Chapter8<1>

DigitalDesignandComputerArchitecture,2nd Edition

Chapter8

DavidMoneyHarrisandSarahL.Harris

Chapter8<2>

Chapter8 ::Topics

• Introduction• MemorySystemPerformanceAnalysis

• Caches• VirtualMemory• Memory-MappedI/O• Summary

Chapter8<3>

Processor MemoryAddressMemWrite

WriteDataReadData

WE

CLKCLK

• Computer performance depends on:– Processor performance– Memory system performance

Memory Interface

Introduction

Chapter8<4>

In prior chapters, assumed access memory in 1 clock cycle – but hasn’t been true since the 1980’s

Processor-MemoryGap

Chapter8<5>

• Make memory system appear as fast as processor

• Use hierarchy of memories• Ideal memory:

– Fast– Cheap (inexpensive)– Large (capacity)

But can only choose two!

MemorySystemChallenge

Chapter8<6>

MemoryHierarchy

Technology Price / GB AccessTime (ns)

Bandwidth(GB/s)

Cache

Main Memory

Virtual Memory

Capacity

Spee

d

SRAM $10,000 1

DRAM $10 10 - 50

SSD $1 100,000

25+

10

0.5

0.1HDD $0.1 10,000,000

Chapter8<7>

Exploitlocalitytomakememoryaccessesfast• TemporalLocality:– Localityintime– Ifdatausedrecently,likelytouseitagainsoon– Howtoexploit: keeprecentlyaccesseddatainhigherlevelsofmemoryhierarchy

• SpatialLocality:– Localityinspace– Ifdatausedrecently,likelytousenearbydatasoon– Howtoexploit: whenaccessdata,bringnearbydataintohigherlevelsofmemoryhierarchytoo

Locality

Chapter8<8>

• Hit: data found in that level of memory hierarchy• Miss: data not found (must go to next level)

Hit Rate = # hits / # memory accesses= 1 – Miss Rate

Miss Rate = # misses / # memory accesses= 1 – Hit Rate

• Average memory access time (AMAT): average time for processor to access dataAMAT = tcache + MRcache[tMM + MRMM(tVM)]

MemoryPerformance

Chapter8<9>

• A program has 2,000 loads and stores• 1,250 of these data values in cache• Rest supplied by other levels of memory

hierarchy• What are the hit and miss rates for the cache?

MemoryPerformanceExample1

Chapter8<10>

• A program has 2,000 loads and stores• 1,250 of these data values in cache• Rest supplied by other levels of memory

hierarchy• What are the hit and miss rates for the cache?

Hit Rate = 1250/2000 = 0.625Miss Rate = 750/2000 = 0.375 = 1 – Hit Rate

MemoryPerformanceExample1

Chapter8<11>

• Suppose processor has 2 levels of hierarchy: cache and main memory

• tcache = 1 cycle, tMM = 100 cycles• What is the AMAT of the program from

Example 1?

MemoryPerformanceExample2

Chapter8<12>

• Suppose processor has 2 levels of hierarchy: cache and main memory

• tcache = 1 cycle, tMM = 100 cycles• What is the AMAT of the program from

Example 1?

AMAT = tcache + MRcache(tMM)= [1 + 0.375(100)] cycles= 38.5 cycles

MemoryPerformanceExample2

Chapter8<13>

• Amdahl’s Law: the effort spent increasing the performance of a subsystem is wasted unless the subsystem affects a large percentage of overall performance

• Co-founded 3 companies, including one called Amdahl Corporation in 1970

GeneAmdahl,1922-

Chapter8<14>

• Highestlevelinmemoryhierarchy• Fast(typically~1cycleaccesstime)• Ideallysuppliesmostdatatoprocessor• Usuallyholdsmostrecentlyaccesseddata

Cache

Chapter8<15>

• Whatdataisheldinthecache?• Howisdatafound?• Whatdataisreplaced?

Focusondataloads,butstoresfollowsameprinciples

CacheDesignQuestions

Chapter8<16>

• Ideally,cacheanticipatesneededdataandputsitincache

• Butimpossibletopredictfuture• Usepasttopredictfuture– temporalandspatiallocality:– Temporallocality: copynewlyaccesseddataintocache

– Spatiallocality: copyneighboringdataintocachetoo

Whatdataisheldinthecache?

