(MPI, PVM, OpenMP, ...)
Just as Mathematica made technical computing immediately accessible without the need to write complicated low-level code, so gridMathematica made grid computing accessible in a similarly convenient and scalable way.
The key is that Mathematica's symbolic architecture—which treats programs and data in a uniform fashion—allows the creation of a unique high-level representation of grid computing. And together with its vast built-in computational capabilities and very high-level core programming language, Mathematica makes it realistic to start doing serious grid computing in an immediate interactive way—then seamlessly scale up to the largest possible computations, using Mathematica's new generation grid-enabled development environment.
Mathematica's symbolic architecture also vastly simplifies the configuration of a grid network, allowing programs to be shared just like data across high-end grid systems, ad hoc heterogeneous networks or simple multicore PCs—and to be integrated with modern grid middleware systems. And Wolfram Research's longstanding relationships with high-performance computing vendors ensure that gridMathematica is consistently optimized for the latest grid hardware and software.
- Full language-level support for parallelization
Support for multiprocessor machines, clusters, and grids »
- High-performance communication optimized for all common configurations
- Efficient, adaptive load balancing
- Scheduling accounts for processor speed and communication latency
- User-programmable scheduling for problem-specific adaptation
- Automatic failure recovery and reassignment of stranded processes
- Speculative parallelization for nondeterministic problems
- Efficient language-level support for synchronization between nodes
Support for all standard and emerging computer platforms »
- Wolfram Research uses gridMathematica extensively for algorithm discovery
- Mathematica ran on 1980s supercomputers
- Mathematica's MathLink protocol has maintained compatibility for two decades