Computer Systems Technology by hY7s1p0X


									                                               Microprocessor Systems Revision Tutorial 2011 1
64ET2520 – Microprocessor Systems - Revision Tutorials 2010-11 (rev a)

Revision Tutorials

The Examination for Micro Systems lasts 2 hours

There are 7 topics

1. Introduction to Modern Computer Systems using the PC Product
2. PC Functional Overview
3. PC Processor Technology
4. PC Memory Technology
5. PC Peripheral Technology
6. The PC Bus Technology
7. PC Disk and Backup Technology

You are required to answer 5 questions out of 7

However for the Exam in May 2011 there are TWO COMPULSORY QUESTIONS
based on Topics 1 to 2 which count for 35/80 marks

You are required to answer THREE more questions (15 each) which count for
45/80 marks

Q3 is based on Topic 3, Q4 is based on Topic 4, Q5 is based on Topic 5, Q6 is
based on Topic 6 and Q7 is based on Topic 7


Suggestions on how to revise

       Select a topic
       Read the presentation handout for that topic
       look at the tutorial for the topic
       Try the tutorial questions yourself or in a small revision group
       Look at the answers

!!!      don't just look at the tutorials and answers
!!!      don’t revise part of a topic

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                               Microprocessor Systems Revision Tutorial 2011 2
1. Introduction to Modern Computer Systems using the PC Product

A1 The PC Product has been produced since 1981:
           Briefly review the major stages in the development of the product
           State three features of the original design which has contributed to the
            success of the product.
           Justify why it has become the de facto standard for home and office users
           Compare the processing power of a modern PC with the original PC –
            why does a modern PC need such a powerful processor ?

A2 Why has the PC Architecture (often called the ISA Industry Standard
   Architecture) been identified as important by so many manufacturers ?

      How does compliance to the specifications from Intel and Microsoft maintain
      standards for the PC architecture ?

      Describe alternative computer products to the PC products:
        for general purpose computing
        for games and multimedia

A3 Legacy has been an important feature of the PC product.
   Explain what this means from the users' point of view.
   State good and bad features of compatibility

A4 Review the features of a professional office PC – include the following
    processor
    chipset/motherboard logic
    memory – ROM and RAM
    disk support
    communications
    display subsystem
    adapter bus technology
    peripheral technology
    security (e.g. Active Management, Trusted platform)

A5 What are the trends in Consumer/ Business Computer Products (e.g.
   desktops, laptops, netbooks, tablets (iPad), the Cloud) ?

A6 What is the future for non-Windows based operating systems and software?
   .. on PCs/ Tablets/ Smartphones

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                               Microprocessor Systems Revision Tutorial 2011 3
2. PC Functional Overview – Chipsets (e.g. Intel Series 5 Q57)

              List the important functional blocks in the current PC architecture
              Why are most of these features now implemented as part of the PC
               motherboard and integrated into the chipset?
              How are peripherals accessed in the PC Architecture ?
              State 5 key PC features used in multimedia applications
              Explain the role of fast microprocessors in multimedia applications (with
               particular reference to processing media files (video: MPEG and audio:
               MP3) and data encryption/ decryption

B2 Chip Sets are as important as processors in the modern PC

           a. Sketch the block diagram of a modern PC based on the chip set
              components (i.e. as in the Intel Q57 Express Chipset
           b. State the data rates for each of the major interfaces (DDR3 Memory
              Bus, PCIe 2.0 x16 bus etc) – Hence confirm that modern chipsets
              provide the core features for modern PCs
           c. List the integrated features provided by modern chip sets (e.g.
              graphics, sound, networking, security) – justify why these products are
              aimed at the budget, business and mobile markets
           d. State the market for dedicated audio, graphics, communications
              adapter cards etc.
           e. What are the advantages of purchasing PCs incorporating the latest
              chipsets ?

B3 List 5 areas that may be developed in the PC product in the next few years.

B4 What is the future for the PC ? - will other technologies replace PCs soon ?

      An application was developed for a PC in 1990 - will this application still
      execute on a modern PC executing Windows XP/ Windows 7 ?

B5 Discuss the modifications in the next generation of chipsets ( six series)
   based on the latest Intel Second Generation Core i3, i5 and i7 processors

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                               Microprocessor Systems Revision Tutorial 2011 4
3. PC Processor Technology (e.g. Core i3/ i5/ i7 – cpu-z - Sandy Bridge)
C1 Explain what is meant by the x86-family of microprocessors
      State the characteristics of typical x86 processors
      List the principal processors in the x86-family of processors
      Explain in simple terms how the original 16-bit x86 architecture has
         been upgraded to 32-bit (80386) and subsequently 64-bit
C2 Outline the basic operation of a microprocessor executing typical C# code
   (e.g. integer arithmetic, real arithmetic, calling a function, outputting to a port)
C3 Outline the features of modern x86 processors which have made these chips
   the de facto standards for modern PCs.
C4 Why have Intel put so much effort into publicizing the Core i3/ i5/ i7 series of
   microprocessors ?
C5 Compatibility has always been an important feature of the x86 family of
   Explain clearly what compatibility means.
   Will future products maintain this compatibility ?
C6 The performance of a PC is closely related to the speed at which memory is
   accessed - Explain how the PC product uses cache memory techniques to
   improve performance.
C7 The cpu-z utility was used in the lab to inspect PC features such as the
    Code name
    Socket technology
    Core Voltage and Operating Frequency
    Number of Cores and HT Hyper-Threading technology - threads
    Turbo mode
    Technology 45nm 32 nm processes
    Intel 64 EM64T 64-bit technology
    SSE (SSE, SSE2, SSE3, SSE4) Streaming Single Instruction Multiple Data
    XD Execute Disable Bit
    Multiplier

      Describe these terms and technologies and how they will affect the
      performance of a PC

C9 Intel has delivered the latest version of the Core i3/i5/i7 processors called
   Sandy Bridge – review the important features of Sandy Bridge

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                               Microprocessor Systems Revision Tutorial 2011 5
4. PC Memory Technology (Core i3/ i5/ i7 processors with integrated RAM
   controllers – UEFI replacing BIOS)

D1     List the memory hierarchy of a typical PC (fastest to slowest – Registers to
      Page File) - hence compare the size, speed and role of each component
      (e.g. using ticks rather than actual values)

D2    List typical memory devices used for the ROM and RAM devices in a PC.
      Explain the difference in the technologies used in static and dynamic RAM.
      State the application areas for dynamic RAM, static RAM and ROM in a PC.
      Explain the basic operation of DDR3 and dual channel DDR3 RAM

      Task Manager is a useful tool in managing a PC:
      Explain what the following terms mean:
            Handles, Threads, Processes
            Physical Memory, Commit Charge/ Page File usage
            Kernel memory, CPU usage

      How can Task Manager help a user to understand the operation of their PC?

D4 Why is memory such an important element in determining the performance of
   a PC?

      Will adding more RAM improve the performance of a PC ?
      Will a dual channel memory configuration improve the overall system
      performance by a factor of two compared with a single channel ?
      Modern processors incorporate memory controllers – how does this improve
      performance ?

      With reference to the Q57 based configuration from Section 2 (Chipsets) -
      calculate the bandwidth of a dual channel DDR3 RAM rated at 1333 MHz

D5 Cache memory is usually quite small.
   Explain why a small amount of memory can have such an effect on overall
   Comment on the benefits of L1, L2 and L3 cache

D6 The latest PCs feature UEFI Unified Extensible Firmware Interface rather
   than classic BIOS.
   Explain what UEFI is and why it offers superior features to the classic BIOS

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                               Microprocessor Systems Revision Tutorial 2011 6
5. PC Peripheral Technology (USB1.1, USB2.0, USB3.0, Firewire, Light Peak)

E1 State the peripheral devices associated with a typical PC and list the features
   of each interface technology
      The printer on a PC has traditionally used the LPT parallel port.. explain why
      this was important for a product aimed at non-specialised user ( the alternative
      was a serial link)
      Explain how the keyboard works on a modern PC
      Explain how the COM ports function on a PC
      How does software access peripherals ?
E2 Explain the move from traditional peripherals e.g. mouse, keyboard, PC
   speaker etc to USB based peripherals thus creating a legacy free PC

      Describe the support for USB by the Q57 chipset
      The Intel Q57 chipset doesn’t support legacy peripherals but even modern
      motherboards are available with legacy support – explain how this is
      implemented and justify this move
E3 Compare polling, interrupts and DMA as techniques to allow access to
   peripheral devices
      Explain how interrupts allow the various peripherals and other devices to
      share the CPU
      Describe the sequence following a typical interrupt event e.g. a real-time
      clock tick on IRQ0
      Explain how the APIC interrupt system facilitates more than 24 interrupts
      Describe how the PCI-bus interrupt system and the legacy ISA PC
      interrupt system (e.g. IRQ0 the system timer) co-exist
      How are interrupts allocated in a typical PC ?

E4      Compare the features of USB and Firewire
        Explain the term hot swapping
        Review the basic technology of USB3.0
        Why is USB3.0 not supported by modern chipsets
        The latest PCs do have USB3.0 ports – how is this facilitated ?

E5      Intel have proposed a general purpose high performance interface
        technology called Light Peak – the original product used optical fibre but
        now a copper version at 10 Gbps is available – what is the future of Light
        Peak ?
29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                               Microprocessor Systems Revision Tutorial 2011 7

6. The PC Bus Technology ( including PCIe x1, x4, x8, x16, x32; PC/104)

F1    List the stages in the development of the extension bus for the PC product.
      Complete a table showing the characteristics of each technology.

F2    Explain why the ISA-bus dominated the PC product until the 90s.

      Many industrial microprocessor systems use a version of the ISA-bus – called
      PC/104 – Explain why these have not upgraded to the industrial versions of
      the PCI-bus.

      Estimate the data rates required to interface to:
        sound card
        graphics card
        a 10/100 Mbps Ethernet card

        in each case, state a suitable bus ( PCI, PCIe, USB1.1, USB2.0)

F3    Modern PCs use the PCIe-bus as well as the PCI-bus.

      Explain the technology of PCIe and justify why it may be implemented with
      various lane configurations (e.g. 1 to 32-lanes or x1, x4, x8, x16 , x32)

      Explain why the PCI-bus was not able to cope with modern video

      Current PCIe graphics cards are listed as x16 - explain this designation.

      In the Intel Q57 chip set the PCIe x16 connection is connected directly to the
      processor. Why isn't it connected to the Q57 chipset?

      Modern PCs use PCIe slots which are designated x1 – what plugs into this

      Chipsets exist which have no PCI support only PCIe. Is the PCI bus obsolete
      and replaced by PCIe x1 for soundcards etc ?

F4    Windows XP/ Vista are described as Plug-and-Play compatible.
      What does the term Plug-and-Play mean and why is it desirable for modern
      cards e.g. video, disk, sound
      Which buses can use Plug-and-Play technology ?

F5    Many users demand faster graphics – explain why the current generation of
      processor include integrated graphics processors as well as integrated
      memory controllers

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                               Microprocessor Systems Revision Tutorial 2011 8

7. Disks and Backup Technology (SATA, SAS, SSDs)

G1 List the major disk and backup technologies used in PCs
   Compare features such as speed, performance, size, cost per GByte

G2 Modern PC disk drives use a technology called Serial ATA – review the major
   features of SATA drives (size (2.5 inch, 3.5 inch), interface and cables, cost,
   RPM, speed (e.g. access time), capacity in GBytes, reliability, buffer size,
   external option eSATA etc)

      SATA drives can feature so called intelligent seeking – describe this
      technology and the potential benefits to a PC user

G3 Draw a block diagram of an SAS Serial Attached SCSI disk subsystem

      Explain why SAS drives are more likely to be used for a server rather than a
      desk top PC

      Compare SAS with SATA

G4 Compare the normal backup technologies used in PCs in terms of cost and
   ease of use.. select the most suitable for home use, small office, large

G5 Review the facilities of a modern file system such as NTFS

G7 Solid State Drives are now available – review these products and discuss the
   areas of application – will conventional hard disk drives become obsolete ?

G8 A simple way of backing up the data on a PC is to use an external disk drive
– compare the technologies available to connect such a drive (e.g. USB2.0, eSATA
(external SATA), Firewire).

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                               Microprocessor Systems Revision Tutorial 2011 9
Outline Answers for Computer Systems Technology

A Introduction to the Introduction to the PC Product

A.1      See Topic One PPT Tutorial Slides 2-5

   Major stages in PC development :
       o IBM-PC upgraded to the IBM-AT
    then a split when IBM marketed the PS/2
       o PS/2 + MCA-bus + OS/2 operating system
    HOWEVER most other PC manufacturers used an update of the IBM-AT called
    the ISA architecture (slide3)
       o ISA Architecture + ISA-bus + MSDOS
    A modern multimedia PC uses the upgraded Industry Standard Architecture
       o Modern ISA Arch + PCI-bus + PCIe (PCI Express) + Windows
      the split was caused by IBM patenting the PS/2 product which required
      companies designing with PS/2 technology to pay royalties to IBM

  The success of PC has been due to:
 - original underwriting of product by IBM which encouraged hardware and
    software developers to design for the PC
 - simplicity of the original PC-bus and the ease with which the video/ disk/
  communications elements of a standard PC can be enhanced for specific
  markets e.g. office, home, engineering, network server
 - stable architecture .. software written in 1981 will still run on the latest PC
 - the performance/ cost ratio of the PC has always been good

   So the PC benefits from economies of scale in a huge global market. The
    hardware is produced in very large volume and is very affordable. There is a vast
    amount of good software (operating systems such as Windows Vista/ 7,
    applications from Word to video editing). Moving to another product e.g. Apple Mac
    is difficult to justify on economic grounds.

