Memory System Architecture
In this section, we’re going to learn about what I believe is the most important part of the computer’s architecture: the memory system. Moving data from point A to point B is one of the main bottlenecks in all of today’s systems. Shown above is a picture of HBM (high-bandwidth memory) which is a relatively new memory technology primarily used in GPUs and other accelerators to try to overcome this memory bottleneck.
Memory intro and Technology
In this section, we’ll be talking about how memory works and memory technology. We’ll find that while there are many different memory technologies, none are “best”.
- Turing Lecture
- Computing technology past to present
- Moore’s Law and Dennard scaling
- Energy and Power of CMOS Devices
- QUIZ CMOS Energy, Power, and Moore’s Law
- Reading for this section
- Introduction: Why does memory architecture matter?
- Memory technology
- Moving data around
- QUIZ Memory technology
Caches and the Memory Hierarchy
Since no one technology is the best, we want to create a memory hierarchy with caches. We will discuss the underlying principles that make caches efficient, and how hardware caches are designed.
- Memory hierarchy introduction
- Basic caching
- Memory hierarchy
- Cache design
- Replacement policies
- Write-back policies
- Summary of caches
Assignment 4 DUE 2/28
Virtual memory
Finally, we’ll step up a level and look at how operating systems allow multiple programs to share a single memory through virtualization.