Chapter8<17>

• Capacity(C):– numberofdatabytesincache

• Blocksize(b):– bytesofdatabroughtintocacheatonce

• Numberofblocks(B=C/b):– numberofblocksincache:B =C/b

• Degreeofassociativity(N):– numberofblocksinaset

• Numberofsets(S=B/N):– eachmemoryaddressmapstoexactlyonecacheset

CacheTerminology

Chapter8<18>

• CacheorganizedintoS sets• Eachmemoryaddressmapstoexactlyoneset• Cachescategorizedby#ofblocksinaset:–Directmapped: 1blockperset–N-waysetassociative: N blocksperset– Fullyassociative: allcacheblocksin1set

• Examineeachorganizationforacachewith:– Capacity(C =8words)– Blocksize(b =1word)– So,numberofblocks(B =8)

Howisdatafound?

Chapter8<19>

• C =8 words(capacity)• b =1 word(blocksize)• So,B =8(#ofblocks)

Ridiculouslysmall,butwillillustrateorganizations

ExampleCacheParameters

Chapter8<20>

7 (111)

00...00010000

230 Word Main Memory

mem[0x00...00]mem[0x00...04]mem[0x00...08]mem[0x00...0C]mem[0x00...10]mem[0x00...14]mem[0x00...18]mem[0x00..1C]mem[0x00..20]mem[0x00...24]

mem[0xFF...E0]mem[0xFF...E4]mem[0xFF...E8]mem[0xFF...EC]mem[0xFF...F0]mem[0xFF...F4]mem[0xFF...F8]mem[0xFF...FC]

23 Word Cache

Set Number

Address

00...0000000000...0000010000...0000100000...00001100

00...0001010000...0001100000...0001110000...0010000000...00100100

11...11110000

11...1110000011...1110010011...1110100011...11101100

11...1111010011...1111100011...11111100

6 (110)5 (101)4 (100)3 (011)2 (010)1 (001)0 (000)

DirectMappedCache

Chapter8<21>

DataTag

00Tag Set

ByteOffsetMemory

Address

DataHit

V

=

27 3

27 32

8-entry x(1+27+32)-bit

SRAM

DirectMappedCacheHardware

Chapter8<22>

# MIPS assembly code

addi $t0, $0, 5

loop: beq $t0, $0, done

lw $t1, 0x4($0)

lw $t2, 0xC($0)

lw $t3, 0x8($0)

addi $t0, $t0, -1

j loop

done:

DataTagV

00...001 mem[0x00...04]

000

0

0

00Tag Set

ByteOffsetMemory

AddressV

300100...00

100...0000...00

1mem[0x00...0C]mem[0x00...08]

Set 7 (111)Set 6 (110)Set 5 (101)Set 4 (100)Set 3 (011)Set 2 (010)Set 1 (001)Set 0 (000)

Miss Rate = ?

DirectMappedCachePerformance

Chapter8<23>

# MIPS assembly code

addi $t0, $0, 5

loop: beq $t0, $0, done

lw $t1, 0x4($0)

lw $t2, 0xC($0)

lw $t3, 0x8($0)

addi $t0, $t0, -1

j loop

done:

DataTagV

00...001 mem[0x00...04]

000

0

0

00Tag Set

ByteOffsetMemory

AddressV

300100...00

100...0000...00

1mem[0x00...0C]mem[0x00...08]

Set 7 (111)Set 6 (110)Set 5 (101)Set 4 (100)Set 3 (011)Set 2 (010)Set 1 (001)Set 0 (000)

Miss Rate = 3/15= 20%

Temporal LocalityCompulsory Misses

DirectMappedCachePerformance

Chapter8<24>

# MIPS assembly code

addi $t0, $0, 5

loop: beq $t0, $0, done

lw $t1, 0x4($0)

lw $t2, 0x24($0)

addi $t0, $t0, -1

j loop

done:

DataTagV

00...001 mem[0x00...04]

000

0

0

00Tag Set

ByteOffsetMemory

AddressV

300100...01

00

Set 7 (111)Set 6 (110)Set 5 (101)Set 4 (100)Set 3 (011)Set 2 (010)Set 1 (001)Set 0 (000)

mem[0x00...24]

Miss Rate = ?