   A modern PC is very much more powerful than the original PC however a
    modern PC is very complex and may be running 50 programs in a multitasking
    environment – the original PC was simple and could only run one program at a
    time – so the extra processing power is used in managing the facilities offered by
    Windows e.g. rendering complex images/ running background tasks

The ISA architecture provides a simple, but relatively high performance, hardware
platform which is now well defined – this ensures interoperability between hardware
(PCs and peripherals), BIOS/ UEFI, OS and Applications

Microsoft hosts Hardware Eng Conferences and Intel hosts similar events

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 10
   Thus the PC product is not managed through a standards authority. Developers
    of hardware, software and peripherals are actively involved in the standards for
    the PC through Microsoft, AMD and Intel

   Thus the whole of the PC product is compliant with these standards
       PC motherboard
       PCI-bus peripherals ( configured via plug-and-play )
       PCIe graphics products
       USB peripherals
       SATA disk drives
       OS e.g. Windows 7
       Shrink-Wrap applications ( load-and-go with little configuration)

    No other computer range has the equivalent selection of Hardware peripherals
    or Software applications or such a good performance/cost ratio

   Alternative products to the PC are
    - the Apple MAC range which is popular with users who like the sleak design
       features of Apple products and are willing to pay more than for a PC with
       similar performance (uses x86 processors)
    - Network Workstations ( Powerful Server + graphics workstation )
       applications run on server - data is stored on server - workstation is used to
       display results - disaster if network is down since no local programs etc.
       - see Question A5 concerning the Cloud
       - also model for many Apps on smartphones

   Alternative products to the PC for surfing web, games etc. are
    - Digital TV channels using set-top boxes to decode channels but also to
       supply email and Internet browsing facilities, e-commerce etc.
    - games consoles with Internet facilities similar to set-top box (Xbox 360 etc)

However users have become familiar with the facilities and frustrations of PCs ..
2010 PC Sales > 300 million worldwide and growing - there seems no doubt that
PCs (desktops, laptops) will survive for another generation – the Tablet market
appears dominated by ARM and OSs such as iOS and Android

Legacy/ Compatibility means 'old software ' runs without modification on new PC
hardware. This is surprisingly important. To use a new PC you can start with all
the old applications and you don't have to learn lots of new ideas. There will be
upgraded features, but since the hardware and the OS will not have changed
radically - most people can start to use a new PC within a few minutes. This has
important retraining implications.
Bad features – old ideas have to be retained, you can’t start again, however
designers may get bored
Good features – as above – ease of use, no retraining, use old software

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 11
I had expected Sandy Bridge products to be delivered by date of this tutorial
(February 2011 ) but this has been delayed by a fault in the Series 6 chipset
(Cougar Point) – So the chipset and processors here are from 2010 !

                                 Professional PC- Dell
                               500 pound professional PC
Intel® Core™ i5-650 (3.2 GHz, 2 Cores, 4 Threads, turbo boost up to 3.46 GHz, 4
MB smart cache), 32 nm, $176, 382 million transistors, Nahalem Clarkdale
released Q1 2010

Turbo Boost Yes
Hyper-Threading Yes
Virtualization Tech Yes
Intel 64 Yes
Speedstep Yes
Ex Disable Bit Yes
Trusted Execution Yes
Active Management Yes
 Q57 chipset (one chip)
-See Q57 chipset diagram
DMI technology – interface processor with chipset
FDI technology – interface to standard graphics products e.g. HDMI
Two DIMM slots providing up to 8GB Dual Channel DDR3 @ 1333MHz
4096MB 1333MHz Dual Channel DDR3 SDRAM [2x2048]
Integrated graphics
Intel® HD Graphics – see Q57 chipset
PCIe option
6 SATA ports
mobile - 320GB (5,400rpm) Hard Drive SATA 3Gb/s
DVD +/- RW Drive (read/write CD & DVD) with DVD Burn software

Intel High Definition Audio 2
USB2.0 – 14 ports – 2 controllers
Integrated -10/100/1000 (Gigabit) Ethernet, integrated WiFi module

 More powerful multimedia environment e.g. codecs using SSE instructions
   (Streaming Single Instruction Multiple Data Extensions) and AES
   encryption/decryption related instructions
 Complex graphics – using say dual graphics cards Core i7-9xx plus x58
   Platform Controller Hub
 Dual/ Quad/ Eight core processors which run slower and hence cooler but still
   perform extremely well
29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 12
   Incorporation of Windows 7 – better security and user interface (Trusted
    Platform and vPro etc)
   64-bit processing (Intel 64)
   Ultra Mobile PCs often called netbooks using Atom processor
   Tablets/ iPad etc problem with battery life – ARM RISC processors (as used in
    mobile phones) may be better than x86
Windows is a complex product which has been updated to meet the requirements
of games users, office users, multimedia/ network users
Dialects of UNIX such as Linux are much simpler than Windows and say for
servers much cheaper to purchase and easier to maintain and promise higher
performance and reliability.
Servers don't require 'office products' or 'games' support.

Outline Answers for Computer Systems Technology
2. PC Functional Overview – see also Topic Two Tutorial
The blocks which make up the current PC architecture are :
 an x86 family processor.. ( e.g. a dual/quad core processor) with on-chip
   Level 1 cache(s) and Level 2 cache (and Level 3 cache in modern Core-i3/
   i5/ i7 processors)
 main RAM ..e.g. DDR3-RAM (e.g. greater than 1 GByte for Windows based
 BIOS ROM.. contains code to test the hardware and initialise the Windows OS
   and may be used to access motherboard peripherals
 Priority Interrupt Controller.. Up to 24 interrupt sources may be identified
   (highest priority = Real Time Clock , next highest priority = keyboard etc. )
   interrupts sources are shared in modern PCs as the number of devices
   requiring interrupt support has increased – setup by PnP BIOS
 DMAC.. used for rapid data transfers e.g. Sound Cards, Hard disks
 On board Peripherals.. mouse port, serial ports COM1:, COM2:, parallel port
   LPT1:, speaker, EIDE disk interface e.g. ATA and SATA
 Timers/Counters Used for real-time-clock, speaker
 PCI interface logic used to provide an extension to the motherboard:
   o PCI-bus (e.g. large disks SCSI , communications .. network card or modem )
   o PCIe (modern graphics cards) – usually one slot x16 – 16 PCIe lanes and
       perhaps 1 slot x1
 Serial bus interface logic
   o Firewire/ 1394 (high performance multimedia peripherals) 400/ 800 Mbps
   o USB ( multimedia peripherals) USB1.1 12 Mbps USB2.0 480 Mbps
       USB3.0? Light Peak ???
   o SATA – modern disk drives (1.5 Gbits/sec; 3 Gbits/sec latest 6 Gbits/sec)

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 13
   These features are implemented as part of motherboard/ chip set because:
       for highest performance these should be connected on a single board (the
         motherboard) to keep distances short
       for low cost manufacturing DDR2/ DDR3 SDRAM interface, PIC, DMAC,
         classic peripherals, Timers, Bus interfaces etc are integrated into the chip
   Peripherals are accessed through the IO map – in the lab you identified PS/2
    keyboard as using IO address 0060 and 0064 hexadecimal – see lab mp210L
   Multimedia applications require:
    ▫ SSE instructions which accelerate image and sound processing
    ▫ fast processor plus PCIe allows high speed video adapter interface
    ▫ chip set allows main memory to be shared with video memory for activities
      requiring large memory (3D textures)
    ▫ cache memory facilitating high speed processing
Multimedia file sizes are very large and cannot be transmitted across the Internet or
stored on disk without compression/ decompression techniques
Using software codecs is cheaper than using dedicated hardware but needs a fast

