Fewer Lines Same Bytes

A 32 KB, 8-way set-associative L1 data cache. A program chases a linked list of 32-byte nodes scattered randomly across a 4 MB heap, so successive nodes never share a line.

The architects propose growing the line size from 64 bytes to 256 bytes, keeping total capacity at 32 KB. What happens to this workload?

The cache behaves as if direct mapped, because a larger line size forces the associativity down to 1

The cache can hold 128 distinct nodes instead of 512, and each miss moves 256 bytes to deliver 32 useful ones, so both miss rate and bandwidth waste rise

Hit rate improves, since each miss now brings in four times as much surrounding data

Nothing measurable changes; 32 KB of capacity is 32 KB either way