Skip to content

jangam-ajay/VLSI_PD_ORCA_TOP

Repository files navigation

✋ Hi, My name is AJAY JANGAM

🎓 Electronics & Communication Engineering Student | VLSI & Physical Design Enthusiast

💡 Passionate about backend VLSI, RTL to GDSII, and physical design automation

🚀 Hands-on with Synopsys ICC2, Design Compiler, PrimeTime, Cadence Innovus for 28nm PnR flows





🔗PROJECTS

PD_ORCA_TOP_28nm

Block-level multi-voltage ASIC physical design at 28nm with 52K std cells & 40 macros. Full RTL-to-signoff flow for multi-power domains & 5 clocks, with TCL automation showcasing senior PD skills.

🎯 Project Overview

PD_ORCA_TOP_28nm implements complete physical design for a multi-voltage ASIC block in 28nm node:

  • 52K std cells, 40 macros
  • 2 power domains
  • 4 clocks: Propagated CLK , Generated clocks - 3 ,Virtual clocks - 2
  • 75% utilization, zero DRC/LVS/timing violations

Theory: Block-level PD targets 65-80% util for routability (congestion <10% overflow). Multi-voltage uses create_voltage_area for isolation. CTS exceptions (nonstop/exclude/stop) ensure skew <50ps on generated clocks.[web:14][web:17]

🛠️ Tools Stack

Tool Purpose
ICC2 PnR
DC Synthesis
STARRC Spef
PrimeTime STA/DMSA

📋 PD Flow & Theory

1. Floorplanning (Core Sizing & Macro Placement)

  • Theory: Aspect ratio 1:1~1.5; util 70-80% leaves routing halo. Macro abutting reduces white-space congestion by 30-40%.
  • TCL: Auto voltage areas aligned to rows.

2. Placement (Global/Detailed Opt)

  • Theory: Path grouping fixes high-depth logic (e.g., I_BLENDER ~40 gates → setup violations via OCV derating).
  • Iterations via fly-lines → zero hotspots.

3. CTS (Clock Tree Synthesis)

  • Theory: Balance propagated/generated clocks; nonstop pins trace thru dividers, exclude test sinks. Target: Balancing the skew
  • Exceptions: don't_touch spines, float pre-routed nets.

4. Routing & ECO

  • Theory: NDR (2x width/spacing) for CLK; track-aware routing at 28nm avoids via shorts.

5. Signoff (STA/TPA)

  • Theory: DMSA models multi-scenario (RC-corner, OCV/AOCV); hold fixes via sizing/del-insert.

🚀 Challenges Solved

  • Voltage Planning: TCL calc w/ margin → no area shortage.
  • Congestion: Macro iters + fly-analysis → routable.
  • Timing: Grouping + opt → clean slacks.

✅ Results Highlights

  • Utilization: 75% (optimal density)
  • Timing: Setup/hold closure
  • Power: Met targets per domain
  • Congestion: Zero overflow

About me

Trained Physical Design Engineer in VLSI , Seeeking opportunity in semiconductor industry to contribute my knowlege and skills.

About

PD_ORCA_TOP_28nm demonstrated a complete block-level ASIC PnR flow using Synopsys ICC2, transforming gate-level netlist to GDSII across floorplanning, power planning, placement, CTS, routing, and signoff stages. The design contains 52K standard cells and 40 macros within a rectangular core at 75% utilization

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors