effective miss penalty. Write-allocate concerns whether a write miss loads the line, and a victim cache is a small buffer for evicted lines; neither relates to transfer order.

  • A. Correct. Sending the needed word first shortens the wait.
  • B. Write-allocate concerns how write misses are handled.
  • C. A victim cache is a buffer for evicted lines.
Practice this topic Back to topic

More Memory Hierarchy questions

  1. Why does fusing the two loops above into one (loop fusion) tend to be faster when N is large?
  2. Which pair gives the reason a large matrix transpose (b[j][i] = a[i][j]) tends to cause many cache misses, and a typical countermeasure?
  3. In a typical computer's memory hierarchy, which property holds more strongly for upper levels (closer to the CPU)?
  4. Why must cache designs be compared by average memory access time rather than miss rate alone?
  5. Running two programs alternately on the same core can make each slower than running alone. What is the best explanation from a cache perspective?