Ultra-Low Energy Domain-Specific Instruction-Set Processors by Francky Catthoor, Praveen Raghavan, Andy Lambrechts, Murali

By Francky Catthoor, Praveen Raghavan, Andy Lambrechts, Murali Jayapala, Angeliki Kritikakou, Javed Absar

Modern shoppers hold many digital units, like a cellphone, camera, GPS, PDA and an MP3 participant. The performance of every of those units has passed through a big evolution over fresh years, with a steep bring up in either the variety of gains as within the caliber of the companies that they supply. in spite of the fact that, supplying the mandatory compute energy to help (an uncompromised mix of) all this performance is extremely non-trivial. Designing processors that meet the difficult requisites of destiny cellular units calls for the optimization of the embedded approach normally and of the embedded processors specifically, as they need to strike the proper stability among flexibility, strength potency and function. ordinarily, a clothier will attempt to reduce the strength intake (as a ways as wanted) for a given functionality, with a adequate flexibility. despite the fact that, reaching this objective is already advanced while the processor in isolation, yet, actually, the processor is a unmarried part in a extra complicated approach. to be able to layout such advanced process effectively, serious judgements in the course of the layout of every person part should still take into consideration impression at the different elements, with a transparent target to maneuver to an international Pareto optimal within the entire multi-dimensional exploration space.

In the advanced, worldwide layout of battery-operated embedded platforms, the focal point of Ultra-Low strength Domain-Specific Instruction-Set Processors is at the energy-aware structure exploration of domain-specific instruction-set processors and the co-optimization of the datapath structure, foreground reminiscence, and guide reminiscence enterprise with a hyperlink to the necessary mapping concepts or compiler steps on the early phases of the layout. via acting an in depth strength breakdown scan for an entire embedded platform, either power and function bottlenecks were pointed out, including the real family among the various parts. in keeping with this data, structure extensions are proposed for all of the bottlenecks.

Show description

Read or Download Ultra-Low Energy Domain-Specific Instruction-Set Processors PDF

Best microprocessors & system design books

Logic-based artificial intelligence

This landmark quantity represents the fruits of over forty years of study within the use of good judgment as a foundation for representing and manipulating difficulties within the box of man-made intelligence. The use of good judgment as a foundation for common-sense reasoning used to be begun by means of John McCarthy in 1959. the gathering comprises either unique examine and surveys of virtually each topic that makes use of common sense in AI, contributed via prime scientists, and grew out of initial paintings awarded at the Workshop on Logic-Based synthetic Intelligence held in Washington, DC, June 1999.

Building Parallel, Embedded, and Real-Time Applications with Ada

The arriving and recognition of multi-core processors has sparked a renewed curiosity within the improvement of parallel courses. equally, the provision of inexpensive microprocessors and sensors has generated an exceptional curiosity in embedded real-time courses. This ebook presents scholars and programmers whose backgrounds are in conventional sequential programming with the chance to extend their features into parallel, embedded, real-time and disbursed computing.

Extra info for Ultra-Low Energy Domain-Specific Instruction-Set Processors

Sample text

It also presents exploration of various architectures and illustrates the various trade-offs between different cost criterion like area, energy and performance between different architectures. 13 Introduction Introduction (Ch 1) Energy Breakdown +High level arch. +Dyn. 5: Dependence flow of this book Chapter 5 presents our distributed loop buffer organisation and the integration with the clustered foreground data memory organisation. Chapter 6 introduces the technique of executing multiple independent threads on a single-threaded processor in an efficient way for both performance as well as energy.

G. 2 tot 4). This can potentially increase the maximal performance significantly. Many VLIWs provide a rather general instruction set and therefore are still quite flexible. g. wireless or media processing. Another class of VLIW style processors is organized as wide or hierarchical VLIW processors, which provides more flexibility than pure vector processors, as different operations can be executed in parallel [SilH, Mon05]. They form very heterogeneous VLIWs. Some VLIW processors however support only quite specific operations that improve the performance for a selected target application or application domain.

1 Processor core The processor core consists of the hardware that executes the operations (the datapath), the foreground memory from which the operands are loaded and to which the results are stored back and the local interconnection between these components. In addition to these components, the processor core also consists of other components like processor pipelining and the issue type. g. register files, pipeline registers. 2. Based on the design decisions made for each component, processor styles can be defined.

Download PDF sample

Ultra-Low Energy Domain-Specific Instruction-Set Processors by Francky Catthoor, Praveen Raghavan, Andy Lambrechts, Murali
Rated 4.30 of 5 – based on 19 votes