NXCompare : Cross Format Layout/Mask Data Equivalence Checker

NxCompare is a fast and versatile comparison tool that can be used to compare any two Layout or Mask databases (Layout vs Layout, Layout vs Mask or Mask vs Mask). NxCompare is a distributed processing application based on SoftJin’s Nirmaan software development platform.

Key Benefits
  • Better price and performance compared to standard DRC tools for comparison of large, crossformat layout/mask databases
  • The most versatile Layout/Mask Data Equivalence Checker
  • Cross format comparison supports all major industry standard Layout and Mask data standards
  • Very fast turnaround time for large, cross format databases
  • Employs special partitioning scheme suitable for parallel processing of inherently flat data
  • Affordable distributed computing on low cost Linux machines
  • Availability
    • 32/64 bit on Linux

Key Features
  • Unlimited data handling capacity
    • Can handle equivalent of hundreds of GBs of GDSII data
    • No restrictions on design hierarchy – handles a mix of flat and hierarchical designs
  • Fast turn-around time using distributed processing on inexpensive compute grids
    • Compares multiple GBs per minute of GDSII equivalent data
  • Cross format comparison support for multiple layout and mask data standards
    • Layout data formats - GDSII, OASIS and OpenAccess
    • E-beam formats - MEBES, OASIS.VSB, VSB-11/12 and JEOL
    • Customized EBDW formats
    • Bitmap data (image data)
  • Rich feature set
    • In-built Boolean functions required for layout equivalence checking of geometrical layers
    • User defined tolerance limit to filter out nuisance errors
    • De-staircasing function to filter out differences caused due to approximation of all-angle geometries in pre-fracture data
    • Scaling, rotation, translation, mirroring and tone reversal
    • Full-chip, Window-specific, Cell-specific, Layer-specific equivalence checking
    • TCL scripting allows quick customization and integration in to existing design and verification flows
  • Supports multiple error database formats for popular Viewers to efficiently navigate and view and analyze the differences
Use Scenarios
NxCompare could be used very effectively in a variety of use scenarios including:
  • Quick equivalence checking of layout databases in the back-end design flow by allowing layout comparisons at different levels of abstractions
    • Full-chip, Window-specific, Cell-specific, Layer-specific comparison
  • Mask Data Flow
    • Independent verification of fracturing results (in formats for Mask Writer or Mask Inspection) with original layout data (in GDSII/OASIS/OpenAccess formats)
      • De-slivering and de-staircasing allow separating the expected differences from unexpected differences
    • Comparing EB data after re-fracturing / format conversion
    • Overlaying and Comparison of Mask Inspection / SEM image data with original Layout
  • Comparison of split data files with merged / original data
    • PSM data, Double patterning
  • Qualification / regression testing of new post-layout tools and internally developed tools and utilities
  • Comparing geometrical databases from different sources
    • Databases generated by different tools
    • Databases in different data formats
    • Databases in different versions of similar data formats
High Speed and Capacity handling
As compared to conventional DRC tools, NxCompare provides better price and performance alternative as it has been specifically designed to handle comparison of large, cross format databases using heterogeneous low-cost workstations.
  • NxCompare employs special partitioning and data access schemes suitable for parallel processing of inherently flat huge geometrical data. For example, for comparing a 138 GB GDSII file with a 13 GB OASIS file, NxCompare took less than 80 minutes on a 1Gpbs network with 16 CPUs, with each CPU (Intel Core2 1.86GHz, 2MB cache) consuming less than 1 GB memory. This is equivalent to processing around 3-4 GB of flat GDSII data per minute. This illustrates speed and capacity handling of NxCompare on low-cost compute environment.
  • For the same example, NxCompare’s has been benchmarked at processing of 12 GB per minute of equivalent flat GDSII data on a high-performance linux cluster having 32 CPUs connected over infiniband with each CPU having 16 GB memory.
  • NxCompare’s Distributed Processing Engine supports Multi-core, Multi-CPU workstations or a network of heterogeneous CPUs with heterogeneous operating systems. NxCompare scales up well on a heterogeneous network of machines as it is designed essentially keeping distributed processing in mind.
Designed by The Scribble