HW/SW Co-design of a PCM/PWM converter: a System Level Approach based in the SpecC Methodology
Daniel G. P. Petrini, Braz Izaias da Silva Junior · 2025
Read HW/SW Co-design of a PCM/PWM converter: a System Level Approach based in the SpecC Methodology by Daniel G. P. Petrini, Braz Izaias da Silva Junior (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-22046.txt.
Introduction
Daniel G. P. Petrini and Braz Izaias da Silva Junior revisit a design exercise in splitting a job between hardware and software. The job is a converter that turns pulse-code signals into pulse-width signals, the core of a Class-D audio amplifier. The question is which parts deserve dedicated circuits and which can run as code.
Working through the SpecC system-level method, first published in two thousand five and reissued with refreshed formatting in twenty twenty-five, the pair modeled the converter, estimated costs, and simulated options quickly. Their chosen partition met real-time demands while avoiding both the price of an all-hardware build and the expensive processor a pure software version would need. The case study argues that system-level co-design pays off early, when architectural insight is cheapest to gain.
Catalog record
| CREATOR | Daniel G. P. Petrini, Braz Izaias da Silva Junior |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 26 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-22046
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
HW/SW Co-design of a PCM/PWM converter: a System Level Approach based in the SpecC Methodology Daniel G. P. Petrini, Braz Izaias da Silva Junior Abstract We present a case study applying the SpecC methodology within a system-level hardware/software co-design flow to a PCM-to-PWM converter, the core of a Class-D audio amplifier. The converter was modeled and explored with SpecC methodology to derive an HW/SW partition. Using system-level estimates and fast functional simulation, we evaluated mappings that meet real-time constraints while reducing estimated cost of an all-hardware solution and avoiding the expense of a purely software implementation on a high-end processor. Despite the design’s moderate complexity, the results underline the value of system-level co-design for early architectural insight, rapid validation, and actionable cost/performance trade-offs. I Introduction The recent requirements of the semiconductors industry has lead the design methodologies evolution in order to fill the design gap detected by the 1999 Roadmap [ 1 ] . Today design has very high integration density and complex functionalities to implement arising the necessity to use a higher level of abstraction, the so-called System Level. The latter has many approaches like: reuse methodologies [ 2 ] , Intellectual Property cores, platform-based-design [ 3 ] , specification and design languages for the system level and hardware/software co-design. In the latter, functional partitioning and mapping to hardware or software are analyzed to optimize objectives such as cost and time-to-market. In this paper we present a case study of a design made in the System Level approach using the system level design language SpecC to obtain a hardware and software partitioning. In the next section we mention the design phases of co-design. In section III, co-design computer aided tools are analyzed and in section IV, SpecC is reviewed. Section V contains the description of the case study followed by its results and conclusion. II Related Work In a hardware/software co-design environment, the project can be divided in phases [ 4 ] . These phases are specification, modeling, representation, estimation, partitioning, co-synthesis, co-simulation and eventually co-verification. This study focuses primarily on the estimation and partitioning phases of the design flow. An estimator is a tool that, when applied to a candidate implementation, produces quantitative predictions for relevant metrics, thereby enabling comparative evaluation of design alternatives [ 5 ] . Partitioning may proceed in a software-oriented manner-starting from a 100% software implementation and progressively migrating selected functional blocks to hardware [ 6 ] -or in the reverse direction, beginning with a hardware-based solution and incrementally mapping functions to software [ 7 ] . The process first divides the system into groups or clusters at an appropriate level of granularity; these resulting blocks are then mapped to processing elements (the physical components). Selecting the specific components constitutes architectural allocation . From there, blocks are assigned to hardware or software according to an objective (cost) function that guides the trade-offs among design goals [ 6 , 8 ] . Co-design methodologies and tools differ in how they produce a hardware/software partition-either automatically or manually (designer-driven). Automatic partitioning integrates algorithms within the tool that ingest the captured system model and propose an optimal split according to a defined cost function, optionally with limited designer guidance. Common algorithmic families include constructive methods [ 8 , 9 ] and interactive approaches [ 7 ] . In manual partitioning -typically the outcome of design space exploration-functional logic blocks are grouped by affinity as expressed in a high-level design-language specification. The identification of these groups (often termed behaviors) is performed by the designer, guided by the architectural and logical characteristics of the system, as in [ 10 ] . III Co-design Tools The primary purpose of design tools is to automate and accelerate key phases of the development flow, enabling modeling, simulation, validation, and synthesis from a unified system representation. A wide range of models of computation underpin these tools, including finite-state machines (FSMs), extended FSMs [ 11 ] , data-flow graphs, statecharts, data-path FSMs [ 12 ] , program FSMs (the basis of SpecC) [ 13 ] and Petri nets. This subject is extensively discussed in the Ptolemy project [ 14 ] . Within co-design, the specification should capture the system’s full functionality. This specification serves as an executable model that encodes system properties and supports simulation. It forms the foundation for system-level design exploration-extracting, analyzing, and experimenting with alternative architectural solutions. Our survey of the area reveals two main classes of co-design tools. The first comprises toolchains originally oriented toward board-level design -i.e., printed-circuit boards integrating a CPU, memories, and an ASIC or FPGA-which later converged toward to a second approach that aims towards to the system-on-chip (SoC) paradigm, where these elements are integrated into a single device. Representative tools in the first group include Cosyma [ 6 ] , Vulcan II [ 7 ] , Ptolemy [ 14 ] , PISH [ 15 ] , Polis [ 16 ] , and SpecSyn [ 17 ] . Tools emblematic of the second, SoC-focused group include CoWare [ 18 ] , SpecC [ 13 ] , SystemC [ 19 ] , and, more recently, Spark, which raises the abstraction level by compiling high-level descriptions into HDL [ 20 ] . For the case study reported in Section V , we compared several of these tools using the following criteria: (a) compatibility of the modeling formalism with our specification (a C program), (b) support for design-space exploration of the hardware/software partition-preferably with automated assistance, (c) quality of documentation, and (d) availability. Based on this evaluation, we selected a tool grounded in the SpecC language and methodology: the System-on-Chip Environment (SCE) , alpha version, which was made available to us by its creators at the University of California, Irvine. IV SpecC Methodology The SpecC methodology is a tool developed to support high levels of abstraction designs. It aims to refine an abstract design specification into a implementation model running in the target architecture. It is based in four models, namely a specification model, an architecture model, a communication model and finally an implementation model. In figure 1 we see these models and the three transformations, architecture Exploration, Communication Synthesis and backend, which comprises hardware and interface synthesis and software compiling, as in [ 13 ] . Fig. 1: SpecC Models. The first tree models are written in SpecC language. Each one is compiled and simulated until its validation. Estimations are obtained by operations counting during simulation together with information of the processing elements delays per operation. The Specification Model is a functional description of the system. To create this model, the requirements must be analyzed and translated into the SpecC language. The system is then divided into blocks (behaviors) with similar functionality, thereby defining the granularity and revealing potential parallelism. Simulation at this stage is purely functional (zero-delay), with no timing considerations. The Architecture Model is obtained after architecture exploration and architecture refinement . Architecture exploration evaluates design alternatives, using behavior estimates and electronic component characteristics to produce a manual hardware/software partition. Architecture refinement then applies transformations such as behavior and variable optimizations. The resulting model reflects the impact of the mapped computation on the chosen architecture and captures the desired hardware-software division. Communication Model is reached after