Decoupled access/execute computer architectures

Abstract

An architecture for high-performance scalar computation is proposed and discussed. The main feature of the architecture is a high degree of decoupling between operand access and execution. This results in an implementation that has two separate instruction streams that communicate via architectural queues. Performance comparisons with a conventional scalar architecture are given, and these show that significant performance gains can be realized. Single-instruction-stream versions, both physical and conceptual, are discussed, with the primary goal of minimizing the differences with conventional architectures. This allows known compilation and programming techniques to be used. Finally, the problem of deadlock in a decoupled system is discussed, and a deadlock prevention method is given.

Deriving Efficient Data Movement from Decoupled Access/Execute Specifications

“…1 shows a CTM filter implementation in our prototype C++ framework. 3 The class method kernel closely resembles the filter's original kernel code, except that accesses to arrays have been replaced with uses of AEcute access descriptors ( §4.1).…”

“…The separation is reminiscent of decoupled access/execute architectures [3], which run (conceptually or physically) separate access and execute instruction streams. Another point of reference is data-prefetching in virtual shared memory, a shared memory abstraction for computers with physically distributed memories [4].…”

Customization of application specific heterogeneous multi-pipeline processors

“…In [16], the author discussed a decoupled Access/Execute architecture, with two computation units each of which contained its own instruction streams. Using a similar architecture, in [15], the authors presented a design approach for dual-pipeline customized processor.…”

Lessons Learned from Model Checking a NASA Robot Controller

“…The term "spatial modularity" was chosen because of a similarity to modular design principles for hardware [41]. The spatial modularity concept leads to two design principles, functional localization and global data encapsulation.…”