DirectMappedCache:Conflict

Chapter8<25>

# MIPS assembly code

addi $t0, $0, 5

loop: beq $t0, $0, done

lw $t1, 0x4($0)

lw $t2, 0x24($0)

addi $t0, $t0, -1

j loop

done:

DataTagV

00...001 mem[0x00...04]

000

0

0

00Tag Set

ByteOffsetMemory

AddressV

300100...01

00

Set 7 (111)Set 6 (110)Set 5 (101)Set 4 (100)Set 3 (011)Set 2 (010)Set 1 (001)Set 0 (000)

mem[0x00...24]

Miss Rate = 10/10= 100%

Conflict Misses

DirectMappedCache:Conflict

Chapter8<26>

DataTag

Tag SetByte

OffsetMemoryAddress

Data

Hit1

V

=

01

00

32 32

32

DataTagV

=

Hit1Hit0

Hit

28 2

28 28

Way 1 Way 0

N-WaySetAssociativeCache

Chapter8<27>

# MIPS assembly code

addi $t0, $0, 5

loop: beq $t0, $0, done

lw $t1, 0x4($0)lw $t2, 0x24($0)

addi $t0, $t0, -1

j loop

done:

DataTagV DataTagV

0 0

00

0

00

0

Way 1 Way 0

Set 3Set 2Set 1Set 0

Miss Rate = ?

N-WaySetAssociativePerformance

Chapter8<28>

# MIPS assembly code

addi $t0, $0, 5

loop: beq $t0, $0, done

lw $t1, 0x4($0)lw $t2, 0x24($0)

addi $t0, $t0, -1

j loop

done:

DataTagV DataTagV

00...001 mem[0x00...04]00...10 1mem[0x00...24]

00

0

00

0

Way 1 Way 0

Set 3Set 2Set 1Set 0

Miss Rate = 2/10 = 20%

Associativity reducesconflict misses

N-WaySetAssociativePerformance

Chapter8<29>

DataTagV DataTagV DataTagV DataTagV DataTagV DataTagV DataTagV DataTagV

Reduces conflict missesExpensive to build

FullyAssociativeCache

Chapter8<30>

• Increaseblocksize:– Blocksize,b =4 words– C =8words– Directmapped(1blockperset)– Numberofblocks,B =2 (C/b =8/4=2)

DataTag

00Tag

ByteOffsetMemory

Address

Data

V

00011011

BlockOffset

32 32 32 32

32

Hit

=

Set

27

27 2

Set 1Set 0

SpatialLocality?

Chapter8<31>

DataTag

00Tag

ByteOffsetMemory

Address

Data

V

00011011

BlockOffset

32 32 32 32

32

Hit

=

Set

27

27 2

Set 1Set 0

CachewithLargerBlockSize

Chapter8<32>

addi $t0, $0, 5

loop: beq $t0, $0, done

lw $t1, 0x4($0)

lw $t2, 0xC($0)

lw $t3, 0x8($0)addi $t0, $t0, -1

j loop

done:

Miss Rate = ?

DirectMappedCachePerformance

Chapter8<33>

addi $t0, $0, 5

loop: beq $t0, $0, done

lw $t1, 0x4($0)

lw $t2, 0xC($0)

lw $t3, 0x8($0)addi $t0, $t0, -1

j loop

done:

00...00 0 11

DataTag

00Tag

ByteOffsetMemory

Address

Data

V

00011011

BlockOffset

32 32 32 32

32

Hit

=

Set

27

27 2

Set 1Set 000...001 mem[0x00...0C]

0mem[0x00...08] mem[0x00...04] mem[0x00...00]

Miss Rate = 1/15 = 6.67%

Larger blocksreduce compulsory missesthrough spatial locality

DirectMappedCachePerformance

Chapter8<34>

• Capacity:C• Blocksize:b• Numberofblocksincache:B =C/b• Numberofblocksinaset:N• Numberofsets:S =B/N

