Chapter 7 – Memory - iiusatech.comiiusatech.com/murdocca/CAO/SlidesPDF/Ch07CAO.pdf · 7-1 Chapter 7- Memory ... Functional Behavior of a RAM Cell Static RAM cell (a) ... Cache Read

7-1 Chapter 7 - Memory

Computer Architecture and Organization by M. Murdocca and V. Heuring © 2007 M. Murdocca and V. Heuring

Computer Architecture andOrganization

Miles Murdocca and Vincent Heuring

Chapter 7 – Memory



Chapter Contents7.1 The Memory Hierarchy7.2 Random-Access Memory7.3 Memory Chip Organization7.4 Case Study: Rambus Memory7.5 Cache Memory7.6 Virtual Memory7.7 Advanced Topics7.8 Case Study: Associative Memory in Routers7.9 Case Study: The Intel Pentium 4 Memory System



The MemoryHierarchy



Functional Behavior of a RAM Cell

Static RAM cell (a) and dynamic RAM cell (b).



Simplified RAM Chip Pinout



A Four-WordMemory withFour Bits perWord in a 2DOrganization



A Simplified Representation of theFour-Word by Four-Bit RAM



2-1/2D Organization of a 64-Word byOne-Bit RAM



Two Four-Word by Four-Bit RAMs areUsed in Creating a Four-Word by

Eight-Bit RAM



Two Four-Word by Four-Bit RAMs Makeup an Eight-Word by Four-Bit RAM



Single-In-LineMemoryModule

• 256 MB dual in-linememory module organizedfor a 64-bit word with 1616M × 8-bit RAM chips(eight chips on each sideof the DIMM).



Single-In-Line Memory

Module

• Schematic diagram of256 MB dual in-linememory module.(Source: adapted fromhttp://www-s.ti.com/sc/ds/tm4en64kpu.pdf.)



A ROM Stores Four Four-Bit Words



A Lookup Table (LUT) Implements anEight-Bit ALU



Flash Memory

• (a) External view of flash memory module and (b) flash moduleinternals. (Source: adapted from HowStuffWorks.com.)



Cell Structure for Flash Memory

• Current flows from source to drain when a sufficient negative charge isplaced on the dielectric material, preventing current flow through theword line. This is the logical 0 state. When the dielectric material is notcharged, current flows between the bit and word lines, which is thelogical 1 state.



Rambus Memory• Comparison of DRAM and RDRAM configurations.



Rambus Memory• Rambus technology on the Nintendo 64 motherboard (left)

enables cost savings over the conventional Sega Saturnmotherboard design (right).

• Nintendo 64 game console:



Placement of Cache Memory in aComputer System

• The locality principle: a recently referenced memory location is likely tobe referenced again (temporal locality); a neighbor of a recentlyreferenced memory location is likely to be referenced (spatial locality).



An Associative Mapping Scheme for aCache Memory



Associative Mapping Example• Consider how an access to memory location (A035F014)16 is mapped to

the cache for a 232 word memory. The memory is divided into 227 blocksof 25 = 32 words per block, and the cache consists of 214 slots:

• If the addressed word is in the cache, it will be found in word (14)16 of aslot that has tag (501AF80)16, which is made up of the 27 mostsignificant bits of the address. If the addressed word is not in thecache, then the block corresponding to tag field (501AF80)16 is broughtinto an available slot in the cache from the main memory, and thememory reference is then satisfied from the cache.



Associative Mapping Area Allocation• Area allocation for associative mapping scheme based on bits stored:



Replacement Policies• When there are no available slots in which to place a block, a

replacement policy is implemented. The replacement policy governsthe choice of which slot is freed up for the new block.

• Replacement policies are used for associative and set-associativemapping schemes, and also for virtual memory.

• Least recently used (LRU)

• First-in/first-out (FIFO)

• Least frequently used (LFU)

• Random

• Optimal (used for analysis only – look backward in time and reverse-engineer the best possible strategy for a particular sequence ofmemory references.)



A Direct Mapping Scheme for CacheMemory



Direct Mapping Example• For a direct mapped cache, each main memory block can be mapped to

only one slot, but each slot can receive more than one block. Considerhow an access to memory location (A035F014)16 is mapped to thecache for a 232 word memory. The memory is divided into 227 blocks of25 = 32 words per block, and the cache consists of 214 slots:

• If the addressed word is in the cache, it will be found in word (14)16 of slot(2F80)16, which will have a tag of (1406)16.



