7 %% http://www.michaelshell.org/
8 %% for current contact information.
10 %% This is a skeleton file demonstrating the use of IEEEtran.cls
11 %% (requires IEEEtran.cls version 1.7 or later) with an IEEE conference paper.
14 %% http://www.michaelshell.org/tex/ieeetran/
15 %% http://www.ctan.org/tex-archive/macros/latex/contrib/IEEEtran/
17 %% http://www.ieee.org/
19 %%*************************************************************************
21 %% This code is offered as-is without any warranty either expressed or
22 %% implied; without even the implied warranty of MERCHANTABILITY or
23 %% FITNESS FOR A PARTICULAR PURPOSE!
24 %% User assumes all risk.
25 %% In no event shall IEEE or any contributor to this code be liable for
26 %% any damages or losses, including, but not limited to, incidental,
27 %% consequential, or any other damages, resulting from the use or misuse
28 %% of any information contained here.
30 %% All comments are the opinions of their respective authors and are not
31 %% necessarily endorsed by the IEEE.
33 %% This work is distributed under the LaTeX Project Public License (LPPL)
34 %% ( http://www.latex-project.org/ ) version 1.3, and may be freely used,
35 %% distributed and modified. A copy of the LPPL, version 1.3, is included
36 %% in the base LaTeX documentation of all distributions of LaTeX released
37 %% 2003/12/01 or later.
38 %% Retain all contribution notices and credits.
39 %% ** Modified files should be clearly indicated as such, including **
40 %% ** renaming them and changing author support contact information. **
42 %% File list of work: IEEEtran.cls, IEEEtran_HOWTO.pdf, bare_adv.tex,
43 %% bare_conf.tex, bare_jrnl.tex, bare_jrnl_compsoc.tex
44 %%*************************************************************************
46 % *** Authors should verify (and, if needed, correct) their LaTeX system ***
47 % *** with the testflow diagnostic prior to trusting their LaTeX platform ***
48 % *** with production work. IEEE's font choices can trigger bugs that do ***
49 % *** not appear when using other class files. ***
50 % The testflow support page is at:
51 % http://www.michaelshell.org/tex/testflow/
55 % Note that the a4paper option is mainly intended so that authors in
56 % countries using A4 can easily print to A4 and see how their papers will
57 % look in print - the typesetting of the document will not typically be
58 % affected with changes in paper size (but the bottom and side margins will).
59 % Use the testflow package mentioned above to verify correct handling of
60 % both paper sizes by the user's LaTeX system.
62 % Also note that the "draftcls" or "draftclsnofoot", not "draft", option
63 % should be used if it is desired that the figures are to be displayed in
67 \documentclass[conference,pdf,a4paper,10pt,final,twoside,twocolumn]{IEEEtran}
68 \IEEEoverridecommandlockouts
69 % Add the compsoc option for Computer Society conferences.
71 % If IEEEtran.cls has not been installed into the LaTeX system files,
72 % manually specify the path to it like:
73 % \documentclass[conference]{../sty/IEEEtran}
75 % Some very useful LaTeX packages include:
76 % (uncomment the ones you want to load)
78 % *** MISC UTILITY PACKAGES ***
81 % Heiko Oberdiek's ifpdf.sty is very useful if you need conditional
82 % compilation based on whether the output is pdf or dvi.
89 % The latest version of ifpdf.sty can be obtained from:
90 % http://www.ctan.org/tex-archive/macros/latex/contrib/oberdiek/
91 % Also, note that IEEEtran.cls V1.7 and later provides a builtin
92 % \ifCLASSINFOpdf conditional that works the same way.
93 % When switching from latex to pdflatex and vice-versa, the compiler may
94 % have to be run twice to clear warning/error messages.
98 % *** CITATION PACKAGES ***
101 % cite.sty was written by Donald Arseneau
102 % V1.6 and later of IEEEtran pre-defines the format of the cite.sty package
103 % \cite{} output to follow that of IEEE. Loading the cite package will
104 % result in citation numbers being automatically sorted and properly
105 % "compressed/ranged". e.g., [1], [9], [2], [7], [5], [6] without using
106 % cite.sty will become [1], [2], [5]--[7], [9] using cite.sty. cite.sty's
107 % \cite will automatically add leading space, if needed. Use cite.sty's
108 % noadjust option (cite.sty V3.8 and later) if you want to turn this off.
109 % cite.sty is already installed on most LaTeX systems. Be sure and use
110 % version 4.0 (2003-05-27) and later if using hyperref.sty. cite.sty does
111 % not currently provide for hyperlinked citations.
