|
|
Emergent System Properties
|
|
|
The Cell Matrix computing architecture is fault tolerant, scalable, distributed, and massively parallel. Not only is data flow massively parallel and fine-grained, but control structures and control flow is
massively parallel and fine-grained. Configuration or programming of Cell Matrices™ is also distributed and fine-grained and is therefore open to being arranged as a massively parallel, distributed operation, and we have developed techniques and software for this.
|
|
|
There is no other computing system on the market today that has complete crossover with this feature set. For example, memory and processor colocation and intermixture can be highly beneficial but is confounded by the architecture of the CPU/memory machine ('C' stands for central, or centralized). As another example, distributed, massively parallel configuration is highly beneficial for large systems or for configuring one region while the other regions continue to run, but this capability is not found in FPGAs.
|
|
|
Cell Matrices are configured by a set of local interactions between cells, completely onboard the hardware itself. Self-configurability translates to the ability of a programmer to load dynamically defined circuitry into Cell Matrix hardware, as well as statically defined circuitry, and circuitry that contains both static and dynamic regions.
|
|
