Opening Caches (Digital Design and Computer Architecture, ETH Zürich, Spring 2025, Lecture 22)
Caches (Digital Design and Computer Architecture, ETH Zürich, Spring 2025, Lecture 22)
- Length
- 1 h 51 min
- Level
- Undergraduate
- Recorded
- ETH Zürich2025
Chapter 1 of 12: Before the lecture
Mutlu shows why true LRU is costly in hardware and when random replacement beats it, measured against Belady's OPT.
Cache memory design begins with the conflict between fast storage and large storage, which a hierarchy resolves by exploiting temporal and spatial locality. Mutlu builds address mapping from direct-mapped through set-associative to fully-associative caches, weighing hit rate against latency and hardware cost. He then works through replacement policies, contrasting true LRU and its hardware approximations with random replacement and Belady's OPT. Write-back versus write-through, dirty bits, write-miss allocation and subblock caches follow. The last part covers multi-level hierarchies, serial versus parallel access, inclusive and exclusive designs, and GPU scratchpad memories, assuming basic digital design and byte-addressable memory.
Cache memory design begins with the conflict between fast storage and large storage, which a hierarchy resolves by exploiting temporal and spatial locality. Mutlu builds address mapping from direct-mapped through set-associative to fully-associative caches, weighing hit rate against latency and hardware cost. He then works through replacement policies, contrasting true LRU and its hardware approximations with random replacement and Belady's OPT. Write-back versus write-through, dirty bits, write-miss allocation and subblock caches follow. The last part covers multi-level hierarchies, serial versus parallel access, inclusive and exclusive designs, and GPU scratchpad memories, assuming basic digital design and byte-addressable memory.

This lecture
Lecture 25 of 35 of Digital Design and Computer Architecture.
Choose a node to see how it connects to this lecture. What comes before is on the leftat the top. What comes after is on the rightat the bottom. The ring is what it teaches. People and places are above. Other lectures and books are below.
Press a chosen node again to open it.
Your comment is kept while you sign up.Sign in

Digital Systems Engineering / P&P
Memory Systems
Slave Memories and Dynamic Storage Allocation
Computing Handbook
A Case for MLP-Aware Cache Replacement
A study of replacement algorithms for a virtual-storage computer