112 % The latest version can be obtained at:
113 % http://www.ctan.org/tex-archive/macros/latex/contrib/cite/
114 % The documentation is contained in the cite.sty file itself.
121 % *** GRAPHICS RELATED PACKAGES ***
124 \usepackage[pdftex]{graphicx}
125 % declare the path(s) where your graphic files are
126 % \graphicspath{{../pdf/}{../jpeg/}}
127 % and their extensions so you won't have to specify these with
128 % every instance of \includegraphics
129 % \DeclareGraphicsExtensions{.pdf,.jpeg,.png}
131 % or other class option (dvipsone, dvipdf, if not using dvips). graphicx
132 % will default to the driver specified in the system graphics.cfg if no
133 % driver is specified.
134 % \usepackage[dvips]{graphicx}
135 % declare the path(s) where your graphic files are
136 % \graphicspath{{../eps/}}
137 % and their extensions so you won't have to specify these with
138 % every instance of \includegraphics
139 % \DeclareGraphicsExtensions{.eps}
141 % graphicx was written by David Carlisle and Sebastian Rahtz. It is
142 % required if you want graphics, photos, etc. graphicx.sty is already
143 % installed on most LaTeX systems. The latest version and documentation can
145 % http://www.ctan.org/tex-archive/macros/latex/required/graphics/
146 % Another good source of documentation is "Using Imported Graphics in
147 % LaTeX2e" by Keith Reckdahl which can be found as epslatex.ps or
148 % epslatex.pdf at: http://www.ctan.org/tex-archive/info/
150 % latex, and pdflatex in dvi mode, support graphics in encapsulated
151 % postscript (.eps) format. pdflatex in pdf mode supports graphics
152 % in .pdf, .jpeg, .png and .mps (metapost) formats. Users should ensure
153 % that all non-photo figures use a vector format (.eps, .pdf, .mps) and
154 % not a bitmapped formats (.jpeg, .png). IEEE frowns on bitmapped formats
155 % which can result in "jaggedy"/blurry rendering of lines and letters as
156 % well as large increases in file sizes.
158 % You can find documentation about the pdfTeX application at:
159 % http://www.tug.org/applications/pdftex
165 % *** MATH PACKAGES ***
167 %\usepackage[cmex10]{amsmath}
168 % A popular package from the American Mathematical Society that provides
169 % many useful and powerful commands for dealing with mathematics. If using
170 % it, be sure to load this package with the cmex10 option to ensure that
171 % only type 1 fonts will utilized at all point sizes. Without this option,
172 % it is possible that some math symbols, particularly those within
173 % footnotes, will be rendered in bitmap form which will result in a
174 % document that can not be IEEE Xplore compliant!
176 % Also, note that the amsmath package sets \interdisplaylinepenalty to 10000
177 % thus preventing page breaks from occurring within multiline equations. Use:
178 %\interdisplaylinepenalty=2500
179 % after loading amsmath to restore such page breaks as IEEEtran.cls normally
180 % does. amsmath.sty is already installed on most LaTeX systems. The latest
181 % version and documentation can be obtained at:
182 % http://www.ctan.org/tex-archive/macros/latex/required/amslatex/math/
188 % *** SPECIALIZED LIST PACKAGES ***
190 %\usepackage{algorithmic}
191 % algorithmic.sty was written by Peter Williams and Rogerio Brito.
192 % This package provides an algorithmic environment fo describing algorithms.
193 % You can use the algorithmic environment in-text or within a figure
194 % environment to provide for a floating algorithm. Do NOT use the algorithm
195 % floating environment provided by algorithm.sty (by the same authors) or
196 % algorithm2e.sty (by Christophe Fiorio) as IEEE does not use dedicated
197 % algorithm float types and packages that provide these will not provide
198 % correct IEEE style captions. The latest version and documentation of
199 % algorithmic.sty can be obtained at:
200 % http://www.ctan.org/tex-archive/macros/latex/contrib/algorithms/
201 % There is also a support site at:
202 % http://algorithms.berlios.de/index.html
203 % Also of interest may be the (relatively newer and more customizable)
204 % algorithmicx.sty package by Szasz Janos:
205 % http://www.ctan.org/tex-archive/macros/latex/contrib/algorithmicx/
210 % *** ALIGNMENT PACKAGES ***
213 % Frank Mittelbach's and David Carlisle's array.sty patches and improves
214 % the standard LaTeX2e array and tabular environments to provide better
215 % appearance and additional user controls. As the default LaTeX2e table
216 % generation code is lacking to the point of almost being broken with
217 % respect to the quality of the end results, all users are strongly
218 % advised to use an enhanced (at the very least that provided by array.sty)
219 % set of table tools. array.sty is already installed on most systems. The
220 % latest version and documentation can be obtained at:
221 % http://www.ctan.org/tex-archive/macros/latex/required/tools/
224 %\usepackage{mdwmath}
226 % Also highly recommended is Mark Wooding's extremely powerful MDW tools,
227 % especially mdwmath.sty and mdwtab.sty which are used to format equations
228 % and tables, respectively. The MDWtools set is already installed on most
229 % LaTeX systems. The lastest version and documentation is available at:
230 % http://www.ctan.org/tex-archive/macros/latex/contrib/mdwtools/
233 % IEEEtran contains the IEEEeqnarray family of commands that can be used to
234 % generate multiline equations as well as matrices, tables, etc., of high
238 %\usepackage{eqparbox}
239 % Also of notable interest is Scott Pakin's eqparbox package for creating
240 % (automatically sized) equal width boxes - aka "natural width parboxes".
242 % http://www.ctan.org/tex-archive/macros/latex/contrib/eqparbox/
248 % *** SUBFIGURE PACKAGES ***
249 %\usepackage[tight,footnotesize]{subfigure}
250 % subfigure.sty was written by Steven Douglas Cochran. This package makes it
251 % easy to put subfigures in your figures. e.g., "Figure 1a and 1b". For IEEE
252 % work, it is a good idea to load it with the tight package option to reduce
253 % the amount of white space around the subfigures. subfigure.sty is already
254 % installed on most LaTeX systems. The latest version and documentation can
256 % http://www.ctan.org/tex-archive/obsolete/macros/latex/contrib/subfigure/
257 % subfigure.sty has been superceeded by subfig.sty.
261 %\usepackage[caption=false]{caption}
262 %\usepackage[font=footnotesize]{subfig}
263 % subfig.sty, also written by Steven Douglas Cochran, is the modern
264 % replacement for subfigure.sty. However, subfig.sty requires and
265 % automatically loads Axel Sommerfeldt's caption.sty which will override
266 % IEEEtran.cls handling of captions and this will result in nonIEEE style
267 % figure/table captions. To prevent this problem, be sure and preload
268 % caption.sty with its "caption=false" package option. This is will preserve
269 % IEEEtran.cls handing of captions. Version 1.3 (2005/06/28) and later
270 % (recommended due to many improvements over 1.2) of subfig.sty supports
271 % the caption=false option directly:
272 %\usepackage[caption=false,font=footnotesize]{subfig}
274 % The latest version and documentation can be obtained at:
275 % http://www.ctan.org/tex-archive/macros/latex/contrib/subfig/
276 % The latest version and documentation of caption.sty can be obtained at:
277 % http://www.ctan.org/tex-archive/macros/latex/contrib/caption/
282 % *** FLOAT PACKAGES ***
284 %\usepackage{fixltx2e}
285 % fixltx2e, the successor to the earlier fix2col.sty, was written by
286 % Frank Mittelbach and David Carlisle. This package corrects a few problems
287 % in the LaTeX2e kernel, the most notable of which is that in current
288 % LaTeX2e releases, the ordering of single and double column floats is not
289 % guaranteed to be preserved. Thus, an unpatched LaTeX2e can allow a
290 % single column figure to be placed prior to an earlier double column
291 % figure. The latest version and documentation can be found at:
292 % http://www.ctan.org/tex-archive/macros/latex/base/
296 %\usepackage{stfloats}
297 % stfloats.sty was written by Sigitas Tolusis. This package gives LaTeX2e
298 % the ability to do double column floats at the bottom of the page as well
299 % as the top. (e.g., "\begin{figure*}[!b]" is not normally possible in
300 % LaTeX2e). It also provides a command:
302 % to enable the placement of footnotes below bottom floats (the standard
303 % LaTeX2e kernel puts them above bottom floats). This is an invasive package
304 % which rewrites many portions of the LaTeX2e float routines. It may not work
305 % with other packages that modify the LaTeX2e float routines. The latest
306 % version and documentation can be obtained at:
307 % http://www.ctan.org/tex-archive/macros/latex/contrib/sttools/
308 % Documentation is contained in the stfloats.sty comments as well as in the
309 % presfull.pdf file. Do not use the stfloats baselinefloat ability as IEEE
310 % does not allow \baselineskip to stretch. Authors submitting work to the
311 % IEEE should note that IEEE rarely uses double column equations and
312 % that authors should try to avoid such use. Do not be tempted to use the
313 % cuted.sty or midfloat.sty packages (also by Sigitas Tolusis) as IEEE does
314 % not format its papers in such ways.
