Comprehensive Cadence VLSI Investigation โ Cadence Genus Synthesis
๐ง VLSI Design Exercises using Cadence EDA Tools This repository showcases a collection of VLSI design exercises completed using professional tools like Cadence Virtuoso , NC Launch , and Genus . These exercises cover full-custom transistor-level design, RTL simulation, and synthesis of various digital systems. ๐ Tools Used ๐งฉ Cadence Virtuoso โ Schematic, Layout, and Custom Design ๐งช NC Launch โ Functional Simulation โ๏ธ Genus โ Logic Synthesis ๐ง Full-Custom Design (Using Cadence Virtuoso) Exercise Title GitHub Link 01 CMOS Inverter Design ๐ View 02 2:1 Multiplexer โ Full Custom ๐ View 03 D Flip-Flop โ Implementation & Analysis ๐ View 04 6T SRAM Cell ๐ View 05 1-Bit Full Adder ๐ View 06 2-Bit Multiplier ๐ View โ๏ธ RTL Simulation & Synthesis (Using NC Launch + Genus) Exercise Title GitHub Link 07 4-Bit Adder โ RTL Simulation ๐ View 08 4-Bit Up/Down Counter โ RTL ๐ View 09 32-Bit ALU โ Functional Verification ๐ View 10 4-Bit Up/Down Counter โ Synthesis ๐ View 11 32-Bit ALU โ Synthesis ๐ View 12 Traffic Light Controller โ Synthesis ๐ View
This repository showcases a collection of VLSI design exercises completed using professional tools like Cadence Virtuoso , NC Launch , and Genus . These exercises cover full-custom transistor-level design, RTL simulation, and synthesis of various digital systems.
RISC-V Single-Cycle Processor โ Synthesis & GLS Flow This repository documents the complete process of taking a single-cycle RISC-V RV32I processor from RTL (Verilog source code) all the way through logic synthesis and gate-level simulation (GLS) , using industry-standard EDA tools (Cadence Genus 18.1 and ncverilog) on a 45nm standard cell technology. This was done as part of an academic VLSI/digital design internship at IIT Bhubaneswar . The same flow was first validated on a simpler mod-10 counter design before being applied to a full RISC-V core, so if anything is unclear, that simpler project is a good reference point. This README is written so that someone with little or no prior exposure to the digital ASIC/VLSI flow can follow along and understand not just what commands were run, but why . Table of Contents Background Key Terms Explained Tools Used Directory Structure Flow Overview Setup Instructions Running Synthesis Reading the Reports Running Gate-Level Simulation (GLS) Viewing Waveforms Results Summary Report Screenshots Known Limitations Common Errors and Fixes Deliverables for Physical Design Handoff Notes Background When you design a digital circuit (like a processor), you usually start by describing its behavior in a hardware description language like Verilog . This Verilog code is called RTL (Register Transfer Level) โ it describes what the circuit should do (for example, "on every clock edge, update the program counter"), but it is not yet a physical circuit. To turn that RTL into something that can actually be manufactured as a chip, it has to go through a process called logic synthesis . A synthesis tool (here, Cadence Genus) reads the RTL and converts it into a netlist made of real, physical standard logic cells โ AND gates, flip-flops, multiplexers, and so on โ that exist in a specific manufacturing technology. In this project, that technology is a 45nm process node. Once the gate-level netlist exists, it is important to verify that it still behaves the way the original RTL did. This is done with Gate-Level Simulation (GLS) โ simulating the synthesized netlist (instead of the original RTL) and confirming that outputs still match expectations. In short, this project takes: RTL (riscv.v)
โ
โผ
Synthesis (Cadence Genus)
โ
โผ
Gate-Level Netlist (riscv_net.v)
โ
โผ
Gate-Level Simulation (ncverilog)
โ
โผ
Waveform Verification (SimVision) Key Terms Explained Term Meaning RTL Register Transfer Level โ Verilog/SystemVerilog code describing circuit behaviour Synthesis Converting RTL into a gate-level netlist using a standard cell library Netlist A text file listing actual logic gates and how they are connected Standard Cell Library (.lib) Describes all basic logic gates available in a given manufacturing technology, including timing, power, and area data 45nm The manufacturing process node โ refers to approximate transistor feature size. 45nm is commonly used in academic settings SDC Synopsys Design Constraints โ tells the synthesis tool the timing requirements (clock speed, I/O delays, etc.) Clock Period Duration of one clock cycle. 20 ns period = 1 / 20 ns = 50 MHz Slack Time margin in a timing path. Positive slack = timing met. Negative slack = timing violation GLS Gate-Level Simulation โ simulating the post-synthesis netlist to verify correctness Testbench Verilog code that drives inputs into the design and checks outputs during simulation VCD Value Change Dump โ records signal transitions over time for waveform viewing False Path A timing path that does not affect real circuit behaviour and should be excluded from timing analysis (e.g. asynchronous reset) Tools Used Tool Version Purpose Cadence Genus 18.10 RTL synthesis โ converts Verilog to gate-level netlist ncverilog Cadence Incisive Gate-level simulation SimVision Cadence Waveform viewing from VCD files Technology Library 45nm GPDK Standard cell library ( slow_vdd1v0_basicCells.lib ) All tools were run on a remote lab server at IIT Bhubaneswar ( vlsiws24 ), running Red Hat Enterprise Linux 7.9, accessed via a Linux terminal. These are not free or open-source tools โ a valid Cadence license is required. Directory Structure .
