fp!0_ppt

Upload: amitbanne

Post on 09-Apr-2018

219 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/7/2019 FP!0_ppt

    1/21

  • 8/7/2019 FP!0_ppt

    2/21

    Floating point Data processing instructions

  • 8/7/2019 FP!0_ppt

    3/21

    Th e ARM floating-point arc h itecture presents: An interpretation of t h e coprocessor instruction set w h en t h e coprocessor

    number is 1 or 2. ( Th e floating-point system uses two logical coprocessornumbers.)

    Eigh t 80-bit floating-point registers in coprocessors 1 and 2 (t h e sameph ysical registers appear in bot h logical coprocessors).

    A user-visible floating-point status register (FPSR) w h ich controls variousoperating options and indicates error conditions.

    Optionally, a floating-point control register (FPCR) w h ich is user-invisible

    and s h ould be used only by t h e support software specific to t h e h ardwareaccelerator.

  • 8/7/2019 FP!0_ppt

    4/21

  • 8/7/2019 FP!0_ppt

    5/21

    Th e coprocessor pipeline followerTh e load/store unit t h at carries out formatconversion on floating-point data typesas t h ey are loaded from and stored to memory.Th e register bank w h ich stores eig h t 80-bitextended precision floating-point operands.

    Th e arit h metic unit w h ich incorporates an adder,a multiplier and a divider,toget h er wit h roundingand normalizing h ardware.

  • 8/7/2019 FP!0_ppt

    6/21

    FPA10 Organization

  • 8/7/2019 FP!0_ppt

    7/21

    U se of Memory

  • 8/7/2019 FP!0_ppt

    8/21

    Run time environmentDivision and remainder functions.

    Since t h e ARM instruction set does not include divide instructions, t h ese areimplemented as library functions.

    Stack-limit c h ecking functions.A minimal embedded system is unlikely to h ave memory management h ardwareavailable for stack overflow detection; t h erefore t h ese library functions are

    needed to ensure programs operate safely.

    Stack and h eap management.All C programs use t h e stack for (many) function calls, and all but t h e most trivialcreate data structures on t h e h eap.

    Program start up.Once t h e stack and h eap are initialized, t h e program starts wit h a call to main ().

    Program termination.Most programs terminate by calling _exit (); even a program w h ich runs foreversh ould terminate if an error is detected.

  • 8/7/2019 FP!0_ppt

    9/21

    ARM memory system

  • 8/7/2019 FP!0_ppt

    10/21

    DRAM memory organization

  • 8/7/2019 FP!0_ppt

    11/21

    ARM address register structure

  • 8/7/2019 FP!0_ppt

    12/21

    DRAM timing diagram

  • 8/7/2019 FP!0_ppt

    13/21

    Tming diagram

  • 8/7/2019 FP!0_ppt

    14/21

    Overview of the AMBA specification

    Th e Advanced Microcontroller Bus Architecture (AMBA) specificationdefines an onchip communications standard for designing h igh -performance embedded microcontrollers.

    Th ree distinct buses are defined wit h in t h e AMBA specification: th e Advanced High-performance Bus (AHB) th e Advanced System Bus (ASB) th e Advanced Peripheral Bus (APB).

    A test met h odology is included wit h th e AMBA specification w h ich

    provides an infrastructure for modular macrocell test and diagnosticaccess.

  • 8/7/2019 FP!0_ppt

    15/21

    Objectives of the AMBA specification

    Th e AMBA specification h as been derived to satisfy four key requirements:

    to facilitate t h e right-first-time development of embedded microcontroller products wit h one or more CP U s or signal processors

    to be technology-independent and ensure that highly reusable peripheral and system macrocells canbe migrated across a diverse range of IC processes and be appropriate for full-custom, standard celland gate array tec h nologies

    to encourage modular system design to improve processor independence,providing a development road-map for advanced cac h ed CP U cores and t h e development of perip h eral libraries

    to minimize t h e silicon infrastructure required to support efficient on-c h ip and off-c h ipcommunication for bot h operation and manufacturing test.

    AMBAh as been proven in and is being designed into:

    PDA microcontrollers, wit h a h igh number of integrated perip h erals but also wit h very low powerconsumption

    multi-media microcontrollers wit h floating-point co-processors, on-c h ip video controller and h igh memory bandwidt h

    complex perip h eral ASICs for consumer products digital mobile communication devices integrating control and signal-processing functions

  • 8/7/2019 FP!0_ppt

    16/21

    A typical AMBA-based microcontroller

    In order to ac h ieve t h eabove, t h e arc h itectureh as t h e key features of:

    h igh ly modularity

    multi-master support

    low-power emp h asis

    innovative testmet h odology

  • 8/7/2019 FP!0_ppt

    17/21

    Th e Advanced Microcontroller Bus Arc h itecture (AMBA)

  • 8/7/2019 FP!0_ppt

    18/21

    Bus Transfer steps

    Th e bus clock, BCLK. This will usually be the same as mclk, the ARM processor clock.Th e bus master w h ich h olds t h e grant t h en proceeds wit h th e bus transactionusing t h e following signals:

    Bus transaction, BTRAN[1:0], indicates whether the next bus cycle will beaddress-only, sequential or non-sequential. It is enabled by t h e grant signal and isah ead of t h e bus cycle to w h ich it refers.

    Th e address bus, BA[31:OJ. (Not all address lines need be implemented in systems wit h modest address-space requirements, and in a multiplexed implementation t h e address is sent down t h e data bus.)

    Bus transfer direction, BWRITE. Bus protection signals, BPROT[1:0], which indicate instruction or data fetches

    and supervisor or user access.Th

    e transfer size, BSIZE[1:0], specifies a byte, half-word or word transfer. Bus lock, BLOK, allows a master to retain the bus to complete an atomic readmodify-write transaction.

  • 8/7/2019 FP!0_ppt

    19/21

    Th e data bus, BD[31:0], used to transmit write data and to receive read data. In animplementation wit h multiplexed address and data, t h e address is also transmitteddown t h is bus.

    A slave unit may process t h e requested transaction immediately, accepting write

    data or issuing read data on ED[31:0], or signal one of the following responses:

    Bus wait, BWAIT, allows a slave module to insert wait states when it cannot complete t h e transaction in t h e current cycle.

    Bus last, BLAST, allows a slave to terminate a sequential burst to force the busmaster to issue a new bus transaction request to continue.

    Bus error, BERROR, indicates a transaction that cannot be completed. If themaster is a processor it s h ould abort t h e transfer.

  • 8/7/2019 FP!0_ppt

    20/21

    H ardware system prototyping tools

  • 8/7/2019 FP!0_ppt

    21/21

    JTAG boundary scan register