0% found this document useful (0 votes)
85 views3 pages

Understanding Memory Hierarchy in Computers

The document discusses the memory hierarchy and how it works. It explains that the memory hierarchy consists of different levels of memory with each level getting slower and larger in size the further it is from the processor. Data is copied between two adjacent levels at a time, with the upper level being faster but smaller and the lower level being slower but larger. There is a growing performance gap between processors and memory due to processors increasing in speed faster than memory. The memory hierarchy aims to bridge this gap by taking advantage of temporal and spatial locality in memory access patterns.

Uploaded by

gandhiraja
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
85 views3 pages

Understanding Memory Hierarchy in Computers

The document discusses the memory hierarchy and how it works. It explains that the memory hierarchy consists of different levels of memory with each level getting slower and larger in size the further it is from the processor. Data is copied between two adjacent levels at a time, with the upper level being faster but smaller and the lower level being slower but larger. There is a growing performance gap between processors and memory due to processors increasing in speed faster than memory. The memory hierarchy aims to bridge this gap by taking advantage of temporal and spatial locality in memory access patterns.

Uploaded by

gandhiraja
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

An Expanded View of the Memory System

ECE4680
Computer Organization and Architecture Processor

Memory Hierarchy Control


Memory
Memory

Memory

Memory
Memory
Datapath

Speed: Fastest Slowest


Size: Smallest Biggest
Cost: Highest Lowest

ECE4680 Memory.1 2002-4-21 ECE4680 Memory.4 2002-4-21

The Big Picture: Where are We Now? Memory Hierarchy: Principles of Operation

°At any given time, data is copied between only 2 adjacent levels:
°The Five Classic Components of a Computer • Upper Level: the one closer to the processor
- Smaller, faster, and uses more expensive technology
Processor • Lower Level: the one further away from the processor
Input
- Bigger, slower, and uses less expensive technology
Control
Memory °Block:
• The minimum unit of information that can either be present or not
Datapath Output present in the two level hierarchy

Lower Level
To Processor Upper Level Memory
°Today’s Topic: Memory System Memory
Blk X
From Processor Blk Y

ECE4680 Memory.2 2002-4-21 ECE4680 Memory.5 2002-4-21

Who Cares About the Memory Hierarchy? Memory Hierarchy: Terminology

Processor-DRAM Memory Gap (latency) °Hit: data appears in some block in the upper level (example: Block X)
• Hit Rate: the fraction of memory access found in the upper level
µProc • Hit Time: Time to access the upper level which consists of
1000 60%/yr.
CPU RAM access time + Time to determine hit/miss
(2X/1.5yr)
“Moore’s Law” °Miss: data needs to be retrieve from a block in the lower level (Block Y)
Performance

100 Processor-Memory • Miss Rate = 1 - (Hit Rate)


Performance Gap: • Miss Penalty: Time to replace a block in the upper level +
(grows 50% / year) Time to deliver the block to the processor
10
DRAM °Hit Time << Miss Penalty
9%/yr.
DRAM
(2X/10 yrs) Lower Level
1 To Processor Upper Level Memory
Memory
2000
2001
2002
1999
1985

1988

1998
1981
1982
1983
1984

1986
1987

1989
1990
1991
1992
1993
1994
1995
1996
1997

Block X
From Processor Block Y

Time
ECE4680 Memory.3 2002-4-21 ECE4680 Memory.6 2002-4-21
Memory Hierarchy: How Does it Work? Technology Trends

°Temporal Locality (Locality in Time): If an item is referenced, it will tend Capacity Speed (latency)
to be referenced again soon.
Logic: 2x in 3 years 2x in 3 years
• Keep more recently accessed data items closer to the processor
DRAM: 4x in 3 years 2x in 10 years
°Spatial Locality (Locality in Space): If an item is referenced, items
whose addresses are close by tend to be referenced soon.
Disk: 4x in 3 years 2x in 10 years
• Move blocks consists of contiguous words to the upper levels

DRAM
Year Size Cycle Time
1980 1000:1! 64 Kb 2:1! 250 ns
Lower Level 1983 256 Kb 220 ns
To Processor Upper Level Memory
Memory 1986 1 Mb 190 ns
Blk X 1989 4 Mb 165 ns
From Processor Blk Y 1992 16 Mb 145 ns
1995 64 Mb 120 ns

ECE4680 Memory.7 2002-4-21 ECE4680 Memory.10 2002-4-21

Memory Hierarchy of a Modern Computer System SPARCstation 20’s Memory System Overview

°By taking advantage of the principle of locality:


• Present the user with as much memory as is available in the Memory
cheapest technology. Controller Memory Bus (SIMM Bus) 128-bit wide datapath

Memory Module 7

Memory Module 5

Memory Module 4

Memory Module 3

Memory Module 1
• Provide access at the speed offered by the fastest technology.

Memory Module 2
Memory Module 6

Memory Module 0
Processor Processor Bus (Mbus) 64-bit wide

Control
Secondary
Storage Processor Module (Mbus Module)
Second Main
(Disk) SuperSPARC Processor
On-Chip
Registers

Level Memory
Cache

Datapath Cache (DRAM) Instruction


(SRAM) External Cache
Cache Register
File
Data
Cache
Speed (ns): 1s 10s 100s 10,000,000s (10s ms)
Size (bytes): 100s Ks Ms Gs

ECE4680 Memory.8 2002-4-21 ECE4680 Memory.11 2002-4-21

Memory Hierarchy Technology SPARCstation 20’s Memory Module


°Supports a wide range of sizes:
°Random Access:
• Smallest: 4 MB = 16 2Mb DRAM chips, 8 KB of Page Mode SRAM
• “Random” is good: access time is the same for all locations
• Biggest: 64 MB = 32 16Mb chips, 16 KB of Page Mode SRAM
• DRAM: Dynamic Random Access Memory
- High density, low power, cheap, slow
DRAM Chip 15
- Dynamic: need to be “refreshed” regularly
512 cols
• SRAM: Static Random Access Memory
256K x 8
- Low density, high power, expensive, fast
= 2 MB
- Static: content will last “forever” DRAM Chip 0 One page
512 rows

°“Non-so-random” Access Technology: 256K x 8 512 × 8 SRAM


• Access time varies from location to location and from time to time = 2 Mb
8 bits bits<127:0>
• Examples: Disk, tape drive, CDROM

°The next two lectures will concentrate on random access technology 512 × 8 SRAM
• The Main Memory: DRAMs bits<7:0> Memory Bus<127:0>
• Caches: SRAMs

ECE4680 Memory.9 2002-4-21 ECE4680 Memory.12 2002-4-21


Summary:

°Two Different Types of Locality:


• Temporal Locality (Locality in Time): If an item is referenced, it will
tend to be referenced again soon.
• Spatial Locality (Locality in Space): If an item is referenced, items
whose addresses are close by tend to be referenced soon.

°By taking advantage of the principle of locality:


• Present the user with as much memory as is available in the
cheapest technology.
• Provide access at the speed offered by the fastest technology.

°DRAM is slow but cheap and dense:


• Good choice for presenting the user with a BIG memory system

°SRAM is fast but expensive and not very dense:


• Good choice for providing the user FAST access time.

ECE4680 Memory.13 2002-4-21

You might also like