SmartOpt An Industrial Strength Framework for Logic Synthesis

Slides:



Advertisements
Similar presentations
Address comments to FPGA Area Reduction by Multi-Output Sequential Resynthesis Yu Hu 1, Victor Shih 2, Rupak Majumdar 2 and Lei He 1 1.
Advertisements

Spartan-3 FPGA HDL Coding Techniques
Simulation of Fracturable LUTs
Reducing the Pressure on Routing Resources of FPGAs with Generic Logic Chains Hadi P. Afshar Joint work with: Grace Zgheib, Philip Brisk and Paolo Ienne.
DAG-Aware AIG Rewriting Alan Mishchenko, Satrajit Chatterjee, Robert Brayton Department of EECS, University of California Berkeley Presented by Rozana.
 Y. Hu, V. Shih, R. Majumdar and L. He, “Exploiting Symmetries to Speedup SAT-based Boolean Matching for Logic Synthesis of FPGAs”, TCAD  Y. Hu,
FPGA Technology Mapping. 2 Technology mapping:  Implements the optimized nodes of the Boolean network to the target device library.  For FPGA, library.
ISE. Tatjana Petrovic 249/982/22 ISE software tools ISE is Xilinx software design tools that concentrate on delivering you the most productivity available.
1 Alan Mishchenko UC Berkeley Implementation of Industrial FPGA Synthesis Flow Revisited.
J. Christiansen, CERN - EP/MIC
FORMAL VERIFICATION OF ADVANCED SYNTHESIS OPTIMIZATIONS Anant Kumar Jain Pradish Mathews Mike Mahar.
Combinational and Sequential Mapping with Priority Cuts Alan Mishchenko Sungmin Cho Satrajit Chatterjee Robert Brayton UC Berkeley.
ABC: A System for Sequential Synthesis and Verification BVSRC Berkeley Verification and Synthesis Research Center Robert Brayton, Niklas Een, Alan Mishchenko,
Cut-Based Inductive Invariant Computation Michael Case 1,2 Alan Mishchenko 1 Robert Brayton 1 Robert Brayton 1 1 UC Berkeley 2 IBM Systems and Technology.
1 Synthesizing Datapath Circuits for FPGAs With Emphasis on Area Minimization Andy Ye, David Lewis, Jonathan Rose Department of Electrical and Computer.
1 Stephen Jang Kevin Chung Xilinx Inc. Alan Mishchenko Robert Brayton UC Berkeley Power Optimization Toolbox for Logic Synthesis and Mapping.
Give qualifications of instructors: DAP
1 Area-Efficient FPGA Logic Elements: Architecture and Synthesis Jason Anderson and Qiang Wang 1 IEEE/ACM ASP-DAC Yokohama, Japan January 26-28,
FPGA Logic Cluster Design Dr. Philip Brisk Department of Computer Science and Engineering University of California, Riverside CS 223.
1 WireMap FPGA Technology Mapping for Improved Routability Stephen Jang, Xilinx Inc. Billy Chan, Xilinx Inc. Kevin Chung, Xilinx Inc. Alan Mishchenko,
A Semi-Canonical Form for Sequential Circuits Alan Mishchenko Niklas Een Robert Brayton UC Berkeley Michael Case Pankaj Chauhan Nikhil Sharma Calypto Design.
Resource Sharing in LegUp. Resource Sharing in High Level Synthesis Resource Sharing is a well-known technique in HLS to reduce circuit area by sharing.
Global Delay Optimization using Structural Choices Alan Mishchenko Robert Brayton UC Berkeley Stephen Jang Xilinx Inc.
Sequential Equivalence Checking for Clock-Gated Circuits Hamid Savoj Robert Brayton Niklas Een Alan Mishchenko Department of EECS University of California,
Reducing Structural Bias in Technology Mapping
Time-borrowing platform in the Xilinx UltraScale+ family of FPGAs and MPSoCs Ilya Ganusov, Benjamin Devlin.
Floating-Point FPGA (FPFPGA)
A New Logic Synthesis, ExorBDS
Topics SRAM-based FPGA fabrics: Xilinx. Altera..
Synthesis for Verification
XILINX FPGAs Xilinx lunched first commercial FPGA XC2000 in 1985
Technology Mapping into General Programmable Cells
Automated Extra Pipeline Analysis of Applications mapped to Xilinx UltraScale+ FPGAs
Power Optimization Toolbox for Logic Synthesis and Mapping
Mapping into LUT Structures
Delay Optimization using SOP Balancing
Alan Mishchenko Department of EECS UC Berkeley
New Directions in the Development of ABC
Topics The logic design process..
Alan Mishchenko Satrajit Chatterjee Robert Brayton UC Berkeley
Logic Synthesis Primer
Magic An Industrial-Strength Logic Optimization, Technology Mapping, and Formal Verification System Alan Mishchenko UC Berkeley.
A Semi-Canonical Form for Sequential AIGs
Versatile SAT-based Remapping for Standard Cells
Integrating an AIG Package, Simulator, and SAT Solver
Topics HDL coding for synthesis. Verilog. VHDL..
Synthesis for Verification
Designing Area-Efficient Dividers and Square Rooters
Standard-Cell Mapping Revisited
LUT Structure for Delay: Cluster or Cascade?
Timing Optimization Andreas Kuehlmann
SAT-Based Area Recovery in Technology Mapping
Polynomial Construction for Arithmetic Circuits
Alan Mishchenko University of California, Berkeley
SAT-Based Optimization with Don’t-Cares Revisited
Scalable and Scalably-Verifiable Sequential Synthesis
Mapping into LUT Structures
FPGA Glitch Power Analysis and Reduction
Integrating Logic Synthesis, Technology Mapping, and Retiming
Integrating an AIG Package, Simulator, and SAT Solver
Improvements in FPGA Technology Mapping
Recording Synthesis History for Sequential Verification
Delay Optimization using SOP Balancing
Magic An Industrial-Strength Logic Optimization, Technology Mapping, and Formal Verification System Alan Mishchenko UC Berkeley.
Innovative Sequential Synthesis and Verification
Robert Brayton Alan Mishchenko Niklas Een
Word-Level Aspects of ABC
Fast Min-Register Retiming Through Binary Max-Flow
Robert Brayton Alan Mishchenko Niklas Een
Alan Mishchenko Robert Brayton UC Berkeley
Presentation transcript:

SmartOpt An Industrial Strength Framework for Logic Synthesis Stephen Jang, Xilinx Inc. Dennis Wu, Xilinx Inc. Mark Jarvin, Xilinx Inc. Billy Chan, Xilinx Inc. Kevin Chung, Xilinx Inc. Alan Mishchenko, UC Berkeley Robert Brayton, UC Berkeley

Example Industrial Circuit Toy synthesized netlist created for Xilinx FPGA with Xilinx MUXFx Xilinx Confidential

Example Industrial Circuit MUXFx is dedicated hardware that multiplexes the outputs of two adjacent LUTs Xilinx Confidential

Example Industrial Circuit MUXFx is dedicated hardware that multiplexes the outputs of two adjacent LUTs Xilinx Confidential

Modeling the Example Xilinx Confidential

Modeling the Example Xilinx Confidential

Modeling the Example .model MUXFx .inputs I0 I1 S .outputs O .attrib comb black box .end Xilinx Confidential

Modeling the Example .model MUXFx .inputs I0 I1 S .outputs O .attrib comb black box .end No information flow between inputs and outputs of MUXFx Xilinx Confidential

Modeling the Example .model MUXFx .inputs I0 I1 S .outputs O .attrib comb black box .end No information flow between inputs and outputs of MUXFx Box I/Os equivalent to PIs/POs Xilinx Confidential

Modeling the Example Xilinx Confidential

Modeling the Example Pin delays added to model .model MUXFx .inputs I0 I1 S .outputs O .attrib comb black box .delay I0 O 0.5 .delay I1 O 0.5 .delay S O 0.1 .end Pin delays added to model Xilinx Confidential

Modeling the Example Pin delays added to model Better timing analysis .model MUXFx .inputs I0 I1 S .outputs O .attrib comb black box .delay I0 O 0.5 .delay I1 O 0.5 .delay S O 0.1 .end Pin delays added to model Better timing analysis Still can’t exploit constant Xilinx Confidential

Modeling the Example Xilinx Confidential

Modeling the Example Model MUXFx functionality .model MUXFx .inputs I0 I1 S .outputs O .attrib comb white box .delay I0 O 0.5 .delay I1 O 0.5 .delay S O 0.1 .names IO I1 S O 1-0 1 -11 1 .end Model MUXFx functionality Xilinx Confidential

Modeling the Example Model MUXFx functionality Trim unused logic .inputs I0 I1 S .outputs O .attrib comb white box .delay I0 O 0.5 .delay I1 O 0.5 .delay S O 0.1 .names IO I1 S O 1-0 1 -11 1 .end Model MUXFx functionality Trim unused logic IP designer’s intent preserved by white box Xilinx Confidential

Results with different models Quality of results obtained by ABC scale with information carried in the integeration model Compared to the reference flow without ABC, using ABC on the fractured netlist, results in 2.5% LUT count reduction and 0.7% design performance improvement. This improvement is due to ABC’s combinational optimization and mapping. Allowing timing information to flow for the entire netlist, ABC’s combinational synthesis is able to improve Fmax an additional 1% and to further reduce LUTs by 0.6% by applying area recovery to the non-critical paths. Allowing timing and functional information to flow for the entire netlist, ABC is able to greatly reduce the # of LUTs and FFs through it sequential optimizations. Xilinx Confidential

Conclusion Introduced BLIF extensions to allowing timing and logical information for the entire circuit to be visible to ABC Industrial design quality of results scale with modeling detail presented to ABC Black Box models: reduce LUTs by 2.5% and improve speed by 0.7% vs. non-ABC flow Box with timing info: reduce LUTs by 3.1% and improve speed by 1.7% SmartOpt (boxes with timing and function information): reduce LUTs by 8.3% and FFs by 7.8% plus improve speed by 2.1% Xilinx Confidential