Skip to content

README.md

Latest commit

 

History

History
37 lines (28 loc) · 3.51 KB

README.md

File metadata and controls

37 lines (28 loc) · 3.51 KB
Raw

32-bit RISC Processor Design & FPGA Implementation

Overview

This repository contains the complete RTL design, simulation, and hardware implementation of a custom 32-bit RISC-like processor written in Verilog. Developed over a 5-week iterative design cycle, the project spans from the initial Instruction Set Architecture (ISA) definition and Arithmetic Logic Unit (ALU) design to a fully autonomous system running custom machine code on a Nexys A7 FPGA.

The processor features a custom hardwired control unit, a robust memory subsystem utilizing FPGA Block RAM (BRAM), and a multiplexed Design-for-Testability (DFT) I/O interface. The final implementation successfully executes complex, multi-cycle algorithmic workloads autonomously at the board's native 100MHz clock speed.

Architecture Specifications

  • Data Path: 32-bit architecture for all operations, memory addresses, and data buses.
  • Register File: Sixteen 32-bit general-purpose registers (R0 to R15), organized with two read ports and one write port. R0 is hardwired to 0.
  • Control Registers: 32-bit Program Counter (PC) and 32-bit Stack Pointer (SP).
  • Memory Subsystem: Byte-addressable memory requiring word-aligned access (multiples of 8) for 32-bit loads and stores. Both Instruction ROM and Data RAM are synthesized using on-chip BRAM.

Instruction Set Architecture (ISA)

The custom ISA supports diverse addressing modes including Register, Immediate, Base, PC-relative, and Indirect.

  • Arithmetic & Logic: ADD, SUB, AND, OR, XOR, NOR, NOT, SL, SRL, SRA, INC, DEC, SLT, SGT, LUI, and immediate variants.
  • Hardware Accelerators: Includes a dedicated HAM instruction to compute the Hamming Weight (population count) of a 32-bit word in hardware.
  • Memory & Transfer: LD, ST (Base addressing), MOVE, and conditional move CMOV.
  • Control Flow: Unconditional branch (BR), conditional branches (BMI, BPL, BZ), and processor halting (HALT).

Repository Structure & Project Milestones

The repository is structured to reflect the systematic, phase-by-phase development of the processor:

  • /Objective1&2_ALU: Contains the custom 32-bit ALU Verilog source, datapath schematics, and instruction encoding formats.
  • /Objective3_DataPath_Integration: Features the integrated core (ALU + 16-register bank) and a multiplexed DFT display utilizing 16 FPGA LEDs to verify 32-bit register-to-register operations via hardware switches.
  • /Objective4_Memory_operations: Houses the BRAM Data RAM modules. Includes manual-stepping logic to verify base-addressing modes for LD and ST instructions via a two-button execute/reset interface.
  • /Objective5: Upgrades the processor to an autonomous Run/Idle Finite State Machine (FSM). Integrates a BRAM Instruction ROM to fetch and execute a "Sum of Integers" program autonomously at the 100MHz system clock.
  • /ObjectiveFinal: The capstone algorithmic validation. Contains the generated bitstreams and BRAM initialization files (.coe) for executing two independent workloads:
    1. Booth's Multiplication: Evaluates a 16-bit signed integer multiplier/multiplicand, storing a 32-bit product.
    2. Total Hamming Weight: Iterates over a 5-word memory array, utilizing the hardware HAM instruction to accumulate the set-bit count.

Technologies Used

  • Hardware Description Language: Verilog
  • Target Hardware: Nexys 4 DDR / Nexys A7 FPGA (Xilinx Artix-7)
  • Development Tools: Xilinx Vivado (Synthesis, Simulation, BRAM IP Generation)