Organisasi Sistem

Komputer

Materi VIII (Input Output)

Input/Output Problems
 Wide variety of peripherals
 Delivering different amounts of data
 At different speeds
 In different formats
 All slower than CPU and RAM
 Need I/O modules
Input/Output Module
 Interface to CPU and Memory
 Interface to one or more peripherals
 GENERIC MODEL OF I/O DIAGRAM 6.1
External Devices
 Human readable
 Screen, printer, keyboard
 Machine readable
 Monitoring and control
 Communication
 Modem
 Network Interface Card (NIC)
I/O Module Function
 Control & Timing
 CPU Communication
 Device Communication
 Data Buffering
 Error Detection
I/O Steps
 CPU checks I/O module device status
 I/O module returns status
 If ready, CPU requests data transfer
 I/O module gets data from device
 I/O module transfers data to CPU
 Variations for output, DMA, etc.
 I/O Module Diagram
Systems Bus Interface External Device Interface
External Data
Data Data Register Device
Lines Status
Interface
Status/Control Register Control
Logic

Address
Lines Input External Data
Output Device
Data Logic Status
Interface
Lines
Logic Control
I/O Module Decisions
 Hide or reveal device properties to CPU
 Support multiple or single device
 Control device functions or leave for CPU
 Also O/S decisions
 e.g. Unix treats everything it can as a file
Input Output Techniques
 Programmed
 Interrupt driven
 Direct Memory Access (DMA)
Programmed I/O
 CPU has direct control over I/O
 Sensing status
 Read/write commands
 Transferring data
 CPU waits for I/O module to complete
operation
 Wastes CPU time
Programmed I/O - detail
 CPU requests I/O operation
 I/O module performs operation
 I/O module sets status bits
 CPU checks status bits periodically
 I/O module does not inform CPU directly
 I/O module does not interrupt CPU
 CPU may wait or come back later
I/O Commands
 CPU issues address
 Identifies module (& device if >1 per module)
 CPU issues command
 Control - telling module what to do
 e.g. spin up disk

 Test - check status

 e.g. power? Error?

 Read/Write
 Module transfers data via buffer from/to device
 Addressing I/O Devices

 Under programmed I/O data transfer is very like memory

access (CPU viewpoint)
 Each device given unique identifier
 CPU commands contain identifier (address)
I/O Mapping
 Memory mapped I/O
 Devices and memory share an address space
 I/O looks just like memory read/write
 No special commands for I/O
 Large selection of memory access commands available
 Isolated I/O
 Separate address spaces
 Need I/O or memory select lines
 Special commands for I/O
 Limited set
 Interrupt Driven I/O
 Overcomes CPU waiting
 No repeated CPU checking of device
 I/O module interrupts when ready
Interrupt Driven I/O
Basic Operation
 CPU issues read command
 I/O module gets data from peripheral whilst
CPU does other work
 I/O module interrupts CPU
 CPU requests data
 I/O module transfers data
CPU Viewpoint
 Issue read command
 Do other work
 Check for interrupt at end of each instruction cycle
 If interrupted:-
 Save context (registers)
 Process interrupt
 Fetch data & store

 See Operating Systems notes

Design Issues
 How do you identify the module issuing the
interrupt?
 How do you deal with multiple interrupts?
 i.e. an interrupt handler being interrupted
 Identifying Interrupting Module (1)
 Different line for each module
 PC
 Limits number of devices
 Software poll
 CPU asks each module in turn
 Slow
Identifying Interrupting Module (2)
 Daisy Chain or Hardware poll
 Interrupt Acknowledge sent down a chain
 Module responsible places vector on bus
 CPU uses vector to identify handler routine
 Bus Master
 Module must claim the bus before it can raise
interrupt
 e.g. PCI & SCSI
Multiple Interrupts
 Each interrupt line has a priority
 Higher priority lines can interrupt lower
priority lines
 If bus mastering only current master can
interrupt
 Example - PC Bus
 80x86 has one interrupt line
 8086 based systems use one 8259A interrupt
controller
 8259A has 8 interrupt lines
Sequence of Events
 8259A accepts interrupts
 8259A determines priority
 8259A signals 8086 (raises INTR line)
 CPU Acknowledges
 8259A puts correct vector on data bus
 CPU processes interrupt
PC Interrupt Layout
8259A 8086
IRQ0
IRQ1
IRQ2
IRQ3 INTR
IRQ4
IRQ5
IRQ6
IRQ7
ISA Bus Interrupt System
 ISA bus chains two 8259As together
 Link is via interrupt 2
 Gives 15 lines
 16 lines less one for link
 IRQ 9 is used to re-route anything trying to
use IRQ 2
 Backwards compatibility
 Incorporated in chip set
 ISA Interrupt Layout
(IRQ 2) 8259A 8259A 80x86