320 % *** PDF, URL AND HYPERLINK PACKAGES ***
323 % url.sty was written by Donald Arseneau. It provides better support for
324 % handling and breaking URLs. url.sty is already installed on most LaTeX
325 % systems. The latest version can be obtained at:
326 % http://www.ctan.org/tex-archive/macros/latex/contrib/misc/
327 % Read the url.sty source comments for usage information. Basically,
334 % *** Do not adjust lengths that control margins, column widths, etc. ***
335 % *** Do not use packages that alter fonts (such as pslatex). ***
336 % There should be no need to do such things with IEEEtran.cls V1.6 and later.
337 % (Unless specifically asked to do so by the journal or conference you plan
338 % to submit to, of course. )
340 % correct bad hyphenation here
341 \hyphenation{op-tical net-works semi-conduc-tor}
343 % Macro for certain acronyms in small caps. Doesn't work with the
344 % default font, though (it contains no smallcaps it seems).
345 \def\acro#1{{\small{#1}}}
346 \def\acrop#1{\acro{#1}s}
347 \def\acrotiny#1{{\scriptsize{#1}}}
348 \def\VHDL{\acro{VHDL}}
350 \def\CLaSH{{\small{C}}$\lambda$a{\small{SH}}}
351 \def\CLaSHtiny{{\scriptsize{C}}$\lambda$a{\scriptsize{SH}}}
353 % Macro for pretty printing haskell snippets. Just monospaced for now, perhaps
354 % we'll get something more complex later on.
355 \def\hs#1{\texttt{#1}}
356 \def\quote#1{``{#1}"}
358 \newenvironment{xlist}[1][\rule{0em}{0em}]{%
360 \settowidth{\labelwidth}{#1:}
361 \setlength{\labelsep}{0.5em}
362 \setlength{\leftmargin}{\labelwidth}
363 \addtolength{\leftmargin}{\labelsep}
364 \addtolength{\leftmargin}{\parindent}
365 \setlength{\rightmargin}{0pt}
366 \setlength{\listparindent}{\parindent}
367 \setlength{\itemsep}{0 ex plus 0.2ex}
368 \renewcommand{\makelabel}[1]{##1:\hfil}
373 \usepackage{paralist}
375 \def\comment#1{{\color[rgb]{1.0,0.0,0.0}{#1}}}
377 \usepackage{cleveref}
378 \crefname{figure}{figure}{figures}
379 \newcommand{\fref}[1]{\cref{#1}}
380 \newcommand{\Fref}[1]{\Cref{#1}}
382 \usepackage{epstopdf}
384 \epstopdfDeclareGraphicsRule{.svg}{pdf}{.pdf}{rsvg-convert --format=pdf < #1 > \noexpand\OutputFile}
386 %include polycode.fmt
389 \newcounter{Codecount}
390 \setcounter{Codecount}{0}
392 \newenvironment{example}
394 \refstepcounter{equation}
405 % can use linebreaks \\ within to get better formatting as desired
406 \title{C$\lambda$aSH: Structural Descriptions \\ of Synchronous Hardware using Haskell}
409 % author names and affiliations
410 % use a multiple column layout for up to three different
412 % \author{\IEEEauthorblockN{Christiaan Baaij, Matthijs Kooijman, Jan Kuper, Arjan Boeijink, Marco Gerards}%, Bert Molenkamp, Sabih H. Gerez}
413 % \IEEEauthorblockA{%Computer Architecture for Embedded Systems (CAES)\\
414 % Department of EEMCS, University of Twente\\
415 % P.O. Box 217, 7500 AE, Enschede, The Netherlands\\
416 % c.p.r.baaij@@utwente.nl, matthijs@@stdin.nl, j.kuper@@utwente.nl}
417 % \thanks{Supported through the FP7 project: S(o)OS (248465)}
420 \author{\IEEEauthorblockN{Blind Review}%, Bert Molenkamp, Sabih H. Gerez}
421 \IEEEauthorblockA{%Computer Architecture for Embedded Systems (CAES)\\
426 \thanks{Supported through: hidden for blind review}
430 % \IEEEauthorblockN{Homer Simpson}
431 % \IEEEauthorblockA{Twentieth Century Fox\\
433 % Email: homer@thesimpsons.com}
435 % \IEEEauthorblockN{James Kirk\\ and Montgomery Scott}
