IAY 0600 Digitaalsüsteemide disain Register Transfer Level Design. FSM Synthesis. Alexander Sudnitson Tallinn University of Technology.

Slides:



Advertisements
Similar presentations
IO Interfaces and Bus Standards. Interface circuits Consists of the cktry required to connect an i/o device to a computer. On one side we have data bus.
Advertisements

Control path Recall that the control path is the physical entity in a processor which: fetches instructions, fetches operands, decodes instructions, schedules.
Give qualifications of instructors: DAP
1 Register Transfer &  -operations Computer Organization Computer Architectures Lab REGISTER TRANSFER AND MICROOPERATIONS Register Transfer Language Register.
Processor System Architecture
CS 151 Digital Systems Design Lecture 37 Register Transfer Level
CHAPTER 4 COMPUTER SYSTEM – Von Neumann Model
Lecture 7 Sept 17 Goals: Complete Chapter 4 Chapters 5 and 6.
4/10/20081 Lab 9 RT methodology introduction Register operations Data Path Control Path ASM Example TA: Jorge Crichigno.
9/15/09 - L25 Registers & Load Enable Copyright Joanne DeGroat, ECE, OSU1 Registers & Load Enable.
Computer Architecture Lecture 08 Fasih ur Rehman.
The Structure of the CPU
IAY 0600 Digitaalsüsteemide disain Course Overview Alexander Sudnitson Tallinn University of Technology.
1 Part V: VHDL CODING. 2 Design StructureData TypesOperators and AttributesConcurrent DesignSequential DesignSignals and VariablesState Machines A VHDL.
Machine Instruction Characteristics
MICROPROCESSOR INPUT/OUTPUT
IAY 0600 Digitaalsüsteemide disain Register Transfer Level Design (GCD example) Lab. 7 Alexander Sudnitson Tallinn University of Technology.
WIKIPEDIA HAS MANY MORE DIVISIBILITY RULES. EXAMPLE Since 52=13(4) is divisible by 4, is divisible by 4 Since 452=56(8)+4 is not divisible.
Computer Design Basics
Finite State Machines VHDL ET062G & ET063G Lecture 6 Najeem Lawal 2012.
11/17/2007DSD,USIT,GGSIPU1 RTL Systems References: 1.Introduction to Digital System by Milos Ercegovac,Tomas Lang, Jaime H. Moreno; wiley publisher 2.Digital.
Lecture 9 RTL Design Methodology. Structure of a Typical Digital System Datapath (Execution Unit) Controller (Control Unit) Data Inputs Data Outputs Control.
VHDL Discussion Finite State Machines
VHDL Discussion Finite State Machines IAY 0600 Digital Systems Design Alexander Sudnitson Tallinn University of Technology 1.
RTL Design Methodology Transition from Pseudocode & Interface
VHDL Discussion Subprograms IAY 0600 Digital Systems Design Alexander Sudnitson Tallinn University of Technology 1.
Overview von Neumann Architecture Computer component Computer function
Lecture 5B Block Diagrams HASH Example.
IAY 0600 Digital Systems Design Digitaalsüsteemide disain Course Overview Alexander Sudnitson Tallinn University of Technology.
Lecture 3 RTL Design Methodology Transition from Pseudocode & Interface to a Corresponding Block Diagram.
IAY 0600 Digital Systems Design Register Transfer Level Design (GCD example) Lab. 7 Alexander Sudnitson Tallinn University of Technology.
Sept. 2005EE37E Adv. Digital Electronics Lesson 4 Synchronous Design Architectures: Control Unit Design (part three)
Processor Organization and Architecture Module III.
ECE 448 Lecture 6 Finite State Machines State Diagrams vs. Algorithmic State Machine (ASM) Charts.
1 3 Computing System Fundamentals 3.2 Computer Architecture.
Digital Computer Concept and Practice Copyright ©2012 by Jaejin Lee Control Unit.
5.3 Sequential Circuits - An Introduction to Informatics WMN Lab. Hey-Jin Lee.
Agenda Review:  Relation Properties Lecture Content:  Divisor and Prime Number  Binary, Octal, Hexadecimal Review & Exercise.
Finite State Machines (part 1)
Basic Computer Organization and Design
RTL Design Methodology Transition from Pseudocode & Interface
Behavioral Style Combinational Design with VHDL
IAY 0600 Digital Systems Design
Introduction Introduction to VHDL Entities Signals Data & Scalar Types
IAY 0600 Digital Systems Design
IAY 0600 Digital Systems Design
Behavioral Style Combinational Design with VHDL
ECE 448 Lecture 6 Finite State Machines State Diagrams vs. Algorithmic State Machine (ASM) Charts.
Computer Science 101 While Statement.
IAY 0600 Digital Systems Design
RTL Design Methodology
Repetition and Loop Statements
RTL Design Methodology
IAS 0600 Digital Systems Design
RTL Design Methodology Transition from Pseudocode & Interface
IAS 0600 Digital Systems Design
IAS 0600 Digital Systems Design
IAS 0600 Digital Systems Design
Finite State Machines (part 1)
RTL Design Methodology
RTL Design Methodology Transition from Pseudocode & Interface
RTL Design Methodology
RTL Design Methodology Transition from Pseudocode & Interface
RTL Design Methodology
RTL Design Methodology Transition from Pseudocode & Interface
ECE 448 Lecture 6 Finite State Machines State Diagrams, State Tables, Algorithmic State Machine (ASM) Charts, and VHDL code ECE 448 – FPGA and ASIC Design.
ECE 448 Lecture 6 Finite State Machines State Diagrams vs. Algorithmic State Machine (ASM) Charts.
COMPUTING.
(Sequential-Circuit Building Blocks)
Presentation transcript:

IAY 0600 Digitaalsüsteemide disain Register Transfer Level Design. FSM Synthesis. Alexander Sudnitson Tallinn University of Technology

2 Register Transfer Level The Register Transfer Level (RTL) is characterized by A digital system is viewed as divided into a data path (data subsystem) and control path (controller); The state of data path consists of the contents of a set of registers; The function of the system is performed as a sequence of transition transfers (in one or more clock cycles). A register transfer is a transformation performed on a datum while the datum is transferred from one register to another. The sequence of register transfers is controlled by the control path (FSM). A sequence of register transfers is representable by an execution graph.

3 Basic units of RT-level design Control Inputs Outputs Control Status InputsSignals Data path InputsOutputs CONTROL UNIT DATA PATH UNIT

4 A word description (example)‏ Digital unit performs an operation of computing the greatest common divisor (GCD) of two integers corresponding to Euclid algorithm: The gist of this algorithm is computing the remainder from division of the greater number with the less one and further exchanging the greater number with the less one and this less number with the division remainder. This converging process is looped until the division remainder is equal to zero. That means the termination of the algorithm with the current less number as the result. The design ranges over several levels of representation. We begin the design process with a word description of an example device.

5 Block diagram (example)‏ Start Ready OP1 OP2 ANSW DISCRETE SYSTEM The interface description entity EUCLID is port (START: in BIT; --The first and the second operand bus OP1, OP2: in INTEGER range 0 to 255; --Answer is ready signal READY: out BIT; --Answer bus ANSW: out INTEGER range 0 to 255); end EUCLID;

6 architecture COMMON of EUCLID is process -- Temporary variables: variable RG1, RG2, temp: INTEGER range 0 to 255; begin -- Waiting for the start: wait on START until START’event and START = ‘1’; RG1 := OP1; RG2 := OP2; if RG1 /= RG2 then if RG1 < RG2 then -- Exchange operands: temp:=RG1; RG1 := RG2; RG2:=temp; end if; while RG1 /= 0 loop -- Calculation of the reminder: RG1 := RG1 rem RG2; if RG1 /= 0 then temp:=RG1; RG1 := RG2; RG2:=temp; end if; end loop; end if; --Answer output: ANSW <= RG2; READY <= ‘1’; end process; end COMMON; Behavioral Description

7 The flowchart (example)‏ Yes RG1 := OP1; RG2 := OP2; RG1 = RG2 Yes No RG1 < RG2 RG1 := RG2; RG2 := RG1; Yes No Remainder Computation Remainder = 0 NoYes READY := 1; ANSW := RG2; END BEGIN START No

8 GCD computation of 15 and 24 OP1OP2 RG1 RG RG1 < RG2 RG1 := RG2; RG2 := RG1; Remainder = RG1 /= 0 RG1 := RG2; RG2 := RG1; 15 9 Remainder = 6 69 RG1 /= 0 RG1 := RG2; RG2 := RG1; 9 6 Remainder = RG1 /= 0 RG1 := RG2; RG2 := RG1; 6 3 Remainder = 0 03 RG1 = 0 READY := 1; ANSW := 3;

9 Data path -1- The data path is specified by the set of operations presented in the behavioral descriptions and by the set of basic elements which it will be implemented by. Notice that remainder computation chip (or macro) doesn’t exist. We need to synthesize it on the next design step basing upon its behavioural description and existing (or virtual) elements of the lower level - e.g. adders, shift registers, counters. It would in its turn lead to appearing the control part of the lower level and so on (top-down design methodology).

