130 likes | 280 Views
Micron Technology Clinic The Algorithms, Tools, and Capabilities of HSimPlus. Presenter: Kristian Blomquist. HSimPlus Simulation Flow. System Coupled simulation is too complex for large circuits & power nets Phase 1 Decoupled circuit simulation is run to create ideal current sources Phase 2
E N D
Micron Technology ClinicThe Algorithms, Tools, and Capabilities of HSimPlus Presenter: Kristian Blomquist
System Coupled simulation is too complex for large circuits & power nets Phase 1 Decoupled circuit simulation is run to create ideal current sources Phase 2 Power net simulation with the current sources included Power Power Net Reliability Analysis (PWRA) Phase 1 IR drop calculated Phase 2 Node voltage and branch current Two Phase Analysis
Hierarchical Netlisting • Provides capability to simulate and analyze: • Large circuit blocks • Groups of large interacting circuit blocks • Full-chip designs • Reduced memory usage
Post Layout Acceleration (PLA) • Flat Netlist Vs. Hierarchical Netlist • Flat Netlist • PLA off • Hierarchical Netlist • PLA on • RC’s distributed throughout
Circuit layout with PLA Circuit layout without PLA
Newton Raphson equation andDC iteration values • Mixed signal simulator converts from digital to analog and vice versa • Example:
Precision • HSimPluscontrols precision and speed • HSIMPARPRECISION • HSIMRCPRECISION • HSIMMOSPRECISION • HSIMVPRECISION • HSIMIPRECISION
HSimPlus Output Precision High Vs. Low Green Line is Highest Precision Purple Line is Lowest Precision
Time Constraints Precision High Vs. Low • Simulation of 3 Bit Adder • 40 ns transient analysis
Monte Carlo Analysis • Simulates changes due to processing irregularities • Gaussian, Uniform, or Limited distribution • Each iteration involves DC matching
Monte Carlo OutputGaussian Distribution Thick Green Line is original Simulation Result
HSimPlus Conclusion • Two-Phase Analysis • Effective model order reduction • Hierarchical netlisting • Is efficient improvement • Precision Vs. Time • Highest precision fails to complete • Monte Carlo • Effective processing simulation