On the working set concept for data-flow machines

Tokoro, Mario; Jagannathan, J.; Sunahara, Hideki

doi:10.1145/800046.801643

Cited by 15 publications

(4 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Too fine-grained (instruction level) multithreading Difficulty in exploiting memory hierarchies and registers Asynchronous triggering of instructions Many researchers have addressed the first two limitations of dataflow architectures [9][10][11]. SDF specifically addresses the third limitation.…”

Section: Dataflow Model and Architecturementioning

confidence: 99%

An efficient way of passing of data in a multithreaded scheduled dataflow architecture

Arul

Yeh

Hsu

et al. 2005

Eighth International Conference on High-Performance Computing in Asia-Pacific Region (HPCASIA'05)

View full text Add to dashboard Cite

The Scheduled Dataflow (SDF) architecture deviates from current trend of building complex hardware to exploit Instruction Level Parallelism (ILP) by exploring a simpler, yet powerful execution paradigm that is based on dataflow, multithreading and decoupling of memory accesses from execution. A program is partitioned into non-blocking threads and all memory accesses are decoupled from the thread's execution. Data is pre-loaded into the thread's context (registers), and all results are post-stored after the completion of the thread's execution. This paper presents an efficient way of storing of data into the thread's register context directly as opposed to storing of data into the frame memory. This way eliminates the need for creating thread frames when there are sufficient register contexts available in the system. Thus, it is possible to explore the scalability of SDF architecture's performance when more register contexts are available on the chip. All the benchmarks ran using these two methods show performance improvement of at least about 20%. This method of allocating data to a consecutive thread in a multithreaded architecture could be applied generally.

show abstract

Section: Dataflow Model and Architecturementioning

confidence: 99%

An efficient way of passing of data in a multithreaded scheduled dataflow architecture

Arul

Yeh

Hsu

et al. 2005

Eighth International Conference on High-Performance Computing in Asia-Pacific Region (HPCASIA'05)

View full text Add to dashboard Cite

show abstract

“…A data flow program in it's pure form is not amenable to a cache, primarily due to the nondeterministic selection of instructions for execution. Reordering of instructions of such a program based on certain criteria [7] , produces synthetic localities, that justify the presence of a cache. The recurrent use of instructions also causes the existence of temporal localities.…”

Section: Locality In Data Flowmentioning

confidence: 99%

“…Thoreson discusses the concept of working set for data flow environments [6]. Tokoro defines the working set for a data flow program as the minimum set of instructions that keep the execution unit busy [7]. The working set concept of a dynamic data flow program could be based on both the principle of locality and the simultaneity of execution.…”

Section: Locality In Data Flowmentioning

confidence: 99%

See 1 more Smart Citation

Cache design for an Explicit Token Store data flow architecture

Shanmugam

Andhare

Kavi

et al.

Proceedings of 1993 5th IEEE Symposium on Parallel and Distributed Processing

View full text Add to dashboard Cite

This paper discusses the performance enhancement of an Explicit Token Store (ETS) architecture as a result of the introduction of hierarchical memory system. It indicates how locality can be exploited in a dataflow environment and evaluates a novel cache subsystem design f o r this purpose. The introduction of an operand cache and an instruction cache, cause a performance enhancement, remarkable enough, to justify their cost. The paper explores various Optimizations and resource management schemes to enhance performance further and discusses the costs and trade-08s involved. Some of the ideas discussed in this paper f o r the E T S architecture are applicable to many finegrained multithreaded architectures

show abstract