Automatic application-specific instruction-set extensions under microarchitectural constraints
Kubilay Atasu, Laura Pozzi, Paolo Ienne
Abstract
Kubilay Atasu, Laura Pozzi, Paolo Ienne
Abstract
Many commercial processors now offer the possibility of extending their instruction set for a specific application---that is, to introduce customised functional units. There is a need to develop algorithms that decide automatically, from high-level application code, which operations are to be carried out in the customised extensions. A few algorithms exist but are severely limited in the type of operation clusters they can choose and hence reduce significantly the effectiveness of specialisation. In this paper we introduce a more general algorithm which selects maximal-speedup convex subgraphs of the application dataflow graph under fundamental microarchitectural constraints, and which improves significantly on the state of the art.
OpenAlex reports 339 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Many commercial processors now offer the possibility of extending their instruction set for a specific application---that is, to introduce customised functional units. There is a need to develop algorithms that decide automatically, from high-level application code, which operations are to be carried out in the customised extensions. A few algorithms exist but are severely limited in the type of operation clusters they can choose and hence reduce significantly the effectiveness of specialisation. In this paper we introduce a more general algorithm which selects maximal-speedup convex subgraphs of the application dataflow graph under fundamental microarchitectural constraints, and which improves significantly on the state of the art.
Key concepts: Dataflow, Computer science, Speedup, Parallel computing, Instruction set, Set (abstract data type), Graph, Theoretical computer science