Avalon® Interface Specifications

ID 683091
Date 1/24/2022
Public

A newer version of this document is available. Customers should click here to go to the newest version.

Document Table of Contents

3.5.5.3. Line–Wrapped Bursts

Processors with instruction caches gain efficiency by using line-wrapped bursts. When a processor requests data that is not in the cache, the cache controller must refill the entire cache line. For a processor with a cache line size of 64 bytes, a cache miss causes 64 bytes to be read from memory. If the processor reads from address 0xC when the cache miss occurred, then an inefficient cache controller could issue a burst at address 0, resulting in data from read addresses 0x0, 0x4, 0x8, 0xC, 0x10, 0x14, 0x18, . . . 0x3C. The requested data is not available until the fourth read. With line-wrapping bursts, the address order is 0xC, 0x10, 0x14, 0x18, . . . 0x3C, 0x0, 0x4, and 0x8. The requested data is returned first. The entire cache line is eventually refilled from memory.