A multi-core processor uses a write-back L1 data cache with a MESI protocol and a 64-byte line size. Core 0 and Core 1 both have a shared copy (State: S) of a cache line containing an array. Core 0 executes a loop writing a 4-byte integer to its local copy of index 0 of the array every 10 cycles, while Core 1 continuously reads a 4-byte integer from index 1 of the same array every 5 cycles. Assuming zero structural hazards and a bus arbitration latency of 2 cycles, what is the primary architectural bottleneck causing severe performance degradation?
Since index 0 and index 1 reside within the same 64-byte cache line, Core 0's writes will repeatedly invalidate the line in Core 1's cache via the MESI protocol. Core 1 then experiences a cache miss on its next read, forcing a line transfer over the interconnect, only for Core 0 to invalidate it again on its next write loop. This phenomenon is known as false sharing.