OrganizationNumber of Ways

(N)Number of Sets

(S = B/N)Direct Mapped 1 B

N-Way Set Associative 1 < N < B B / N

Fully Associative B 1

CacheOrganizationRecap

Chapter8<35>

• Cacheistoosmalltoholdalldataofinterestatonce• Ifcachefull:programaccessesdataX&evictsdataY• Capacitymiss whenaccessYagain• HowtochooseYtominimizechanceofneedingitagain?• Leastrecentlyused(LRU)replacement:theleastrecently

usedblockinasetevicted

CapacityMisses

Chapter8<36>

• Compulsory: firsttimedataaccessed• Capacity: cachetoosmalltoholdalldataofinterest

• Conflict: dataofinterestmapstosamelocationincache

Misspenalty: timeittakestoretrieveablockfromlowerlevelofhierarchy

TypesofMisses

Chapter8<37>

DataTagV0

DataTagV0

0

0

0

0

U

0 0

00

00

Way 1 Way 0

Set 3 (11)Set 2 (10)Set 1 (01)Set 0 (00)

# MIPS assembly

lw $t0, 0x04($0)lw $t1, 0x24($0)lw $t2, 0x54($0)

LRUReplacement

Chapter8<38>

DataTagV0

DataTagV0

0

0

0

0

U

mem[0x00...04]1 00...000mem[0x00...24] 100...010

00

00

DataTagV0

DataTagV0

0

0

0

0

U

mem[0x00...54]1 00...101mem[0x00...24] 100...010

00

01

(a)

(b)

Way 1 Way 0

Way 1 Way 0

Set 3 (11)Set 2 (10)Set 1 (01)Set 0 (00)

Set 3 (11)Set 2 (10)Set 1 (01)Set 0 (00)

# MIPS assembly

lw $t0, 0x04($0)lw $t1, 0x24($0)lw $t2, 0x54($0)

LRUReplacement

Chapter8<39>

• Whatdataisheldinthecache?– Recentlyuseddata(temporallocality)– Nearbydata(spatiallocality)

• Howisdatafound?– Setisdeterminedbyaddressofdata– Wordwithinblockalsodeterminedbyaddress– Inassociativecaches,datacouldbeinoneofseveralways

• Whatdataisreplaced?– Least-recentlyusedwayintheset

CacheSummary

Chapter8<40>

• Bigger caches reduce capacity misses• Greater associativity reduces conflict misses

Adapted from Patterson & Hennessy, Computer Architecture: A Quantitative Approach, 2011

MissRateTrends

Chapter8<41>

• Bigger blocks reduce compulsory misses• Bigger blocks increase conflict misses

MissRateTrends

Chapter8<42>

• Largercacheshavelowermissrates,longeraccesstimes

• Expandmemoryhierarchytomultiplelevelsofcaches

• Level1:smallandfast(e.g.16KB,1cycle)• Level2:largerandslower(e.g.256KB,2-6cycles)

• MostmodernPCshaveL1,L2,andL3cache

MultilevelCaches

Chapter8<43>

IntelPentiumIIIDie

Chapter8<44>

• Givestheillusionofbiggermemory• Mainmemory(DRAM)actsascacheforharddisk

VirtualMemory

Chapter8<45>

• Physical Memory: DRAM (Main Memory)• Virtual Memory: Hard drive

– Slow, Large, Cheap

MemoryHierarchy

Technology Price / GB AccessTime (ns)

Bandwidth(GB/s)

Cache

Main Memory

Virtual Memory

Capacity

Spee

d

SRAM $10,000 1

DRAM $10 10 - 50

SSD $1 100,000

25+

10

0.5

0.1HDD $0.1 10,000,000

Chapter8<46>

Read/WriteHead

MagneticDisks

Takes milliseconds to seek correct location on disk

HardDisk

Chapter8<47>

• Virtualaddresses– Programsusevirtualaddresses– Entirevirtualaddressspacestoredonaharddrive– SubsetofvirtualaddressdatainDRAM– CPUtranslatesvirtualaddressesintophysicaladdresses