(a) For Core i5 block diagram see the Core i5 processor connected to the Q57
    Chipset (single chip)

(b) Data Rates: - in Q57 chipset:
 The microprocessor connects to the Q57 hub using the Direct Media Interface
   bus at 2 MByte/sec)
 The dual channel DDR3 memory connects at say 1333 MHz – 10.7 GBytes/sec
 the PCIe 2.0 x16 connects at 10 Gbps times 16 say 20 GBytes/sec reduced by
   the 10:8 protocol so usually quoted as16 GBytes/sec

Data Rates - the Q57 chipset provides the following integrated features:
   14 USB serial channels (USB2.0 at 480 Mbps)
   6 SATA serial disk channels (say SATA 300 - each channel 3 Gbps)
   LAN Ethernet serial interface 10/100/ 1000 Mbps
   HDA High Definition Audio – requires external sound chip
   Hardware based security – Active Management and Trusted Platform

(c) the Core i5 processor provides an x86 processor plus a graphics processing
Integrated graphics - integrated graphics (e.g. Intel graphics) uses unified memory
(the graphics shares main memory rather than having a dedicated memory) – the
frame rate for games with integrated graphics will normally be poor but adequate
for all other applications including Windows 7

the Q57 chipset can provide integrated features such as sound and networking
29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 14

Integrated audio – integrated audio will be good enough for home cinema etc but
a top of the range sound card will outperform an integrated HDA sound sub-system
Integrated Ethernet – a network card normally has additional features to allow fine
tuning of network performance, but this is not required on PCs (but useful for
(d) A PC designed for the games, multimedia editing, specialist CAD
applications will benefit from a dedicated graphics card and perhaps a dedicated
sound card

(e) Advantages of purchasing a PC with the latest chipset
Supports latest processors e.g. Core i7 - which has an integrated memory controller
facilitating direct and hence fast interfaces to DDR3 RAM modules, Core i7
replaces FSB with a new interface called QuickPath and the Core i7 chipset has
multicard PCIe x16 support etc
Areas of development
 attempt to move Windows XP users to Windows 7
 Voice operated applications.. improved voice recognition software
 Media Center products - PC integrated into Entertainment unit.. TV programs
   stored on hard disk – music channels processed at same time
 Netbooks - Ultra Mobile PCs small form factor (< 1 Kg) with WiFi, WiMax ? in
    future, SSDs etc
 iPad products with improved multimedia facilities

Future of PCs
Although sales of desktop are slowing, overall there is growth in the overall PC
market .. laptop and other mobile PCs continue to incorporate more and more
desktop features/performance – overall PC sales are growing (2010 sales
estimated 300 million) – no obvious competition at present
1990 Application – will it run on modern PC ? - In general YES - although a
program for a modern PC won't run on the 1990 PC. The old code will run on the
latest PC hardware since the core architecture is still based on the original ISA
 Intel have updated their processor range with a range of products named Core i7-
9xx series and Core i7-8xx / i5 / i3

     Core i7-9xx (Nehalem) has an integrated memory controller in the processor
     chip facilitating fast interfaces to DDR3 RAM modules, the FSB is replaced by a
     new interface called QuickPath, and the Northbridge chip has multicard PCIe
     x16 support – the current Core i7-9xx processors are quad-core with
     HyperThreading * 8 threads) and have a large L3 cache shared between the
29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 15

      Core i7-8xx / i5 / i3 (Nehalem) have features such as dual core, integrated
      graphics in processor module and a single chipset component to interface to
      USB/ SATA/ Ethernet etc and selected versions operate using a management
      environment called vPro

3. The x86 Family of Microprocessors ( Core i3 /i5 /i7 Sandy Bridge))
The x86 family of processors all have the same basic features as the 8088
- internal architecture
- register set
- instruction set
- these features allow all x86 processors to execute MSDOS
Later members of the family (IA32) have enhanced features and from 1985 and
the 80386, all members have supported the Windows Operating System as well as
8088 – 8086 – 80188 – 80186 – 80286 – 80386 – 80486 – Pentium - Pentium MMX
- Pentium Pro - Pentium II - Pentium III - Pentium 4 – Pentium D - Core 2 Duo –
Core 2 Quad – Atom – Core i7-9xx – Core i3/ 5/ 7-8xx – Second Generation i3/
i5/ i7
Upgrade of x86 16-bit to 32-bit to 64-bit
The original x86 architecture was 16-bit – so the AX register was 16-bit wide
However addresses were processed by the address-adder to give 20-bit addresses
(1 MByte)
The 80386 microprocessor upgraded the registers to 32-bit by adding an extra 16-
bits to the 16-bit registers – a 32-bit register can be used to access 4 GBytes of
memory – extra 32-bit instructions were added
Subsequently processors such as the Core 2 products were upgraded to 64-bit by
adding an additional 32-bits to the 32-bit registers
This actually allows programs to access up to 2 ^64 which is a huge address range
– extra 64-bit instructions were added
Current 64-bit processors only use 2^48 = 256 TBytes !!!
Microprocessors must basically
     read instructions from the program area of memory, queue and decode the
       instructions then execute the instructions
Instructions will involve
     transferring data from one memory area/ register to another memory area/ register
     transferring data to/from an IO area
     interacting with the stack - CALL and RETURN also Interrupts
     perform arithmetic and logic (ADD, MUL, Mask)
     test conditions and branch if condition is true etc.
     machine control operations (HALT etc.)

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 16
a = b + c ; Assume the variables are 32-bits and stored in RAM. The data must be
copied from RAM locations b and c to internal x86 scratchpad registers (e.g. EAX,
EBX etc). The arithmetic operation occurs using the ALU Arithmetic and Logic Unit
and the result is stored in one of the scratchpad registers and is then copied to the
memory location a in RAM. The flag register will store the status of the arithmetic
(zero etc)
The x86 products can all execute MS-DOS when operating in the REAL-mode
However these processors are also capable of executing in the Windows or
The protected mode supports in hardware the features required to implement a
complex multi-tasking Operating System (Windows NT/ 2000/ XP/ Vista/ Windows

Features ( e.g. Core 2 Duo, Core i3/ i5/ i7 )

▫  high clock speeds e.g. 3 GHz and now dual/ quad core (even six-core plus HT)
▫  complex internal queuing and decoding of instructions
▫  super scalar architecture ( more than one instruction at a time )
▫  out of order execution of instructions
▫  extra hidden registers ( aliases ) .. several instruction can access the same
   register at the same time ( the processor sorts out the resulting problems)
▫ integrated Level 1 cache memory, integrated Level 2 cache and on-chip L3
▫ High speed interface to motherboard (Core 2 Duo has a FSB speed 1333 MHz)
▫ SSE 1,2,3,4 Streaming SIMD (Single Instruction Multiple Data) Extensions -
   ideal for 3D graphics – 128-bit arithmetic
- Intel require successful sales of Core i range to fund new product development
- Intel require customers to insist on Intel inside or similar logo on their PC
- much of this is an advertising issue rather than a technical one !
- Intel have changed their logo and processor logos to become more memorable

Compatibility means old code will run on a new computer
It has the disadvantages that you can't start again with say a RISC instruction set
rather than the complex x86 instruction set
There is no hint that Intel will change from x86 processors

