GE Intelligent Platforms AXISLib-AVX DSP & Math Libraries for Intel Processors with Advanced Vector Extensions (AVX) AXISLIB-AVX from GE Intelligent Platforms is Features AXIS Advanced Multiprocessor a family of high performance DSP and math Integrated Software • 600+ DSP and Vector Math Functions libraries that deliver world class performance AXISLIB DSP and Math libraries can be used • VSIPL API Core 1.0+ for the latest Intel 2nd generation Core i7 on their own, or within GE’s integrated • RSPL API for maximum performance multicore, multithreaded platforms with multiprocessor application development Advanced Vector Extensions (AVX). • Generic C libraries framework that includes AXISFLOW inter- • Development and production library builds processor communication (IPC) middleware Portability and AXISVIEW integrated GUI. These tools • AVX optimized functions The Vector Signal Image Processing (VSIPL) enable fast prototyping and application • Target Platforms: Intel 32 & 64bit multi- open standard application programing scaling across multiple CPUs, boards and threaded, multicore CPUs with AVX 256bit interface (API) facilitates code portability system fabrics. ™ wide SIMD extensions for Windows, across multiple CPU generations and archi- ™ Linux & VxWorks tectures to support technology refresh AXISView screen shot during the entire program life cycle. Performance GE’s RSPL API gives the programmer more control with lower CPU overheads to meet very challenging performance objectives. Benchmarks GE can supply performance benchmarks for a suite of common DSP functions on the latest, Intel, NVIDIA GPGPU and PowerPC platforms. GE single board computers (SBCs) and mul- Reduced cost of ownership tiprocessing boards leverage the latest high AXISLIB gets the best performance out performance computing (HPC) architectures of the deployed system without the need and switched fabrics onto rugged COTS to hand craft libraries for each processor form factors such as 3U & 6U OpenVPX. architecture thereby reducing project work These platforms allow system integrators load, cost of ownership and shortening time to move desk top and HPC applications into to solution. mission critical pay loads to meet expanded operational requirements for a range of Flexibility & performance tuning intelligence, surveillance and reconnais- sance (ISR) platforms. AXISLIB delivers more than 600 standard functions and our team of expert math- Typical applications include radar, sonar, ematicians and programmers can offer image processing, SIGINT, ELINT, EW and optimization services to meet the most counter measures for deployed airborne, demanding customer requirements. ground and naval platforms. DSP & Math Libraries for Intel Processors with Advanced Vector Extensions (AVX) Functionality Function Set Description Scalar Complex Scalar 40 functions for performing complex scalar math Index Scalar 4 functions for indexing matrix elements Random Number Generation Random Numbers 11 functions for generating random numbers, vectors and complex vectors Vector and Elementwise Operations Elementary Mathematical 42 functions performing elementary vector math (sin, cos, tan, atan, exp, log, sqrt, etc.) Unary Operations 48 functions for operating on a single vector or matrix Binary Operations 60 functions for operating either two vectors or matrices or one vector and a scalar Ternary Operations 24 functions for operations requiring three inputs Logical Operations 20 functions for performing logical operations on vectors or matrices Selection Operations 23 functions for selecting a subset of a vector or matrix Bitwise and Boolean Logical Operators 16 functions for performing Bitwise and Boolean operations on vectors and matrices Element Generation and Copy 40 functions for copying and generating vector elements Manipulation Operations 28 functions for vector and matrix manipulation (e.g. scatter, gather and swap) Signal Processing FFTs 42 functions for performing 1D and 2D FFTs (real-complex, complex-real, complex-complex in place and out-of-place) Windowing 4 windowing functions (Blackman, Hanning, Kaiser, Chebyshev) SBC624 and DSP280 are the first in a new generation Filter 8 functions for FIR filtering of rugged, high performance embedded computing Convolution 8 functions convolutions (1D & 2D) (HPEC) OpenVPX solutions that benefit from the Correlation 8 functions correlations (1D & 2D) new AXISLIB-AVX libraries bringing best-in-class Histogram 1 function histograming performance to SWaP sensitive defense and aerospace applications for a range if ISR missions. Linear Algebra Matrix and Vector Operations 75 functions for performing linear algebra on vectors and matrices Linear System Solvers 45 functions **Note: List does not account for all data type varieties of functions. Standard VSIPL management functions are not included in this listing. Ordering Information AXISLIB-AXV-01M Maintenance Agreement. Includes the right to use license and 1 x runtime license. Annually renewable. AXISLIB-AXV-01R Runtime license Performance Bench Marks: Sample Function Times (single thread unless stated ) * – Function Description Time in Microseconds vsip_ccfftip_f 1K complex-complex in-place FFT 2.7 vsip_ccfftip_f 128K complex-complex in-place FFT 773 (4-threads) vsip_mtrans_f 256*256 real matrix transpose 32.3 Results obtained on a Intel 2nd Generation Core i7 @ 2.16 GHz & measured in µs. Data in cache where possible. GE Intelligent Platforms Contact Information Americas: 1 800 433 2682 or 1 434 978 5100 Global regional phone numbers are listed by location on our web site at www.ge-ip.com/contact www.defense.ge-ip.com ©2011 GE Intelligent Platforms, Inc. All rights reserved. All other brands or names are property of their respective holders. Specifications are subject to change without notice. 03.11 GFA-1798