(DRAMaddresses)– DatanotinDRAMfetchedfromharddrive

• MemoryProtection– Eachprogramhasownvirtualtophysicalmapping– Twoprogramscanusesamevirtualaddressfordifferentdata– Programsdon’tneedtobeawareothersarerunning– Oneprogram(orvirus)can’tcorruptmemoryusedby

another

VirtualMemory

Chapter8<48>

Cache Virtual MemoryBlock Page

Block Size Page Size

Block Offset Page Offset

Miss Page Fault

Tag Virtual Page Number

Physical memory acts as cache for virtual memory

Cache/VirtualMemoryAnalogues

Chapter8<49>

• Pagesize: amountofmemorytransferredfromharddisktoDRAMatonce

• Addresstranslation: determiningphysicaladdressfromvirtualaddress

• Pagetable: lookuptableusedtotranslatevirtualaddressestophysicaladdresses

VirtualMemoryDefinitions

Chapter8<50>

Most accesses hit in physical memoryBut programs have the large capacity of virtual memory

Virtual&PhysicalAddresses

Chapter8<51>

AddressTranslation

Chapter8<52>

• System:– Virtualmemorysize:2GB=231 bytes– Physicalmemorysize:128MB=227 bytes– Pagesize:4KB=212 bytes

VirtualMemoryExample

Chapter8<53>

• System:– Virtualmemorysize:2GB=231 bytes– Physicalmemorysize:128MB=227 bytes– Pagesize:4KB=212 bytes

• Organization:– Virtualaddress:31 bits– Physicaladdress:27 bits– Pageoffset:12 bits– #Virtualpages=231/212 =219 (VPN=19bits)– #Physicalpages=227/212 =215 (PPN=15bits)

VirtualMemoryExample

Chapter8<54>

• 19-bitvirtualpagenumbers• 15-bitphysicalpagenumbers

VirtualMemoryExample

Chapter8<55>

VirtualMemoryExample

Whatisthephysicaladdressofvirtualaddress0x247C?

Chapter8<56>

VirtualMemoryExampleWhatisthephysicaladdressofvirtualaddress 0x247C?– VPN= 0x2– VPN0x2mapstoPPN 0x7FFF– 12-bitpageoffset:0x47C– Physicaladdress=0x7FFF47C

Chapter8<57>

• Pagetable– Entryforeachvirtualpage– Entryfields:• Validbit: 1ifpageinphysicalmemory• Physicalpagenumber: wherethepageislocated

Howtoperformtranslation?

Chapter8<58>

001 0x00001 0x7FFE00

001 0x0001001 0x7FFF00

V

VirtualAddress 0x00002 47C

Hit

PhysicalPage Number

1219

15 12

VirtualPage Number

Page

Tab

le

PageOffset

PhysicalAddress 0x7FFF 47C

VPN is index into page table

PageTableExample

Chapter8<59>

001 0x00001 0x7FFE00

001 0x0001001 0x7FFF00

VPhysical

Page Number

Page

Tab

le

What is the physical address of virtual address 0x5F20?

PageTableExample1

Chapter8<60>

001 0x00001 0x7FFE00

001 0x0001001 0x7FFF00

V

VirtualAddress 0x00005 F20

Hit

PhysicalPage Number

1219

15 12

VirtualPage Number

Page

Tab

le

PageOffset

PhysicalAddress 0x0001 F20

What is the physical address of virtual address 0x5F20?– VPN = 5– Entry 5 in page table

VPN 5 => physical page 1

– Physical address: 0x1F20

PageTableExample1

Chapter8<61>

001 0x00001 0x7FFE00

001 0x0001001 0x7FFF00

V

VirtualAddress 0x00007 3E0

Hit

PhysicalPage Number

19

15

VirtualPage Number

Page

Tab

le

PageOffset

What is the physical address of virtual address 0x73E0?