 There is no point in having a fast processor which operates with slow memory
 ideally the code memory and data memory operate at the same speed as the
 such memory must be on the same chip as the microprocessor and as large as

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 17
    this high speed memory is usually divided into code and data Level 1 cache
     memories – usually called L1 cache
    a logic unit called a cache controller is used to move blocks of code and data
     into the Level 1 cache memories.. most program are localised (using code in
     next location and a small block of data) hence cache technology works well
    a further improvement in cache memory operation is to have a Level 2 cache
     between the Level 1 cache and main memory ( RAM )
 Latest versions of the Core i have an on-chip Level 2 cache (2/4 MB) using
  Advanced Transfer Cache technology (level 3 cache as well)

    Code Name – usually released before processor is launched
    Package/ Socket Technology – new processors normally have new socket
     configurations and will only fit into the latest motherboards
    Core Voltage and Frequency – for reliable operation the processor should be
     used as specified by the manufacturer and set up by the BIOS/ EFI –
     enthusiasts can change these to get better performance for games but the
     processor may crash due to overheating etc
    Number of cores – even basic processors can have four cores with clever
     frequency concepts –see Turbo mode
    HT Hyper-Threading – extra hardware fools OS and application into seeing two
     processors this can give a performance boost of 30% but only needs 5% extra
     chip area – there is only one physical processor
    Turbo Mode – if only one core is active, then the processor clock speed can be
     raised to give extra performance – if this happened with four cores active the
     chip would overheat

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 18
   45nm 32nm – each generation of processor can benefit from using a finer
    silicon geometry – this makes the transistors smaller and hence cooler – so you
    can have more transistors for cache etc
   Intel 64 EM64T – modern applications such as large databases or complex
    CAD can benefit from more than 4 GBytes of RAM which was the traditional 32-
    bit Pentium limit – by allowing 64-bit addressing the direct address range is now
    massive – needs 64-bit BIOS, 64-bit OS, 64-bit Applications etc
   SSE –Streaming Single Instruction Multiple Data Extensions allow complex data
    sets to be processed at say 8 points at a time – ideal for rendering images in
    games etc
   XD Execute Disable Bit - many users will always fail to update virus signatures
    leaving their PCs vulnerable – the Execute Disable Bit feature, marks code as
    read-only, so viruses cannot alter code
   Speedstep – processors run at reduced power consumption which is ideal for
    notebooks using batteries – this works by lowering the frequency, reducing the
    voltage when idle – the same ideas are being used for other Core i class chips
Sandy Bridge processors are Second Gen i3/ i5/ i7:
    Quad core, HT, Turbo Boost features
    On die memory controller supporting at least two channels of DDR3 RAM
    On die GPU features
    Dedicated hardware to support video conversion
    A ring bus which links cores/ GPU and L3 cache
    Graphics good enough for all general purpose PC applications

Outline Answers to Topic 4
PC Memory Technology

D1 Memory Hierarchy
   Page File - on the hard disk – increasingly used as applications cannot fit in
    physical memory
   Main System Memory SDRAM (Synchronous Dynamic RAM) currently in
    three versions DDR SDRAM, DDR2 SDRAM and DDR3 SDRAM
    DDR RAM = Double Data Rate RAM two data activities / clock
    DDR2 RAM = Second generation DDR RAM allowing higher frequencies
    DDR2 uses lower voltages (1.8V) and a differential interface to reduce noise
    DDR3 uses even lower voltages and doubles up the external clock speed

       L3 cache –extension of level 2 cache now used in Core i3 / i5 / i7 processors
        – originally intended for Server Applications rather than desktop PCs – uses
        Static RAM

       L2 Cache memory – on chip operates at processor clock speed - uses
        Static RAM

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 19
       L1 Cache memory: operates at the full processor speed - may have very
        wide data bus 256 bits wide - uses Static RAM
       Registers – direct access at full processor speed – fastest of all - uses
        Static RAM

Memory type                  size                          speed       role
Processor registers          √ EAX etc                     √√√√√√√     Part of x86 arch
L1 cache                     √√     64 kB                  √√√√√√      Fastest cache
L2 cache                     √√√    256 kB                 √√√√√       May be shared by cores
L3 cache*                    √√√√ 8 MB                     √√√√        Usually shared by cores
System RAM                   √√√√√ 2 GB                    √√√         Main memory
Turbo Memory                 √√√√√ 1 GB                    √√          Used by W7 ReadyDrive
Page File                    √√√√√√ (500MB)                √           Used by OS

Note: Turbo Memory does not appear to be supported by Intel Series 6 chipsets

Used for BIOS Basic Input Output System which holds code to test and boot up
the PC after reset

The BIOS can be shadowed i.e. it can be copied into fast main memory which is
then electronically switched to occupy the same address range as the BIOS

ROMs include :
   Flash memory: reprogrammable in situ .. useful since it is possible to update
    contents using a Internet link etc.

RAM: Main memory now usually connected to the processor chip
Static memory uses flip-flops – say 6 transistors per memory cell – used in cache
Dynamic memory uses capacitors – say 1 transistor per cell – used for system
So dynamic memory devices are much larger than static memory
DDR, DDR2 and DDR3 SDRAM has internal logic which buffers data and thus
produces external speeds which are faster than the memory chips
Serial Presence Detect: the chips in a DIMM socket may be able to be interrogated
to determine the module type and speed .. this uses the IIC-bus
DDR3 is the latest dynamic memory technology – uses lower voltages than DDR2
hence lower power, but can potentially offer higher data bandwidth since it can
transfer data at x2 the rate of DDR2 due to smart internal buffering – latest
processors only support DDR3 e.g. Corei3,i5 and i7
Handles – resources managed by the OS (Windows XP) e.g. a serial port
Threads – a single sequential activity in a program
Processes – a program such as Word, IE7
29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 20
Physical memory – the memory from which the processor is currently executing
Commit Charge (Page File Usage) – basically the totally ‘memory’ which the OS has
committed to all programs – if the physical memory is large enough accesses to the Page
File should be limited
Kernel Memory – RAM used by the OS for drivers etc
CPU Usage – a measure of how much time in the last few seconds the CPU has been
active – if the CPU usage is high then the CPU is too busy and tasks will execute slowly
 Software has three essential elements :
    code/ data/ stack
However clever the processor is, a significant part of the time will be spent in
accessing main memory either directly or indirectly through the cache memories
(see D5) - A slow main memory will mean that all these activities occur at a rate
which has a dramatic effect on program execution time.
This effect is at its worst when all the main memory is filled by applications
because the only way the Operating System can get addition memory is to swap
RAM based software elements to disk... the main memory is fast whereas the disk
transfer rates are many times slower than this ( remember the lab exercise to
measure disk access times )

Adding more RAM will improve the performance of a PC if the performance was
being limited by too small a RAM – this may be caused by opening too many
programs at the same time resulting in the excessive use of the page file on the
hard disk – if a PC already has enough RAM indicated in Task Manager by a large
amount of available RAM or a Commit Charge which is less than the Physical RAM
then the immediate result will be a very limited improvement in performance

