Storaasli 5/9/03 Analytical and Computational Methods Computing Faster without CPUs Scientific Applications on FPGA-based* Reconfigurable Hypercomputers by Dr. Olaf Storaasli Analytical & Computational Methods Branch Structures and Materials for May Seminar, Electronic Systems Branch * Field-Programmable Gate Array

Storaasli 5/9/03 Analytical and Computational Methods 2 NASA Research Background FEA: NASTRAN Viking ==> Mars IPAD: Integrated Design Finite Element Machine: Early // Computer Cray GigaFLOP Award: Shuttle SRB Matrix Equation Solver: FORTRAN, C, Java Lanczos Eigensolver: 88x Speedup Intel: Supercomputer Users Board, P6 Award Symposia: Large-Scale Apps. (5) NASA Software of the Year Award Creativity & Innovation Awards NASA Fellowship: Norway

Storaasli 5/9/03 Analytical and Computational Methods 3 Exploring Scientific Applications on Reconfigurable Hypercomputers ‘02‘03 Creativity & Innovation 62K gates/FPGA 6M gates/FPGA

Storaasli 5/9/03 Analytical and Computational Methods 4 Computing Faster Without CPUs GOAL: Evaluate FPGA*-based Hypercomputer Potential for NASA Scientific Computations TEAM: Dr. Olaf Storaasli, Principal Investigator PARTNERS: Star Bridge Systems, NSA, USAF, MSFC, William Fithian-Harvard, Siddhartha Krishnamurthy-VT Shaun Foley-MIT, Cris Kania-GS, Neha Dandawate-GS, Patrick Butler-VT Kristin Barr-JPMorgan, Robert Lewis-Morehouse, Vincent Vance-VT Jarek Sobieski, Robert Singleterry, Dave Rutishauser, Joe Rehder Garry Qualls

Storaasli 5/9/03 Analytical and Computational Methods 5 William Fithian* (Harvard, Merit Scholar, Oracle Award) * NASA-NHGS mentorship ‘00-’02

Storaasli 5/9/03 Analytical and Computational Methods 6 First Langley Hypercomputers 10 FPGAs each

Storaasli 5/9/03 Analytical and Computational Methods 7

Storaasli 5/9/03 Analytical and Computational Methods 8 FPGA Programming User controls gates: middle man removed Code options: –1-D Text, sequential FORTRAN-like: C-to-Gate, VHDL parallelism esoteric –3-D Graphic, parallel drag & drop: Viva Parallelism inherent data flow like analog computer NASA Hypercomputers

Storaasli 5/9/03 Analytical and Computational Methods 9 FPGA: New Computing Paradigm Traditional CPU Gateware: VIVA Icons & Transports 26 MFLOPS /250 MHz SGI Reconfigurable FPGA Sequential: 1 operation/cycle Fixed gates & data types Wasteful: 99% gates idle/cycle yet all draw power Software: Text - 1D do i = 1, billion c= a+b end do Parallel: Inherent Dynamic gates & data types Efficient: Optimizes gates to task 392+ MFLOPS /64 MHz FPGA GFLOPS /10 FPGA board

Storaasli 5/9/03 Analytical and Computational Methods 10 Select-Drag-Drop to Code “icon” Primitives Add new code to library Complex algorithms “drill in”

Storaasli 5/9/03 Analytical and Computational Methods 11 VIVA:Custom Chip Design Gateware What: Graphics tool to “route” FPGAs (VHDL cumbersome) Growth in VIVA Capability Extensive Data Types Trig, Logs, Transcendentals File Input/Output Vector-Matrix Support Access to Multiple FPGAs Extensive Documentation Stable Development Few “bugs” NO Floating Point NO Scientific Functions NO File Input/Output NO Vector-Matrix Support Access to One FPGA Primitive Documentation Weekly Changes Frequent “bugs” VIVA 1 (Feb ’01) VIVA2 (July ’02) How: Converts icon-transport “gateware” to circuit logic Why: Achieve near-ASIC speed (w/o chip design $)

Storaasli 5/9/03 Analytical and Computational Methods 12

Storaasli 5/9/03 Analytical and Computational Methods 13 Viva User1 Viva User2 Viva User3 Viva User4 Viva User5 Viva User6 Windows Server HC-38m Hypercomputer 7 FPGAs 6M Gates Parallel Use of Parallel FPGAs

Storaasli 5/9/03 Analytical and Computational Methods 14 Algorithms Developed* * In AIAA & Military & Aerospace Programmable Logic Device (MAPLD) papers. n! => Probability: Combinations/Permutations AirSC Cordic => Transcendentals: sin, log, exp, cosh…  ∂y/∂x & ∫ f(x)dx => Runge-Kutta: CFD, Newmark Beta: CSM  Matrix Equation Solver : [A]{x} = {b} - Gauss & Jacobi Nonlinear Analysis : Analog simulation avoids NLT devp’t time  Matrix Algebra : {V}, [M], {V} T {V}, [M]x[M],GCD,… Dynamic Analysis : [M]{ ü } + [C]{u} + [K]{u} + NLT = {P(t)} Analog Computing : digital accuracy

