CGIntroAndHardware_0b

download CGIntroAndHardware_0b

of 16

Transcript of CGIntroAndHardware_0b

  • 7/30/2019 CGIntroAndHardware_0b

    1/16

    Computer Graphics

    Prof. Muhammad Saeed

  • 7/30/2019 CGIntroAndHardware_0b

    2/16

    2

    Hardware

    ( Graphic Cards )

    II

    Hardware II Computer Graphics

    1 August 2012

  • 7/30/2019 CGIntroAndHardware_0b

    3/16

    3

    With a PCI-based graphics

    card, every texture map has tobe stored twice. First, the

    texture map is loaded from the

    hard drive to the system

    memory (RAM) until it has to be

    used. Once it is needed, it is

    pulled from memory and sent to

    the CPU to be processed. Once

    processed, it is sent through the

    PCI bus to the graphics card,

    where it is stored again in the

    card's framebuffer. All of thisstoring and sending between

    the system and the card is very

    draining to the overall

    performance of the computer.

    Hardware II Computer Graphics

    PCI(Peripheral Component Interface)

    1 August 2012

  • 7/30/2019 CGIntroAndHardware_0b

    4/16

    AGP( Accelerated Graphics Port) Card

    4

    AGP is built on the idea ofimproving the ways that PCItransports data to the CPU. Intelachieved this by addressing allof the areas where PCI transfers

    were causing data bottlenecksin the system. By clearing thetraffic jams of data, AGPincreases the speed at whichmachines can render graphics

    while using the system'sresources more efficiently toreduce overall drag.

    Hardware II Computer Graphics

    It achieves the goal by:

    1 August 2012

  • 7/30/2019 CGIntroAndHardware_0b

    5/16

    5

    1. Dedicated Port

    There are no other devices connected to the AGP otherthan the graphics card. With a dedicated path to the CPU,the graphics card can always operate at the maximumcapacity of the connection.

    2. PipeliningThis method of data organization allows the graphics cardto receive and respond to multiple packets of data in asingle request. Here's a simplified example of this:With AGP, the graphics card can receive a request for all ofthe information needed to render a particular image andsend it out all at once. With PCI, the graphics cardwould receive information on the height of the image andwait... then the length of the image, and wait... then thewidth of the image, and wait... combine the data, andthen send it out.

    Hardware II Computer Graphics

    1 August 2012

  • 7/30/2019 CGIntroAndHardware_0b

    6/16

    6

    3. Sideband addressingLike a letter, all requests and information sent from one part

    of your computer to the next must have an addresscontaining "To" and "From." The problem with PCI is thatthis "To" and "From" information is sent with the workingdata all together in one packet. This is the equivalent ofincluding an address card inside the envelope when yousend a letter to a friend: Now the post office has to open the

    envelope to see the address in order to know where to sendit. This takes up the post office's time. In addition, theaddress card itself takes up room in the envelope, reducingthe total amount of stuff you can send to your friend.With sideband addressing, the AGP issues eight additionallines on the data packet just for addressing. This puts the

    address on the outside of the envelope, so to speak, freeingup the total bandwidth of the data path used to transferinformation back and forth. In addition, it unclogs systemresources that were previously used to open the packet toread the addresses.

    Hardware II Computer Graphics

    1 August 2012

  • 7/30/2019 CGIntroAndHardware_0b

    7/16

    7

    AGP: Saves RAMAGP improves the process of storing texture maps byallowing the Operating System to designate RAM foruse by the graphics card on the fly. This type of

    memory is called AGP memory or non-local video

    memory. Using the much more abundant and fasterRAM used by the operating system to store texture

    maps reduces the number of maps that have to be

    stored on the graphics card's memory. In addition, the

    size of the texture map your computer is capable of

    processing is no longer limited to the amount of RAM

    on the graphics card.

    Hardware II Computer Graphics

    1 August 2012

  • 7/30/2019 CGIntroAndHardware_0b

    8/16

    8

    AGP saves RAM is by only storing texture mapsonce. It does this with a little trickery. This trickerytakes the form of a chipset called the GraphicsAddress Remapping Table (GART). GART takes theportion of the system memory that the AGPborrows to store texture maps for the graphicscard and re-addresses it. The new addressprovided by GART makes the CPU think that thetexture map is being stored in the card's frame

    buffer. GART may be putting bits and pieces of themap all over the system RAM; but when the CPUneeds it, as far as it's concerned the texture map isright where it should be.

    GART

    Hardware II Computer Graphics

    1 August 2012

  • 7/30/2019 CGIntroAndHardware_0b

    9/16

    PCI (Peripheral Component Interconnect ) Express Card

    9

    PCI Express is a serialconnection that operates more

    like a network than a bus.