Certain chipsets/ processors allow dual channel system memory – essentially two
64-bit RAM modules operate in parallel so the potential gain in performance is X2
for RAM accesses – HOWEVER not all activities require a high memory bandwidth
so the performance will depend on the type of activity – is it processor/ RAM/
network/ disk/ graphics intensive ? Core i7-9xx has three DDR3 RAM channels

Cache memory works so well because typical programs do not make spurious
accesses to random parts of the memory

Typically the next instruction is in the next memory location .. not true for a jump,
call or interrupt - but normally the case
Typically the next data item to be read is in the next location in RAM e.g. in a
data base there will be a small block of data in use at any one time
Overall the conclusion is that this localization of code and data makes it worth
while loading the current code and data areas into fast cache memories

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 21
As we have seen, the goal of processor designers is to get cache memories which
are large and operating at full processor speed
L1 caches are positioned very close to the cpu and can be accessed at full
processor speed and very efficiently
L2 caches are usually larger than L1 caches and operate at full processor speed
but are only accessed if the code or data cannot be found in the L1 cache
If a particular activity cannot find what is required in the L1 cache it looks in the L2
cache – the L1 cache will be updated
If the L2 doesn’t have the data then the cpu will go to main memory – the L2 and L1
cache will be updated

Modern processors often have a L3 cache on chip – this provides an additional
stage before going to the slow main memory
Cache memories require lots of transistors and these chips become increasingly
complex and expensive see D7
The BIOS has been used to provide support for booting up a PC for thirty years
The BIOS ideas are very much based on a 16-bit MSDOS environment
Unified Extensible Firmware Interface provides
    a more sophisticated 32-bit/ 64-bit environment for modern processors
    allows graphics and networking to be used at an early stage in the bootup
    should facilitate faster and more secure booting up
    cpu independent
    cpu independent drivers
    no file system is specified by EFI so works for alternative OSs to Windows

Answers to Topic 5
PC Peripheral Technology – see in-class tutorial

E1      Legacy Peripheral Devices include:
         PS/2 Keyboard
         PS/2 mouse
         LPT1: Parallel Port Printer
         COM ports ( Serial Ports ) e.g. COM1: to 4:

       LPT1: the Parallel Printer Port
The Parallel Port has very little which can be configured by the end user which is
ideal for the PC Product which aimed to de-skill activities such as configuration
The original PC Printer standard was designed by Centronics who manufactured
The standard basically consisted of :
        8-data lines
29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 22
         a small number of control/ status lines such as Strobe, Acknowledge,
          Busy, Out-of-Paper, Off-line
         the data transfer rate is controlled by the PC and uses the Strobe signal
          to indicate the presence of valid data on the 8-data lines

Thus there is no need to configure Baud rate, data bits, parity, stop bits, hardware
or software flow control which are needed for the Serial COM ports

        the PS/2 keyboard uses a serial link to the keyboard interface on the PC
motherboard ( this is supported on the P35 chipset by the legacy support chip) –
this is classically supported by IRQ1

The keyboard contains its own microcontroller which monitors key activity – e.g. key
down and key up – serial messages (called scan codes) are sent to the
motherboard which interprets this information, since Windows has been configured
with the information about the installed keyboard.

      COM ports provide a serial communications link usually to the RS232
standard. Basically it is necessary to set up the baud rate, word length, parity etc.
as well as flow control. The voltages used are typically +/- 9 volts

Software accesses peripherals through IO ports – remember the first lab –
COM1: is situated at IO port address 037F hex for example

      The mouse, keyboard and speaker are legacy components which are difficult
      to update without breaking the compatibility with existing software

      In addition the position of the legacy peripheral connectors (at the back of the
      PC) is inconvenient

      The USB keyboard allows a fresh start.. the USB keyboard might also act as a
      HUB for further peripherals e.g. a USB mouse, USB scanner, a USB games
      console, USB printer

      In addition a modern display system could also be controlled through USB
      allowing Control Panel style display management

      Since modern chipsets (e.g. P35) doesn’t support a mouse/ keyboard/ printer/
      COM1/ floppy disk – these are supported by a legacy support chip – Intel
      hoped all peripherals would use USB by 2005 ! – however PS/2 keyboard and
      PS/2 mice are often very cheap (mass production) and many PCs use these
      rather than more versatile but more expensive USB peripherals
There are three basic techniques for servicing peripheral devices
 polling ( checking a device such as a modem for new data)
29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 23
 interrupt servicing ( on getting new data the device requests servicing)
 Direct Memory Access ( where a device is directly connected to memory)
The modern PC interrupt structure supports 24 IRQ (interrupt sources) using a
device called an APIC Advanced Priority Interrupt Controller. Some devices are
connected internally and some externally - traditionally these are connected :
IRQ0 Timer to provide a real-time-clock                                see lab mpl210L
IRQ1 keyboard

possible interrupt sources:
system timer, keyboard, hard disk, floppy disk, sound card, serial ports, USB sub-
system, mouse, printer, network card etc.
Sequence following an interrupt event ( e.g. system timer IRQ0)
    The system clock hardware generates an interrupt request to the APIC
    The APIC receives and queues the interrupt request
    If the interrupt is of higher priority than any current interrupt, then the APIC
      signals to the processor via an INTR that there is one (or more) pending
Note: IRQ0 the System Timer is the highest priority interrupt
    assuming the interrupts are enabled, then the processor will complete the
      current instruction and respond with an INTA signal
    The APIC will then transfer the interrupt number associated with the System
      Timer interrupt handler to the processor
    The processor will get the address of the Interrupt Handler from the interrupt
      vector table
    the body (i.e. code) of the Interrupt handler will execute
    when complete the interrupt handler will inform the APIC it has completed
Shared interrupts
With shared interrupts, there is a second stage in checking the specific source
(which of the shared resources is actually generating an interrupt request)
PCI and ISA Interrupt Systems
The classic interrupts System Timer (IRQ0), Keyboard (IRQ1), COM1, PS/2 mouse
etc are termed ISA interrupts and have individual inputs – these IRQ sources have
changed very little from the original IBM PC – all PCs have IRQ0 as the System
Timer as a legacy feature
PCI interrupts from graphics card, sound card, USB etc use a more flexible
configuration where a single interrupt source can be shared by several sources
The Plug-and-Play BIOS – interrupt configuration
When a PC boots up the BIOS identifies the devices which require interrupt support
and configures the APIC – each device on the PCI bus has a ROM which allows
the device to be identified – obviously device drivers will be required – this is not a
29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 24
problem in a stable system but when new hardware is added and the system is
rebooted then requests will occur for device drivers from CD or the internet
USB – all modern PCs have USB ports - simple with very robust connector – low
cost smart silicon interfaces – USB requires assistance from main processor and is
usually master node is usually integrated into chipset – a tiered star technology
with up to 5 tiers and up to 127 USB nodes – USB2.0 480 Mbps – cable has two
power and two signal lines - can supply power to a peripheral max 5 Volts 500 mA
– standard cable 5 metres - typical peripherals: general purpose kb, mouse, printer,
webcam, modem etc
Firewire – few budget PCs have Firewire ports - standalone technology – no
requirement for a processor – high data rates 400/800 Mbps – often used for
specialist video peripherals – usually requires extra chip on motherboard -
USB3.0 – Higher transfer rates (up to 4.8 Gbps) -Increased maximum bus power and
increased device current provision to better accommodate power-hungry devices – new
connectors and cables for higher speed data transfer...although they are backwards
compatible with USB 2.0 devices and computers – ideal for high speed backup.
But see comments on Light Peak

PCs with USB3.0 are now available but the major application is fast data transfer to disk
Hot swapping means a peripheral can be connected or disconnected without
powering down the complete PC or Server – usually the power connectors are
arranged to connect before the data connectors by having pins of different lengths
– many hot swap peripherals expect the device to be effectively powered down
through a hot swap utility before removal
-   disconnecting an active hot swap disk drive may be disastrous !