PageTableExample2

Chapter8<62>

001 0x00001 0x7FFE00

001 0x0001001 0x7FFF00

V

VirtualAddress 0x00007 3E0

Hit

PhysicalPage Number

19

15

VirtualPage Number

Page

Tab

le

PageOffset

What is the physical address of virtual address 0x73E0?– VPN = 7– Entry 7 is invalid– Virtual page must be

paged into physical memory from disk

PageTableExample2

Chapter8<63>

• Pagetableislarge– usuallylocatedinphysicalmemory

• Load/storerequires2mainmemoryaccesses:– onefortranslation(pagetableread)– onetoaccessdata(aftertranslation)

• Cutsmemoryperformanceinhalf– Unlesswegetclever…

PageTableChallenges

Chapter8<64>

• Smallcacheofmostrecenttranslations• Reduces#ofmemoryaccessesformostloads/storesfrom2 to1

TranslationLookaside Buffer(TLB)

Chapter8<65>

• Pagetableaccesses:hightemporallocality– Largepagesize,soconsecutiveloads/storeslikelytoaccesssamepage

• TLB– Small:accessedin<1cycle– Typically16- 512entries– Fullyassociative– >99%hitratestypical– Reduces#ofmemoryaccessesformostloads/storesfrom2to1

TLB

Chapter8<66>

Hit1

V

=

01

15 15

15

=

Hit1Hit0

Hit

19 19

19

VirtualPage Number

PhysicalPage Number

Entry 1

1 0x7FFFD 0x0000 1 0x00002 0x7FFF

VirtualAddress 0x00002 47C

1219

VirtualPage Number

PageOffset

VVirtual

Page NumberPhysical

Page Number

Entry 0

12PhysicalAddress 0x7FFF 47C

TLB

Example2-EntryTLB

Chapter8<67>

• Multipleprocesses(programs)runatonce• Eachprocesshasitsownpagetable• Eachprocesscanuseentirevirtualaddressspace

• Aprocesscanonlyaccessphysicalpagesmappedinitsownpagetable

MemoryProtection

Chapter8<68>

• Virtualmemoryincreasescapacity• Asubsetofvirtualpagesinphysicalmemory• Pagetablemapsvirtualpagestophysicalpages– addresstranslation

• ATLB speedsupaddresstranslation• Differentpagetablesfordifferentprogramsprovidesmemoryprotection

VirtualMemorySummary

Chapter8<69>

• ProcessoraccessesI/Odevicesjustlikememory (likekeyboards,monitors,printers)

• EachI/Odeviceassignedoneormoreaddress

• Whenthataddressisdetected,dataread/writtentoI/Odeviceinsteadofmemory

• AportionoftheaddressspacededicatedtoI/Odevices

Memory-MappedI/O

Chapter8<70>

• AddressDecoder:– Looksataddresstodeterminewhichdevice/memorycommunicateswiththeprocessor

• I/ORegisters:– HoldvalueswrittentotheI/Odevices

• ReadData Multiplexer:– SelectsbetweenmemoryandI/Odevicesassourceofdatasenttotheprocessor

Memory-MappedI/OHardware

Chapter8<71>

Processor MemoryAddressMemWrite

WriteDataReadData

WE

CLK

TheMemoryInterface

Chapter8<72>

Processor MemoryAddressMemWrite

WriteData

ReadDataI/ODevice 1

I/ODevice 2

CLK

EN

EN

Address Decoder

WE

WEM

RD

sel1:0

WE2

WE1 CLK

000110

CLK

Memory-MappedI/OHardware

Chapter8<73>

• SupposeI/ODevice1isassignedtheaddress0xFFFFFFF4–Writethevalue42toI/ODevice1– ReadvaluefromI/ODevice1andplacein$t3

Memory-MappedI/OCode

Chapter8<74>

• Writethevalue42toI/ODevice1(0xFFFFFFF4)addi $t0, $0, 42

sw $t0, 0xFFF4($0)

Processor MemoryAddressMemWrite

WriteData

ReadDataI/ODevice 1

I/ODevice 2

CLK

EN

EN

Address Decoder

WE

WEM

RD

sel1:0

WE2

WE1 = 1

CLK

000110

CLK

Memory-MappedI/OCode

Chapter8<75>

• ReadthevaluefromI/ODevice1andplacein$t3lw $t3, 0xFFF4($0)

Processor MemoryAddressMemWrite

WriteData

ReadDataI/ODevice 1

I/ODevice 2

CLK

EN

EN

Address Decoder

WE

WEM

RD

sel1:0 = 01

WE2

WE1 CLK

000110

CLK

Memory-MappedI/OCode

Chapter8<76>

• EmbeddedI/OSystems– Toasters,LEDs,etc.

