SIMtone ASPEED offers several key features that expand on a very simple application acceleration concept. The concept is that programs include serially dependent and independent components that can be overlapped in order to reduce elapsed time and improve the systems effective capacity.

If a program is written so that each component must be executed serially and does not start the next component until the previous is completed, this program is referred to as single thread. These single thread applications take the longest elapsed time but are the easiest to write and debug.

A program can be written to overlap the execution of components that are independent in their execution. These multi-thread and multi-process programs are much more complex to write but can reduce the elapsed time by overlapping processing and waits using spare cycles or multiple cores, CPUs or systems.

SIMtone ASPEED uses a multi-process approach, referred to as heavy weight processes. This approach expands on the notion of threads and processes and enables hybrid applications using GPUs, FPGAs and other auxiliary processors  to provide an efficient level of distribution in the same system or across systems.

SIMtone ASPEED dynamically distributes the iterations and rebalances the execution processes to optimize the response time at execution time.

SIMtone ASPEED uniquely offers this in a way that does not require the programmer to have multi-threading, multi-processing programming or application integration skills. ASPEED's self-optimizing capability is invoked at run time using ISV integrated interfaces or using SIMtone's ASPEED Software Developer's kit. It automatically provides the application with the best capabilities of both multiple processes and multi-thread.  

SIMtone's ASPEED processes are applied when writing new applications or adapting a single thread or multi-thread application. The result is that computations can be spread across a single system or across multiple systems thereby significantly increasing the concurrent processing and leveraging overlapping processing to reduce the effective elapsed time. The result is significantly reduced elapsed time, more efficient use of CPU utilization and automatic application scalability.

SIMtone's ASPEED real-time intelligent dynamic workload balancing functionality distributes the relevant iterations and data objects based upon the type of computational interdependencies specified in the API.   

Typical SIMtone ASPEED applications see linear and super-linear acceleration of the accelerated portions.