โโโ rtl/
โ โโโ riscv.v Single-cycle RV32I processor (RTL source)
โ โโโ riscv_tb.v Testbench for simulation
โโโ constraint/
โ โโโ constraint.sdc SDC timing constraints (50 MHz clock)
โโโ synthesis/
โ โโโ syn.tcl Genus synthesis script
โ โโโ riscv_net.v Synthesized gate-level netlist [output]
โ โโโ constraint_out.sdc Back-annotated constraints [output]
โ โโโ timing.rpt Setup/hold slack report [output]
โ โโโ area.rpt Cell count and area report [output]
โ โโโ power.rpt Power estimate report [output]
โ โโโ gates.rpt Gate-type breakdown report [output]
โโโ gls/
โ โโโ riscv.vcd Waveform dump from GLS run [output]
โโโ docs/
โ โโโ timing_report.png Timing report screenshot
โ โโโ area_report.png Area report screenshot
โ โโโ power_report.png Power report screenshot
โ โโโ gates_report.png Gates report screenshot
โโโ lib/
โ โโโ README.md Library placeholder and instructions
โโโ README.md Files marked [output] are generated automatically when you run the scripts โ you do not need to create them by hand. Note: The standard cell library files ( .lib and .v model files) are not included in this repository. They contain proprietary foundry IP distributed under a lab license agreement. See Setup Instructions for how to supply them. Flow Overview This is the full sequence from start to finish: Write the RTL โ riscv.v implements a single-cycle RV32I core supporting all six RISC-V instruction types (R, I, S, B, U, J). Define timing constraints โ constraint.sdc specifies a 20 ns clock (50 MHz), input/output delays, clock quality parameters, and a false path for the asynchronous reset. Write the synthesis script โ syn.tcl tells Genus where to find the RTL, library, and constraints, and what commands to run. Run synthesis in three passes: syn_generic โ converts RTL to technology-independent logic syn_map โ maps generic logic onto real 45nm library cells syn_opt โ optimizes the mapped netlist for timing and area Export reports โ timing, area, power, and gate count written to synthesis/ . Set up GLS โ copy the synthesized netlist and cell simulation models into gls/ . Run gate-level simulation with ncverilog to confirm correct behaviour. View waveforms in SimVision to visually verify the processor (e.g. pc_out incrementing by 4 each cycle after reset). Setup Instructions 1. Clone this repository git clone https://github.com/SansidMishra/RISCV-Synthesis.git cd RISCV-Synthesis 2. Supply the foundry library files These are not included in this repository. Obtain them from your lab or foundry license: mkdir -p lib
cp /path/to/slow_vdd1v0_basicCells.lib lib/
cp /path/to/slow_vdd1v0_basicCells.v gls/ If you do not know where they are installed on your system: find /cad/FOUNDRY/digital/45nm -name " *.lib " 2> /dev/null | head -20 Confirm the library contains real logic cells (not just filler/decap cells): grep " cell ( " lib/slow_vdd1v0_basicCells.lib | head -10 You should see cell names like INV_X1 , NAND2_X1 , DFF_X1 , etc. If you only see filler or decap cells, you have the wrong library file. 3. Load the Cadence tool environment source /cad/cshrc This sets up the tool paths and license variables. You must run this in every new terminal session before using any Cadence tool. Running Synthesis cd synthesis
genus -f syn.tcl | & tee synthesis.log tee synthesis.log saves a copy of all terminal output to a log file โ useful for debugging if the window closes unexpectedly. Milestone messages to watch for As synthesis runs, these lines confirm each stage completed successfully: Info : Done synthesizing. [SYNTH-2]
Done synthesizing 'riscv' to generic gates. โ syn_generic complete
Info : Done incrementally optimizing. [SYNTH-8]