10 Remainder computation No RG1 := RG1 - RG2; RG2(7) = 1 Yes RG1(8) = 1 L1(RG2.0); C := C + 1; YesNo Remainder Computation RG1 := RG1 + RG2; C = 0 No R1(0.RG2); C := C - 1; Yes

11 The flowchart Yes RG1 := OP1; RG2 := OP2; RG1 = RG2 Yes No RG1 < RG2 RG1 := RG2; RG2 := RG1; Yes No Remainder = 0 NoYes READY := 1; ANSW := RG2; END BEGIN START No Remainder Computation No RG1 := RG1 - RG2; RG2(7) = 1 Yes RG1(8) = 1 L1(RG2.0); C := C + 1; YesNo RG1 := RG1 + RG2; C = 0 No R1(0.RG2); C := C - 1; Yes

12 Data path -2- Consider in our example the data path that is based upon some ALU which completes four arithmetic operations (addition, subtraction, left shift and right shift) with registers RG1 and RG2 for storing the intermediate results, with up/down counter and with control buses for data transfer. It is considered that RG1 and RG2 are Master- Slave registers that allows to exchange their contents during one clock cycle. Input operands are 8-bit wide. For this example it is assumed that input operands are positive and none of them is 0. Note, that RG1 and RG2 have a sign bit, as remainder computation algorithm deals with negative values as well.

13 The structure of GCD device ALU x1 x2 RG1 RG2 Counter y9y8 y5y4 y0 x4 x6 y7y6 y1 x3 y10 DATA PATH UNIT OP1 ANSW OP2 y3y2 x5 FSM ●●●●●● y0 y1 y10 x1 x0 X5 ●●●●●● READY START CONTROL UNIT

14 ALU OP1 OP2 x2 x1 OP1 >= OP2 0 OP1 < OP2 1 0 OP1 /= OP2 1 OP1 = OP2 Result y R1L1–+ y8 y9 y8

15 Registers and Counter y0 x4 RG1 enable SignSign NOR x6 y1 x3 RG1 enable SignSign y2 y3 Counter enable NOR x5 0 C C – 1

16 Multiplexers RG1 Input OP1 ALU RG2 y4 y5y4 y5 RG1 Input OP – 00ALU RG2 RG2 Input OP2 ALU RG1 y6 y7y6 y7 RG2 Input OP – 00ALU RG1

17 Control bus 8 A B y y10 &&& RG(0)‏ ANSW(0)‏ANSW(1)‏ANSW(7)‏ RG(1)‏RG(7)‏

18 Control part At every description level after the (regular) structure of data path is defined it is possible to extract the remaining control part from the current level of behavioral description. Naturally this extracted control part description may be at first only behavioral one and the methods of finite automata synthesis are required for control part (controller) implementation. In this stage it is convenient to represent the extracted control behavior by means of graph- scheme of algorithm (GSA). The flowchart corresponding to our algorithm was obtained as the first step of GSA synthesis. In this flowchart simultaneously executed statements are grouped into common blocks. The GSA we got from the flowchart by replacing the computational statements (actions of ALU and counter) with the corresponding control signals (y-s) and the conditions - with binary conditions signals (x-s).

19 Graph-scheme of algorithm BEGIN y9 y8 y3 y2 y1 x0 0 1 y7 y5 y1 y0 x1 x2 01 x3 x4 x5 1 y6 y4 y1 y0 0 y9 y2 y1 0 1 y8 y0 0 1 y0 0 x y10 END

20 Moore type FSM synthesis ‏ Step 1. The construction of marked GSA. At this step, the vertices “Begin”, “End” and oerator vertices are marked by the symbols s1, s2, … as follows: vertices “Begin”, “End” are marked by the same symbol s1; the symbols s2, s3, … mark all operator vertices; all operator verteces should be marked; Note that while synthesizing a Moore FSM symbols of states mark not inputs of vertices following the operator ones but operator vertices. Step2. The construction of transition list (state diagram) of a controller. SpresSnext X(Spres, Snext) Y(Spres)‏ Y(Snext)‏

21 Moore type FSM GSA BEGIN y9 y8 y3 y2 y1 x0 0 1 y7 y5 y1 y0 x1 x2 01 x3 x4 x5 1 y6 y4 y1 y0 0 y9 y2 y1 0 1 y8 y0 0 1 y0 0 x y10 END S1 S2 S3 S4 S5 S6 S7 S8 S1

22 The transition list (Moore FSM)‏

23 Microoperation and microinstruction Let a microoperation be an elementary indivisible step of data processing in the datapath and let Y be a set of microoperations. Microoperations are induced by the binary signals y 1, …,y T from a controller. To perform the microoperation y i (i = 1, …, T) the signal y i = 1 has to appear at the output y i. A set of microoperations executed concurrently in the datapath is called a microinstruction. Thus if  h = {y h 1, …, y h t } is microinstruction, then  h is represented as subset of Y and the microoperations y h 1, …, y h t are executed at the same clock period. The Y t could be empty and we denote such an empty microinstruction Y 0 (“-“).