Direct Mapping Area Allocation• Area allocation for direct mapping scheme based on bits stored:



A Set Associative Mapping Schemefor a Cache Memory



Set-Associative Mapping Example• Consider how an access to memory location (A035F014)16 is mapped to

the cache for a 232 word memory. The memory is divided into 227 blocksof 25 = 32 words per block, there are two blocks per set, and the cacheconsists of 214 slots:

• The leftmost 14 bits form the tag field, followed by 13 bits for the set field,followed by five bits for the word field:



Set Associative Mapping AreaAllocation

• Area allocation for set associative mapping scheme based on bits stored:



Cache Read and Write Policies



Hit Ratios and Effective Access Times• Hit ratio and effective access time for single level cache:

• Hit ratios and effective access time for multi-level cache:



Direct Mapped Cache Example• Compute hit ratio and

effective access time fora program that executesfrom memory locations48 to 95, and then loops10 times from 15 to 31.

• The direct mappedcache has four 16-wordslots, a hit time of 80 ns,and a miss time of 2500ns. Load-through isused. The cache isinitially empty.



Table of Events for Example Program



Calculation of Hit Ratio and EffectiveAccess Time for Example Program



Multi-level Cache MemoryAs an example, consider a two-level cache in which the L1 hit time is 5 ns,the L2 hit time is 20 ns, and the L2 miss time is 100 ns. There are 10,000memory references of which 10 cause L2 misses and 90 cause L1 misses.Compute the hit ratios of the L1 and L2 caches and the overall effectiveaccess time.

H1 is the ratio of the number of times the accessed word is in the L1 cacheto the total number of memory accesses. There are a total of 85 (L1) and15 (L2) misses, and so:

(Continued on next slide.)



Multi-level Cache Memory (Cont’)H2 is the ratio of the number of times the accessed word is in the L2 cacheto the number of times the L2 cache is accessed, and so:

The effective access time is then:

= 5.23 ns per access



Neat Little LRU Algorithm• A sequence is shown for the Neat Little LRU Algorithm for a cache with

four slots. Main memory blocks are accessed in the sequence: 0, 2, 3,1, 5, 4.



Cache Coherency• The goal of cache coherence is to ensure that every cache sees the

same value for a referenced location, which means making sure thatany shared operand that is changed is updated throughout the system.

• This brings us to the issue of false sharing, which reduces cacheperformance when two operands that are not shared betweenprocesses share the same cache line. The situation is shown below.The problem is that each process will invalidate the other’s cache linewhen writing data without a real need, unless the compiler prevents this.



Overlays• A partition graph for a program with a main routine and three subroutines:



Virtual Memory• Virtual memory is stored in a hard disk image. The physical memory

holds a small number of virtual pages in physical page frames.

• A mapping between a virtual and a physical memory:



Page Table• The page table maps between virtual memory and physical memory.



Using the Page Table

• A virtual address istranslated into a physicaladdress:

Typical page table entry



Using the Page Table (cont’)• The

configuration ofa page tablechanges as aprogramexecutes.

• Initially, the pagetable is empty.In the finalconfiguration,four pages are inphysicalmemory.



Segmentation• A segmented memory allows two users to share the same word

processor code, with different data spaces:



Fragmentation

• (a) Free areaof memoryafter initial-ization; (b)afterfragment-ation; (c)aftercoalescing.



Translation Lookaside Buffer• An example TLB holds 8 entries for a system with 32 virtual pages and

16 page frames.



Putting it All Together• An example TLB holds 8 entries for a system with 32 virtual pages and

16 page frames.



Content Addressable Memory –Addressing

• Relationships between random access memory and contentaddressable memory:



Overview of CAM

• Source: (Foster,C. C., ContentAddressableParallelProcessors, VanNostrandReinholdCompany, 1976.)



Addressing Subtrees for a CAM



Associative Memory in Routers

• A simple network withthree routers.

• The use of associativememories in high-end routersreduces the lookup time byallowing a search to be performed in a single operation.

• The search is based on the destination address, rather than thephysical memory address.

• Access methods for this memory have been standardized into aninterface interoperability agreement by the Network Processing Forum.



Block Diagram of Dual-Read RAM• A dual-read or dual-port

RAM allows any twowords to besimultaneously readfrom the same memory.



The Intel 4 Pentium Memory System

Documents

Chapter 7 – Memory - iiusatech.comiiusatech.com/murdocca/CAO/SlidesPDF/Ch07CAO.pdf · 7-1 Chapter 7- Memory ... Functional Behavior of a RAM Cell Static RAM cell (a) ... Cache Read