Multicore math engine for science and engineering
MtxVec is an object oriented numerical library for Delphi and C++ Builder developers with complete matrix/vector arithmetic. It features a large set of vectorized mathemathical functions which cover complex numbers, sparse matrices, math parser, probabilities, optimization unit, SVD, QR, LQ, and LU solvers, special functions, and more. All applications based on this library take advantage of CPU-specific code optimization and symmetric multiprocessing. Efficient memory and CPU cache management further enhance its performance.
Overview
- Use pure pascal (100% full source code) or high performance external dlls
- With FireMonkey use common source to compile your applications for Windows, Mac OS, iOS / iPhone, Android and Linux. The full source version is called MtxVec Core Edition.
- Full hardware acceleration for Windows (32 and 64bit) and for Linux OS (64bit).
- Maps to native OS math lib on Android, iOS, MacOS and Linux for maximum speed of scalar math functions.
- Comprehensive mapping to Apple's Accelerator pack for all functions on iOS and MacOS for maximum performance on all Apples devices. (The Accelerator pack includes all LAPACK functions among many other features).
- Runs natively on Apples devices with Intel x86 and Apple M1 CPUs without the need for Rosetta.
- Supports a large set of compilers from Delphi XE3 up to latest Rad Studio XE Alexandria (v11). FireMonkey support requires at least XE 10.3.
- Operator overloading support for vectors and matrices from including Delphi 2006 and C++Builder 2006.
- Super conductive memory management allows linear scaling of numerical algorithms with number of cores when multithreading.
- 64bit compiler support for Embarcadero Delphi XE2 and later.
- 64bit compiler support for Embarcadero C++Builder XE4 and later.
- Brute-force K-NN algorithm with excellent performance
- Powerfull scripting support with Matlab/Scilab like syntax and roughly 3x faster than Python (or other scripting languages) for vector/matrix math.
- Includes Debugger Visualizer, a debugger plug-in for faster debugging of numerical algorithms.
- Ability to select the floating point precision (single or double), in which to run the algorithm, at run-time.
- Provides and uses a fair critical section to support linear scaling with core count when multi-threading.
- Support for 4GB address space for 32bit applications under 64bit OS and 3GB address space for 32bit OS. For more info how to enable this support for your project see here.
Most users observe typical performance improvements of 2x to 3x for vector functions, but speed increases up to 10x are often obtained. The matrix multiplication, for example, is up to 20x faster. Many times we can speed-up customers algorithms by 100-200x by rewriting with MtxVec.