IRQ0 (8) IRQ0

IRQ1 (9) IRQ1
IRQ2 (10) IRQ2
IRQ3 (11) IRQ3 INTR
IRQ4 (12) IRQ4
IRQ5 (13) IRQ5
IRQ6 (14) IRQ6
IRQ7 (15) IRQ7
Direct Memory Access
 Interrupt driven and programmed I/O require
active CPU intervention
 Transfer rate is limited
 CPU is tied up
 DMA is the answer
DMA Function
 Additional Module (hardware) on bus
 DMA controller takes over from CPU for I/O
DMA Operation
 CPU tells DMA controller:-
 Read/Write
 Device address
 Starting address of memory block for data
 Amount of data to be transferred
 CPU carries on with other work
 DMA controller deals with transfer
 DMA controller sends interrupt when finished
DMA Transfer
Cycle Stealing
 DMA controller takes over bus for a cycle
 Transfer of one word of data
 Not an interrupt
 CPU does not switch context
 CPU suspended just before it accesses bus
 i.e. before an operand or data fetch or a data write
 Slows down CPU but not as much as CPU doing
transfer
Aside
 What effect does caching memory have on
DMA?
 Hint: how much are the system buses
available?
 DMA Configurations (1)

DMA I/O I/O Main

CPU
Controller Device Device Memory

 Single Bus, Detached DMA controller

 Each transfer uses bus twice
 I/O to DMA then DMA to memory
 CPU is suspended twice
 DMA Configurations (2)
DMA DMA Main
CPU
Controller Controller Memory
I/O
I/O I/O Device
Device Device
 Single Bus, Integrated DMA controller
 Controller may support >1 device
 Each transfer uses bus once
 DMA to memory
 CPU is suspended once
 DMA Configurations (3)
CPU DMA Main
Controller Memory

I/O I/O I/O I/O

Device Device Device Device
 Separate I/O Bus
 Bus supports all DMA enabled devices
 Each transfer uses bus once
 DMA to memory
 CPU is suspended once
I/O Channels
 I/O devices getting more sophisticated
 e.g. 3D graphics cards
 CPU instructs I/O controller to do transfer
 I/O controller does entire transfer
 Improves speed
 Takes load off CPU
 Dedicated processor is faster
Interfacing
 Connecting devices together
 Bit of wire?
 Dedicated processor/memory/buses?
 E.g. SCSI, FireWire
Small Computer Systems Interface (SCSI)
 Parallel interface
 8, 16, 32 bit data lines
 Daisy chained
 Devices are independent
 Devices can communicate with each other as
well as host
SCSI - 1
 Early 1980s
 8 bit
 5MHz
 Data rate 5MBytes.s-1
 Seven devices
 Eight including host interface
SCSI - 2
 1991
 16 and 32 bit
 10MHz
 Data rate 20 or 40 Mbytes.s-1

 (Check out Ultra/Wide SCSI)

 SCSI Signaling (1)
 Between initiator and target
 Usually host & device
 Bus free? (c.f. Ethernet)
 Arbitration - take control of bus (c.f. PCI)
 Select target
 Reselection
 Allows reconnection after suspension
 e.g. if request takes time to execute, bus can be released
SCSI Signaling (2)
 Command - target requesting from initiator
 Data request
 Status request
 Message request (both ways)
 SCSI Bus Phases

Reset

Command,
Bus Data,
free Arbitration (Re)Selection
Status,
Message
SCSI Timing Diagram
Configuring SCSI
 Bus must be terminated at each end
 Usually one end is host adapter
 Plug in terminator or switch(es)
 SCSI Id must be set
 Jumpers or switches
 Unique on chain
 0 (zero) for boot device
 Higher number is higher priority in arbitration
IEEE 1394 FireWire
 High performance serial bus
 Fast
 Low cost
 Easy to implement
 Also being used in digital cameras, VCRs
and TV
FireWire Configuration
 Daisy chain
 Up to 63 devices on single port
 Really 64 of which one is the interface itself
 Up to 1022 buses can be connected with bridges
 Automatic configuration
 No bus terminators
 May be tree structure
FireWire 3 Layer Stack
 Physical
 Transmission medium, electrical and signaling
characteristics
 Link
 Transmission of data in packets
 Transaction
 Request-response protocol
FireWire - Physical Layer
 Data rates from 25 to 400Mbps
 Two forms of arbitration
 Based on tree structure
 Root acts as arbiter
 First come first served
 Natural priority controls simultaneous requests
 i.e. who is nearest to root

 Fair arbitration
 Urgent arbitration
FireWire - Link Layer
 Two transmission types
 Asynchronous
 Variable amount of data and several bytes of transaction data
transferred as a packet
 To explicit address
 Acknowledgement returned
 Isochronous
 Variable amount of data in sequence of fixed size packets at
regular intervals
 Simplified addressing
 No acknowledgement