    Instead of one bus that handles

    data from multiple sources,PCI-e has a switch that controls

    several point-to-point serial

    connections. These connections

    fan out from the switch, leading

    directly to the devices wherethe data needs to go. Every

    device has its own dedicated

    connection

    Hardware II Computer Graphics

    1 August 2012

    http://computer.howstuffworks.com/pci.htmhttp://computer.howstuffworks.com/pci.htm
  • 7/30/2019 CGIntroAndHardware_0b

    10/16

    10

    Each lane of a PCI Express

    connection contains two pairs ofwires -- one to send and one to

    receive. Packets of data move

    across the lane at a rate of one bit

    per cycle. A x1 connection, thesmallest PCI-e connection, has

    one lane made up of four wires. It

    carries one bit per cycle in each

    direction. A x2 link contains eight

    wires and transmits two bits at

    once, a x4 link transmits four bits,

    and so on. Other configurations

    are x12, x16 and x32.

    PCI Express Lanes

    Hardware II Computer Graphics

    A x16 PCI-e connector can move 6.4 GB

    of data per second in each direction.

    1 August 2012

  • 7/30/2019 CGIntroAndHardware_0b

    11/16

    11

    Computer Graphics

    The video BIOS or firmware contains the basicprogram, which is usually hidden, that governs

    the video card's operations and provides theinstructions that allow the computer andsoftware to interact with the card. It may containinformation on the memory timing, operatingspeeds and voltages of the graphics processor,RAM, and other information.

    Video BIOS

    1 August 2012

    Hardware II

  • 7/30/2019 CGIntroAndHardware_0b

    12/16

    12

    Computer Graphics

    Graphics Processing Unit (GPU )

    AGPU is a dedicated processor optimized for acceleratinggraphics. The processor is designed specifically toperform floating-point calculations, which are fundamental to3D graphics rendering and 2D picture drawing. The mainattributes of the GPU are the core clock frequency, which

    typically ranges from 250 MHz to 4 GHz and the number ofpipelines (vertex and f ragment shaders), which translate a 3Dimage characterized by vertices and lines into a 2D imageformed by pixels.Modern GPUs are massively parallel, and fully programmable.Their computing power is orders of magnitude greater than that

    of CPUs for certain kinds of operations. This has led to theemergence of general-purpose computing on graphicsprocessing units (GPGPU).Occasionally called visual processing unit orVPU.

    1 August 2012

    Hardware II

  • 7/30/2019 CGIntroAndHardware_0b

    13/16

    13

    Computer Graphics

    Compute Unified Device Architecture(CUDA)

    CUDAis a parallel computing architecture developed by Nvidia forgraphics processing. CUDAis the computing engine inNvidia graphics processing units (GPUs) that is accessible tosoftware developers through variants of industry standardprogramming languages.

    CUDAgives developers access to the virtual instruction set andmemory of the parallel computational elements in CUDAGPUs.Using CUDA, the latest Nvidia GPUs become accessible forcomputation like CPUs. Unlike CPUs however, GPUs have a parallel

    throughput architecture that emphasizes executing manyconcurrent threads slowly, rather than executing a single threadvery quickly. This approach of solving general purpose problems onGPUs is known as GPGPU.

    1 August 2012

    Hardware II

  • 7/30/2019 CGIntroAndHardware_0b

    14/16

    14

    Computer Graphics

    Power demandAs the processing power of video cards has increased, sohas their demand for electrical power. Current high-performance video cards tend to consume a great deal ofpower. While CPU and power supply makers have recentlymoved toward higher efficiency, power demands of GPUs

    have continued to rise, so the video card may be thebiggest electricity user in a computer. Although powersupplies are increasing their power too, the bottleneck isdue to the PCI-Express connection, which is limited to

    supplying 75 Watts. Modern video cards with a powerconsumption over 75 Watts usually include a combinationof six-pin (75W) or eight-pin (150W) sockets that connectdirectly to the power supply.

    1 August 2012

    Hardware II

  • 7/30/2019 CGIntroAndHardware_0b

    15/16

    15

    Computer Graphics

    High-Definition Multimedia Interface

    (HDMI)

    High-Definition Multimedia Interface (HDMI)An advanced digital audio/video interconnect released in

    2003 and is commonly used to connect gameconsoles and DVD players to a display. HDMI supportscopy protection through HDCP(High-bandwidth DigitalContent Protection) .

    1 August 2012

    Hardware II

    http://en.wikipedia.org/wiki/File:HDMI_Connector_Pinout.svg
  • 7/30/2019 CGIntroAndHardware_0b

    16/16

    16

    END

    Hardware II

    16

    Hardware II Computer Graphics

    1 August 2012