Light Peak (an optical fibre technology using low cost optical fibre techniques) was
originally announced as a general purpose replacement for many interface
technologies e.g. HDMI – the technology would be introduced at 10 Gbps and could
be upgraded to 100 Gbps

The current position (Feb 2011) is that a copper version is to be delivered at 10
Gbps which obviously simplifies the motherboard interface etc

Light Peak is perhaps an alternative to USB3.0 which has not been implemented on
mainstream chipsets e.g. Q67

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 25
Answers to Topic 6
The PC bus Technology

F1 see in-class tutorial
name      width     clock                   transfer char              data rate       typical use
PC-bus 8-bit        5 MHz                   1 transfer/ 2 clocks       2.5             everything
ISA-bus       16-bit        8.3 MHz         1 transfer/ 2 clocks       8 MByte/sec     not memory
PCI-bus       32-bit        33 MHz          1 transfer/ clock          132             audio,
                                                                       MByte/sec       graphics,
AGP-          32-bit        66 MHz          x2 x4 etc transfers/       264             graphics
bus                                         clock                      MByte/sec       card
PCI-X         64-bit        133 MHz         1 transfer/ clock                          Servers
PCI-          1, 4, 8,      2.5 GHz         1 transfer/ clock          2.5 Gbps        graphics and
Express       16, 32                                                   Per lane each   low cost
1.0           lanes                                                    direction       cards
USB 1.1       1-bit         12 MHz          1 transfer/ clock          12 Mbps         simple mm
USB 2.0       1-bit         480 MHz 1 transfer/ clock                  480 Mbps        video data
USB 3.0       1-bit         4800 MHz 1 transfer/ clock                 4800 Mbps       Fast backup

In the original PC, the expansion bus was not a limiting factor in the speed of the
Once main memory was placed on the motherboard next to the processor, the
traffic to video, disks and communications cards was light because MS-DOS is lean
and mean unlike Windows which puts huge demands on graphics etc.

The PC/104 bus is low cost, well supported by industrial processor modules and IO
modules. Systems can be implemented by stacking IO modules on a processor
module. There is no graphics/ audio requirement so such systems have sufficient
processing power and are relatively low cost – industrial x86 products have very
long products compared with the PC market

data rate for a sound card is low -
CDs require sampling at 44.1 kHz and 16-bit resolution and stereo
(sampling rate) times (resolution) times (channels) = 44100 times 2-bytes times 2=
176.4 kBytes/sec ( PCI and USB ok – PCIe x1 ok )

data rate for a graphics card is high
say 1000 by 1000 with 24-bit colour and 50 frames /second
= 1000 times 1000 time 3 times 50 = 150 MBytes /sec or 1.2 Gbps (PCIe only)

data rate for a 10/ 100 Mbps Ethernet card is 10 Mbps = 1.25 MBytes/sec to 100
Mbps = 12.5 MBytes/sec (PCI ok; PCIe ok ; USB2.0 ok for 10 and 100 Mbps)
29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 26
PCIe ok for Gigabit Ethernet at 1000 Mbps !!

The PCI Express bus uses a number of bidirectional serial lanes which can each
allow data transfers at 2.5 Gbps ( 2.5 Giga-bits per second) in each direction. Each
lane provides a data transfer rate of 250 MBytes per second since 10 bits are used
for each byte including error checking (so called 10-bit 8b10b encoding )

Many applications such as a sound or network card could be connected to PCIe x1

Graphics requires a huge bandwidth so graphics cards currently require PCIe x16
so 16 lanes give a bandwidth of 250 Mbytes/sec times 16 = 4000 MBytes/sec

The latest chipsets allow 32 lanes so two PCIe x16 graphics cards can be

The x 4 and x8 interfaces are not used on desktop systems but may be used for
connecting disk drives etc to a server motherboard (Raid etc)

Graphics processing increases in power by about 30% per year. This is more
than the increase in cpu processing (20% per year)

Thus the PCI bus which had been designed to be future proof, can't meet the
speed requirements of video.. the story is more complicated than that because
graphics memory requirements are huge.

How can graphics memory be expanded ? Memory must be configurable.. the
idea is that the graphics card can store a texture map of the object it is displaying in
main RAM, giving the graphics card the whole of fast graphics RAM for use as the
local frame buffer

PCI Express (PCIe) allows high speed data transfers – each lane can provide
2.5 Gbps (serial) which can be configured as x1, x4, x16 x32 etc.
So PCIe x16 operates at 2.5 Gbps times 16 or about 4 GBytes/sec – enough for
two graphics cards

The Memory Controller Hub is at the high speed end of the chipset and is
adjacent to the CPU – the IO Controller Hub is much lower performance – only
supports PCIe x1 , SATA etc

The PCIe x1 connector is simple, cheap, small and ideal for PCs with a small
form factor. There are now adapter cards which plug into PCIe x1 slots e.g.
soundcard, network card, wireless card etc

It would appear to be Intel’s intention to remove direct PCI support from its chipsets
since all PCI functions can be implemented by PCIe x1 which has a smaller
connector and simplifies motherboard design. Some motherboard companies have
29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                                Microprocessor Systems Revision Tutorial 2011 27
reinstated PCI for chipsets which don’t support PCI by using a dedicated PCI chip.
Same issue as the legacy free PC – customers like familiar interfaces !!

Plug and Play means that adapter cards or peripheral interfaces have a built in ID
 Thus on booting up a system the Plug and Play enumerator checks the devices
    which are installed in the PC and confirms that appropriate device drivers
    (special interface software) are available for each card.. if unavailable the user is
    requested to provide the driver via a CD-ROM, floppy disk or the URL of the

This makes reconfiguring a PC or upgrading a PC simple.