Storaasli 5/9/03 Analytical and Computational Methods 15 Numeric Integration f(x)=x 2 f(x)*  x xx  f(x)*  x x i+1 = x i +  x Control Output (Area under curve) f(x) x xx

Storaasli 5/9/03 Analytical and Computational Methods 16 VIVA Sparse Matrix Equation Solver Jacobi Iterative (3x3 Demo) Control 3 Row Loads 3 // Dot Products [A]{x}={b} x 1 = 1/A 11 *(b 1 - A 12 *x 2 - A 13 *x 3 )

Storaasli 5/9/03 Analytical and Computational Methods 17 z b P a x - / y d M e x - / z y y i = (P - b z i-1 )/a z i = (M - dy i-1 )/e - initialized output y output z Analog Diagram of 2x2 Equations Solution input bz P-bz bz

Storaasli 5/9/03 Analytical and Computational Methods 18 Fixed-Point Iteration: VIVA Diagram

Storaasli 5/9/03 Analytical and Computational Methods 19 Year 2: Exploit Latest FPGAs Plans: - Millions of Matrix Equations: Structures, Electromagnetics & Acoustics - Rapid Static & Dynamic Structural Analyses - Cray Vector Computations in Weather Code (VT PhD) - Robert on Administrator’s Fellowship at Star Bridge Systems - Simulate advanced computing concepts using VIVA - Collaborate with SBS, NSA, A&T… to expand VIVA libraries - Tailor VIVA development for NASA applications - Target applications to NASA programs (e.g. EDB Collaboration??) Rapid Growth in FPGA Capability FPGA (Feb ’01) FPGA (Oct ’02) Xilinx FPGA Gates Multiplies on chip Clock Speed MHz Memory on chip Memory Speed Reconfigure Time GFLOPS Total GFLOPs XC K Kb 466 Gb/s 100ms (10 FPGAs) XC2V million (97x) (3x) 3.5 Mb (175x) 5 Tb/s (11x) 40ms (2.5x) 47 (120x) 329 (7 FPGAs)

Storaasli 5/9/03 Analytical and Computational Methods 20 Summary What We’re Learning We like FPGA promise – accomplished much Hardware: Testing 3 futuristic FPGA systems FPGAs: Inherently //, flexible, efficient, & fast, dramatic advances VIVA: Powerful & growing (tailor to NASA needs) Applications: ‘02 - Diverse “pathfinder” algorithms developed Speed: Year 1: 4 GFLOPS => Year 2: 329 GFLOPS Future: exploit capability on NASA “cutting edge” innovations ‘03 - Comprehensive NASA engineering applications

Storaasli 5/9/03 Analytical and Computational Methods 21 Langley Reconfigurable Computing Research 1. Singleterry, Robert C., Jaroslav Sobieszczanski-Sobieski, and Samuel Brown. “Field-Programmable Gate Array Computer in Structural Analysis: an Initial Exploration.” 43 rd American Institute of Aeronautics and Astronautics (AIAA) Structures, Structural Dynamics, and Materials Conference. April 22-25, Storaasli, Olaf O., Robert C. Singleterry, and Samuel Brown. “Scientific Computations on a NASA Reconfigurable Hypercomputer.” Abstract accepted for 5 th Military and Aerospace Programmable Logic Devices (MAPLD) Conference, Paper in preparation. September 10-12, Fithian, William, Samuel Brown, and Tyler Reed. “Object Synchronization in VIVA 1.5.” Briefing prepared for VIVA users at NASA Marshall, Eglin AFB, Progress Forge, Inc., and Star Bridge Systems, Inc. March 26, Barr, Kristen, Shaun Foley, and Robert A. Lewis II. “Hypercomputing with the CORDIC Algorithm.” August, Presentation of research conducted under Dr. Olaf O. Storaasli, June-August, Butler, Patrick. New Horizons Governors School Mentorship Project. May, Presentation of research conducted under Dr. Olaf O. Storaasli, September 2000 – May Dandawate, Neha. “Reckless Speeding: The Investigation of the Programming Capabilities of the HAL Hypercomputer.” July, Presentation of research conducted under Dr. Olaf O. Storaasli, June – July, Dandawate, Neha. “The Investigation of the Programming Capabilities of the HAL-15 Hypercomputer.” July, Paper on research conducted under Dr. Olaf O. Storaasli, June – July, Fithian, William. “Developing a Matrix Equation Solver for the HAL-15 Hypercomputer.” December, Proposal for research to be conducted under Dr. Olaf O. Storaasli, September 2001 – May Fithian, William. “Developing a Matrix Equation Solver for the HAL-15.” May, Presentation of research conducted under Dr. Olaf O. Storaasli, September 2001 – May Fithian, William. “Jacobi Iterative Matrix Equation Solver for Star Bridge Systems FPGA Hypercomputer.” September, In preparation. 11. Foley, Shaun. “Scientific Hypercomputing.” August, Paper describing research conducted under Dr. Olaf O. Storaasli, June – August, Krishnamurthy, Siddhartha. “Development of an Integration Algorithm for Field Programable Gate Arrays using VIVA.” July, Paper describing research conducted under Dr. Robert C. Singleterry, June – Aug Further Information: Google: “olaf acmb”