436 % \IEEEauthorblockA{Starfleet Academy\\
437 % San Francisco, California 96678-2391\\
438 % Telephone: (800) 555--1212\\
439 % Fax: (888) 555--1212}}
441 % conference papers do not typically use \thanks and this command
442 % is locked out in conference mode. If really needed, such as for
443 % the acknowledgment of grants, issue a \IEEEoverridecommandlockouts
444 % after \documentclass
446 % for over three affiliations, or if they all won't fit within the width
447 % of the page, use this alternative format:
449 %\author{\IEEEauthorblockN{Michael Shell\IEEEauthorrefmark{1},
450 %Homer Simpson\IEEEauthorrefmark{2},
451 %James Kirk\IEEEauthorrefmark{3},
452 %Montgomery Scott\IEEEauthorrefmark{3} and
453 %Eldon Tyrell\IEEEauthorrefmark{4}}
454 %\IEEEauthorblockA{\IEEEauthorrefmark{1}School of Electrical and Computer Engineering\\
455 %Georgia Institute of Technology,
456 %Atlanta, Georgia 30332--0250\\ Email: see http://www.michaelshell.org/contact.html}
457 %\IEEEauthorblockA{\IEEEauthorrefmark{2}Twentieth Century Fox, Springfield, USA\\
458 %Email: homer@thesimpsons.com}
459 %\IEEEauthorblockA{\IEEEauthorrefmark{3}Starfleet Academy, San Francisco, California 96678-2391\\
460 %Telephone: (800) 555--1212, Fax: (888) 555--1212}
461 %\IEEEauthorblockA{\IEEEauthorrefmark{4}Tyrell Inc., 123 Replicant Street, Los Angeles, California 90210--4321}}
466 % use for special paper notices
467 %\IEEEspecialpapernotice{(Invited Paper)}
472 % make the title area
477 \CLaSH\ is a functional hardware description language that borrows both its
478 syntax and semantics from the functional programming language Haskell.
479 Polymorphism and higher-order functions provide a level of abstraction and
480 generality that allow a circuit designer to describe circuits in a more
481 natural way than possible with the language elements found in the traditional
482 hardware description languages.
484 Circuit descriptions can be translated to synthesizable VHDL using the
485 prototype \CLaSH\ compiler. As the circuit descriptions, simulation code, and
486 test input are also valid Haskell, complete simulations can be done by a
487 Haskell compiler or interpreter, allowing high-speed simulation and analysis.
489 % \CLaSH\ supports stateful descriptions by explicitly making the current
490 % state an argument of the function, and the updated state part of the result.
491 % This makes \CLaSH\ descriptions in essence the combinational parts of a
494 % IEEEtran.cls defaults to using nonbold math in the Abstract.
495 % This preserves the distinction between vectors and scalars. However,
496 % if the conference you are submitting to favors bold math in the abstract,
497 % then you can use LaTeX's standard command \boldmath at the very start
498 % of the abstract to achieve this. Many IEEE journals/conferences frown on
499 % math in the abstract anyway.
506 % For peer review papers, you can put extra information on the cover
508 % \ifCLASSOPTIONpeerreview
509 % \begin{center} \bfseries EDICS Category: 3-BBND \end{center}
512 % For peerreview papers, this IEEEtran command inserts a page break and
513 % creates the second title. It will be ignored for other modes.
514 \IEEEpeerreviewmaketitle
516 \section{Introduction}
517 Hardware description languages (\acrop{HDL}) have not allowed the productivity
518 of hardware engineers to keep pace with the development of chip technology.