USB also features Plug and Play but additionally USB peripherals are hot
swappable.. there is no need to switch the PC off or reset the PC to use a new USB
peripheral e.g. a modem

In general the PCI bus and USB feature Plug and Play

     The latest Core i3, i5, i7-8xx processors have an integrated memory controller
     to avoid bottlenecks in the RAM interface – these processors also have an
     integrated graphics unit (GPU) and so there are fast connections between the
     CPU/ GPU and system memory (system memory is also shared with the GPU)

Outline Answers for Topic 7
Disk and Backup Technology
         technology           transfer rate             Size/cost      use for backup         comments
Floppy   Generic NEC          < 1 Mbps                  1.44 Mbytes/   Low cost but           obsolete
Disk                                                    25p            unreliable             technology
Hard     basic drive unit -   e.g. SATA 3 Gbps at       500 Gbytes     useful as an on-line   very cost effective
Disk     is still EIDE but    300 MBytes/sec            /£50           backup - Raid          – cache + smart
         interface is                                                                         disk access
         Serial ATA           typical data rate up to
                              50 MBytes/sec
SSD      NAND flash           Interface SATA/ 3         Kingston 40    problems reported      Useful for fast
                              Gbps (300 Mbytes/s)       GB SSD/        in retrieving data     boot-up
                                                        £70            from crashed drives
                              Read speeds reported                                            Emerging
                              at 250 MB/sec                                                   technology
CD-      ATAPI                150 kbps x n              640 Mbytes     ok for limited         Disks can be
ROM                                                     /25p           backup                 difficult to read on
                                                                                              different PCs
DVD      DVD-R                150 kbps x n              5 Gbytes       ok for system          Readability
                                                        /25p           backup                 problems
Flash    flash memory         20-100 Mbps               8 MB to 32     useful for personal    easy to use and
Drive    device(s) with       (well below USB2.0        GB/ £10 for    backup but can be      getting larger all
29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                                  Microprocessor Systems Revision Tutorial 2011 28
              USB2.0          limit)                  smaller          unreliable              the time
              interface                               drives
Tape          SATA/ USB 2.0   30/ 20 MBytes/sec       40/80/120        still main enterprise   300 pounds plus
                                                      GBytes           backup technology       for simple tape
                                                      compressed                               unit plus 20
                                                                                               pounds /tape

Serial ATA
Simple parallel cable systems can only operate at relatively low frequencies before
the error rate becomes very high due to cross talk between signal lines.
The solution is to send the data on a serial line, which can run much faster and be
properly screened and use matched loads to stop reflections.
The problem is that each end is more complex – however this is integrated into the
VLSI chip sets used on the motherboard and in the disk drive

                              SATA feature                      comment
Application areas             PCs and general purpose           SAS outperforms in terms of reliability
Interface technology          serial                            4 signal lines
Signalling technology         differential                      Avoids crosstalk
Cabling/ connectors           simple                            Rather flimsy construction
Hot swapping                  possible                          Useful in servers
Cable length                  1 metre                           Easy access within a PC or server case
External Connection           eSATA                             Limited take up - See below question G8
2.5 “ and 3.5” drives         Use same connector                2.5” lower power but currently slower
Data rates/ Data Transfer     SATA 1.5 Gbps/ 1.2 Gbps           Coding system means 8 data bits can
rates                                                           be extracted from 10-bit 8b10b
                              SATA 3.0 Gbps/ 2.4 Gbps           Current generation product
                              SATA 6.0 Gbps/ 4.8 Gbps           Next generation product
                                                                e.g. Intel Series 6 chipsets
                                                                Cougar Point
Advanced Host Control         Full feature SATA requires        Windows XP only supports IDE
Interface                     motherboard and chipset           emulation mode
NCQ (Intelligent seeking)     Requires AHCI                     Requires Vista/ Windows 7 not XP
Rotational Speed - RPM        5400, 7200 – some 10000           7200 for desktop
Latency/ seek/ write          4.2 mS/ 11 mS/ 13 mS              For RPM = 7200
Cache/ data buffer            8 MB typical size                 Not much advantage in larger caches
Typical size                  500GB and more
cost                          50 pounds                         Good value for money
reliability                   good                              Not so good as SAS

Intelligent Seeking
Modern disk drive controllers can order requests for data from a disk to be more
efficient – say there are 10 requests pending which are alternate inner-outer, the
disk heads will move in-out-in-out etc – why not alter the data access to 5 inner and
then 5 outer – this is Native Command Queuing or Intelligent Seeking

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 29
G3 (diagram see lecture notes)
The small computer systems interface SCSI is a relatively old standard. It has
never been very popular in PCs. In general for large disks at high speed, high
reliability - the best choice is a SCSI drive but the cost is many times that of a
SATA drive
SAS – uses the same interface cables/connectors as SATA but the drive
specification is that of a SCSI drive – typically high speed, fast access
Server technology will benefit from a state-of-the-art disk drive ( big and fast)
whereas a Server doesn't require state of the art graphics
SATA – used in PCs and entry level servers; drive uses ATA technology with a
serial interface; low cost; large capacity; 7200 RPM; good latency, seek time and
write time; typical size 500 GB; NCQ; good reliability
SAS - used in expensive servers; based on SCSI drive technology with serial
interface; expensive; typically smaller than SATA; 10,000/ 15,000 RPM; excellent
latency, seek time and write time; entry level size about 80 GB; Intelligent Access,
excellent reliability
Floppy Disk            Simple, can ZIP                   not very big
Second hard            relatively expensive              may get corrupted
disk - Raid
RW CD-ROM              cheap and quite                   doesn't easily facilitate an
                       reliable                          incremental backup
Tape                   traditional enterprise            low cost, small, slow, facilitates
                       backup medium                     incremental backup

Cost – very approximate prices
   Hard Disk 500 GByte @ 50 pounds                                    = 10 GByte/ pound
   Floppy Disk 1 MByte @ 20p                                          = 5 MByte/ pound
   USB Flash RAM 4GByte @ 10 pounds                                   = 400 Mbyte/ pound
   Tape 20 GBytes @ 5 pounds                                          = 4 GByte/ pound
   DVD 5 GByte @ 50p                                                  = 10 GByte/ pound

home use     DVD, Flash, external USB i/f hard drives
small office DVD, tape, dedicated removable hard disks                           Cloud ?
large org    complex tape drives/ complex network
             attached storage units

File Allocation Table
 simple way to catalogue files on a disk
gets in a mess if system crash
Modern File Systems e.g. NTFS ( NT File System)
 allow journal files.. updates are stored allowing recovery back tracking
 system can repair file damage
29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc
                                             Microprocessor Systems Revision Tutorial 2011 30
no G6

Disk drives cause problems in mobile systems because of weight and vibration
An attractive alternative is to use solid state (Flash) memory as used in memory sticks –
the problem is storage size and cost
The smallest useful SSD solid state drive is 4 GB as used in netbooks – the cost of 4 GB
of Flash RAM is about 20 pounds (5 pounds/ GByte)

An option is to use a small SSD to boot the OS from and a larger hard disk to store user
Cache: A flash-based SSD uses a small amount of DRAM as a cache, similar to the cache
in Hard disk drives. A directory of block placement and wear leveling data is also kept in
the cache while the drive is operating

There are reports that a crashed SSD can be difficult to recover data from – however this a
new area whereas tape and disk recovery is mature
eSATA (external SATA) – potentially a good solution since the data does not need to be
converted from on board SATA format so 1500 Mbps – however requires a separate
power supply – more of a problem is that few PCs have an eSATA connector so an
eSATA adapter is needed – extra cost !!

USB - data needs to be ‘repackaged’ by USB interface so this will limit the data rate –
max is 480 Mbps – nearly every PC has a USB adapter so all is needed is device driver
software – USB can supply 5 volts at 500 mA so no external supply may be needed

Firewire – data needs to be repackaged for Firewire so this will limit data rate – 800 Mbps
maximum data rate – need Firewire port on PC – most budget PCs don’t support Firewire !

29-09-12 10:36 AM G E South 0156365e-702a-4779-9393-fc319d90847b.doc

To top