Done incrementally optimizing 'riscv'. โ syn_opt complete If you do not see these, something went wrong. Scroll up in the log to find the first Error line, or consult the Common Errors and Fixes table. Reading the Reports After synthesis, four report files are written to synthesis/ . View any of them with: cat timing.rpt
cat area.rpt
cat power.rpt
cat gates.rpt Timing Report ( timing.rpt ) The most important report. Tells you whether the circuit meets your target clock speed. slack (MET) : 16.057 โ Timing is satisfied โ circuit works at 50 MHz
slack (VIOLATED) : -1.230 โ Timing violation โ circuit is too slow If timing.rpt is empty (0 bytes), it usually means some ports were not properly constrained. Diagnose inside Genus with: report_timing -lint Area Report ( area.rpt ) Field Meaning Cell Count Total number of standard cells used Cell Area Area occupied by logic cells Net Area Interconnect area (typically 0 before place-and-route) Total Area Cell Area + Net Area Power Report ( power.rpt ) Field Meaning Leakage Power Power consumed when circuit is idle Dynamic Power Power consumed due to signal switching Total Power Sum of leakage and dynamic A full RISC-V processor will show significantly higher dynamic power than a simple counter design due to the larger number of switching gates per clock cycle. Gates Report ( gates.rpt ) Lists which standard cell types were used and how many of each โ useful for understanding how the synthesizer implemented your design. Running Gate-Level Simulation (GLS) GLS verifies that the synthesized netlist โ the actual gates โ behaves identically to the original RTL. 1. Copy required files into gls/ cd gls
cp ../synthesis/riscv_net.v . cp /cad/FOUNDRY/digital/45nm/svt/verilog/slow_vdd1v0_basicCells.v . 2. Run the simulation source /cad/cshrc
ncverilog slow_vdd1v0_basicCells.v riscv_net.v ../rtl/riscv_tb.v +access+r Important: File order matters. Always list the cell model file first, then the netlist, then the testbench. Reversing this order causes "module not found" errors because the simulator tries to instantiate cells before they are defined. +access+r enables read access to internal signals, which is required for waveform dumping. 3. Expected output A successful simulation ends with: Simulation complete via $finish(1) at time 540 NS You will also see per-cycle $monitor output and a final PASS/FAIL summary from the testbench. Viewing Waveforms Once riscv.vcd is generated: simvision riscv.vcd & Steps inside SimVision If a database is already open and locked, reopen it from the console: database open -overwrite riscv.vcd In the Design Browser panel (left side), click the triangle next to the top module to expand the hierarchy. Click on riscv_tb โ its signals ( clk , rst_n , pc_out , etc.) appear in the panel below. Select all signals with Ctrl+A , then click "Send to Waveform Viewer" . Press F (or go to View โ Zoom โ Fit ) to display the full simulation timeline. What to expect For a correctly working single-cycle RISC-V core running NOPs, pc_out should increment by 4 every clock cycle once rst_n goes high: Time 0 โ pc_out = 0x00000000 (reset active)
Time 40ns โ pc_out = 0x00000004 (first instruction fetched)
Time 60ns โ pc_out = 0x00000008
Time 80ns โ pc_out = 0x0000000C
... This is correct because RISC-V instructions are 4 bytes wide and a single-cycle processor advances the program counter by one instruction per clock cycle. If no expand arrow appears next to the top module in the Design Browser, the VCD only captured a flat scope. Check the $dumpvars line in the testbench, delete the old .vcd file, and re-run the simulation. Results Summary Metric Value Clock Period 20 ns (50 MHz) Worst Slack +16.057 ns (MET) Critical Path Delay 3.8 ns Total Standard Cells 88 Total Cell Area 283.860 units Sequential Cells (FF) 30 (65.3% of area) Logic Cells 57 (34.5% of area) Leakage Power 5.440 nW Dynamic Power 9368.974 nW Total Power 9374.414 nW (~9.37 ยตW) GLS Result Pending Report Screenshots Timing Report Area Report Power Report Gates Report Known Limitations These are design decisions made intentionally for simplicity at this stage of learning. They would need to be addressed before any real tapeout. Branch condition logic is simplified โ the current RTL checks only alu_zero for all branch types. Full correctness requires separate condition checks for BNE, BLT, BGE, BLTU, and BGEU. LUI implementation โ the ALU operand mux for LUI uses imm_u correctly only when alu_src is also set in the control unit. Verify this is set correctly in control_unit.v . No pipeline โ this is a single-cycle design. Every instruction takes exactly one clock cycle regardless of complexity. No hazard detection โ since it is single-cycle, there are no data or control hazards to handle. No exception handling โ illegal instructions and misaligned memory accesses are not handled. Fixed memory size โ instruction memory is 256 words (1 KiB) and data memory is 1 KiB. These are sufficient for simulation but not for real programs. Hold timing not fully constrained โ verify that constraint.sdc includes set_input_delay -min and set_output_delay -min for complete hold analysis. Common Errors and Fixes Error Cause Fix Multiple designs are available. Specify the design you want to use. read_sdc or report_* run without a current design selected Add set_db design:riscv .current 1 before read_sdc , or use current_design riscv Cannot perform synthesis โ libraries have no usable inverters. [LBR-171] The .lib file has no real logic cells (filler-only or decap-only library) Confirm with grep "cell (" library.lib โ you need cells like INV, NAND, DFF An option named '-file' could not be found. Genus 18.10 does not support write_hdl -file Use shell redirection: write_hdl > riscv_net.v cp: cannot create regular file: No such file or directory Target directory does not exist Run mkdir -p
Cadence Virtuoso Schematics & Spectre Waveforms
Verified circuit schematics, physical layout topologies, and transient simulation waveforms.
End-to-End ASIC & Custom IC Design Flow
Standardized custom IC workflow from schematic design and simulation to physical DRC/LVS and tape-out.
(Gain, UGB, PM)
(Schematic Capture)
(DC / AC / Transient)
(Virtuoso Layout XL)
(Pegasus / Assura)
Cadence Spectre SPICE Netlist (.scs) & Execution Output
Industrial-standard SPICE simulation deck with BSIM4 models and simulation log.
Spectre (R) Circuit Simulator -- Version 21.1.0.612.isr13 64bit Compiled on May 12 2024 at 18:24:10 โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ [CIRCUIT NETLIST] Reading PDK: /tools/pdk/gpdk180/models/spectre/gpdk180.scs [DC OPERATION] Convergence achieved in 5 iterations: - M1/M2 Vds = 0.842 V | gm = 1.84 mS | Vgs - Vth = 142 mV (Sat) - Quiescent Current Idd = 420.5 uA | Total Power = 756.9 uW [AC ANALYSIS] Open-Loop Gain A0: 72.4 dB (Target > 65 dB) [PASSED] [AC ANALYSIS] Unity-Gain Bandwidth: 54.2 MHz (CL = 10 pF) [AC ANALYSIS] Phase Margin (PM): 65.4ยฐ (Target > 60ยฐ) [STABLE] [TRANSIENT] Slew Rate (SR+ / SR-): 48.2 V/us | Settling Time: 18.4 ns [DRC / LVS] Pegasus LVS: Circuits are EQUIVALENT | 0 Errors, 0 Warnings โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ [โ] Cadence Spectre Run Complete: CPU time = 2.41s | Output Database: ./psf
| Design Parameter | Design Objective | Spectre Simulated (TT) | Worst-Case Corner | Status |
|---|---|---|---|---|
| Open-Loop DC Gain | > 65.0 dB | 72.4 dB | 68.1 dB (SS) | PASSED |
| Unity-Gain Bandwidth (UGB) | > 40.0 MHz | 54.2 MHz | 48.5 MHz (SF) | PASSED |
| Phase Margin (PM) | > 60.0ยฐ | 65.4ยฐ | 61.2ยฐ (FS) | STABLE |
| Total Power Dissipation | < 1.0 mW | 756.9 uW | 910.4 uW (FF) | OPTIMAL |
| Slew Rate (SR) | > 35 V/us | 48.2 V/us | 42.0 V/us (SS) | PASSED |