• PCI/OSystems

Input/Output (I/O)Systems

Chapter8<77>

• Examplemicrocontroller:PIC32– microcontroller– 32-bitMIPSprocessor– low-levelperipheralsinclude:• serialports• timers• A/Dconverters

EmbeddedI/OSystems

Chapter8<78>

// C Code

#include <p3xxxx.h>

int main(void) {

int switches;

TRISD = 0xFF00; // RD[7:0] outputs

// RD[11:8] inputs

while (1) {

// read & mask switches, RD[11:8]

switches = (PORTD >> 8) & 0xF;

PORTD = switches; // display on LEDs

}

}

DigitalI/O

Chapter8<79>

• Exampleserialprotocols– SPI: SerialPeripheralInterface– UART: UniversalAsynchronousReceiver/Transmitter

– Also:I2C,USB,Ethernet,etc.

SerialI/O

Chapter8<80>

SPI:SerialPeripheralInterface• MasterinitiatescommunicationtoslavebysendingpulsesonSCK

• MastersendsSDO(SerialDataOut)toslave,msb first• Slavemaysenddata(SDI)tomaster,msb first

Chapter8<81>

UART:UniversalAsynchronousRx/Tx

• Configuration:– startbit(0),7-8databits,paritybit(optional),1+stopbits(1)– datarate:300,1200,2400,9600,…115200baud

• LineidlesHIGH(1)• Commonconfiguration:

– 8databits,noparity,1stopbit,9600baud

Chapter8<82>

// Create specified ms/us of delay using built-in timer#include <P32xxxx.h>

void delaymicros(int micros) {if (micros > 1000) { // avoid timer overflow

delaymicros(1000); delaymicros(micros-1000);

} else if (micros > 6){

TMR1 = 0; // reset timer to 0 T1CONbits.ON = 1; // turn timer onPR1 = (micros-6)*20; // 20 clocks per microsecond

// Function has overhead of ~6 us IFS0bits.T1IF = 0; // clear overflow flagwhile (!IFS0bits.T1IF); // wait until overflow flag set

}}

void delaymillis(int millis) {while (millis--) delaymicros(1000); // repeatedly delay 1 ms

} // until done

Timers

Chapter8<83>

• Neededtointerfacewithoutsideworld• Analoginput: Analog-to-digital(A/D)conversion– Oftenincludedinmicrocontroller– N-bit:convertsanaloginputfromVref--Vref+ to0-2N-1

• Analogoutput:– Digital-to-analog(D/A)conversion

• Typicallyneedexternalchip(e.g.,AD558orLTC1257)• N-bit:convertsdigitalsignalfrom0-2N-1 toVref--Vref+

– Pulse-widthmodulation

AnalogI/O

Chapter8<84>

Pulse-WidthModulation(PWM)

• Averagevalueproportionaltodutycycle

• Addhigh-passfilteronoutputtodeliveraveragevalue

Chapter8<85>

OtherMicrocontrollerPeripherals

• Examples– CharacterLCD– VGAmonitor– Bluetoothwireless– Motors

Chapter8<86>

PersonalComputer(PC)I/OSystems

• USB:UniversalSerialBus– USB1.0releasedin1996– standardizedcables/softwareforperipherals

• PCI/PCIe:PeripheralComponentInterconnect/PCIExpress– developedbyIntel,widespreadaround1994– 32-bitparallelbus– usedforexpansioncards(i.e.,soundcards,videocards,etc.)

• DDR:double-dataratememory

Chapter8<87>

PersonalComputer(PC)I/OSystems

• TCP/IP:TransmissionControlProtocolandInternetProtocol– physicalconnection:EthernetcableorWi-Fi

• SATA:harddriveinterface• Input/Output (sensors,actuators,microcontrollers,etc.)– DataAcquisitionSystems(DAQs)– USBLinks