519 While traditional \acrop{HDL}, like \VHDL~\cite{VHDL2008} and
520 Verilog~\cite{Verilog}, are very good at describing detailed hardware
521 properties such as timing behavior, they are generally cumbersome in
522 expressing the higher-level abstractions needed for today's large and complex
523 circuit designs. In an attempt to raise the abstraction level of the
524 descriptions, a great number of approaches based on functional languages have
525 been proposed \cite{Cardelli1981,muFP,DAISY,T-Ruby,HML2,Hydra,Hawk1,Lava,
526 Wired,ForSyDe1,reFLect}. The idea of using functional languages for hardware
527 descriptions started in the early 1980s \cite{Cardelli1981,muFP,DAISY}, a
528 time which also saw the birth of the currently popular \acrop{HDL}, such as
529 \VHDL. Functional languages are especially well suited to describe hardware
530 because combinational circuits can be directly modeled as mathematical
531 functions and functional languages are very good at describing and composing
534 In an attempt to reduce the effort involved with prototyping a new
535 language, such as creating all the required tooling like parsers and
536 type-checkers, many functional \acrop{HDL} \cite{Hydra,Hawk1,Lava,Wired} are
537 embedded as a domain specific language (\acro{DSL}) within the functional
538 language Haskell \cite{Haskell}. This means that a developer is given a
539 library of Haskell functions and types that together form the language
540 primitives of the \acro{DSL}. The primitive functions used to describe a
541 circuit do not actually process any signals, they instead compose a large
542 graph (which is usually hidden from the designer). This graph is then further processed by an embedded circuit compiler which can perform e.g. simulation or synthesis. As Haskell's choice elements (\hs{case}-expressions, pattern-matching, etc.) are evaluated at the time the graph is being build, they are no longer visible to the embedded compiler that processes the datatype. Consequently, it is impossible to capture Haskell's choice elements within a circuit description when taking the embedded language approach. This does not mean that circuits specified in an embedded language can not contain choice, just that choice elements only exists as functions, e.g. a multiplexer function, and not as syntactic elements of the language itself.
544 This research is uses (a subset of) the Haskell language \emph{itself} for the purpose of describing hardware. By taking this approach, this research \emph{can} capture certain language constructs, like all of Haskell's choice elements, within circuit descriptions. Advanced features of Haskell, such as polymorphic typing and higher-order functions, are also supported.
546 % supporting polymorphism, higher-order functions and such an extensive array
547 % of choice-elements, combined with a very concise way of specifying circuits
548 % is new in the domain of (functional) \acrop{HDL}.
549 % As the hardware descriptions are plain Haskell
550 % functions, these descriptions can be compiled to an executable binary
551 % for simulation using an optimizing Haskell compiler such as the Glasgow
552 % Haskell Compiler (\GHC)~\cite{ghc}.
554 Where descriptions in a conventional \acro{HDL} have an explicit clock for the
555 purposes state and synchronicity, the clock is implicit for the descriptions
556 and research presented in this paper. A circuit designer describes the
557 behavior of the hardware between clock cycles, as a transition from the
558 current state to the next. Many functional \acrop{HDL} model signals as a
559 stream of values over time; state is then modeled as a delay on this stream of
560 values. Descriptions presented in this research make the current state an
561 additional input and the updated state a part of their output. This
562 abstraction of state and time limits the descriptions to synchronous hardware.
563 However, there is room with the language to eventually add an abstraction
564 mechanism that allows modeling of asynchronous and multi-clock systems.
566 Likewise as with the traditional \acrop{HDL}, descriptions made in a
567 functional \acro{HDL} must eventually be converted into a netlist. This
568 research also features a prototype compiler, which has the same name as the
569 language: \CLaSH\footnote{\CLaSHtiny: \acrotiny{CAES} Language for Synchronous
570 Hardware, where \acrotiny{CAES} % the acronyom of our chair.}
571 is hidden for blind review.}
572 (pronounced: clash). This compiler converts the Haskell code to equivalently
573 behaving synthesizable \VHDL\ code, ready to be converted to an actual netlist
574 format by an (optimizing) \VHDL\ synthesis tool.
576 To the best knowledge of the authors, \CLaSH\ is the only (functional)
577 \acro{HDL} that allows circuit specification to be written in a very concise
578 way and at the same time support such advanced features as polymorphic typing,
579 user-defined higher-order functions and pattern matching.
581 The next section will describe the language elements of \CLaSH, and \Cref{sec:compiler} gives a high-level overview of the \CLaSH\ compiler. \Cref{sec:usecases} discusses two use-cases, a \acro{FIR} filter, and a higher-order \acro{CPU} design. At the end, \Cref{sec:relatedwork} compares \CLaSH\ to existing (functional) \acrop{HDL}, conclusions are given in \Cref{sec:conclusion}, and future work is discussed in \Cref{sec:futurework}.
583 \section{Hardware description in Haskell}
584 This section describes the basic language elements of \CLaSH\ and the support
585 of these elements within the \CLaSH\ compiler. In various subsections, the
586 relation between the language elements and their eventual netlist
587 representation is also highlighted.
589 \subsection{Function application}
590 Two basic elements of a functional program are functions and function
591 application. These have a single obvious translation to a netlist format:
593 \item every function is translated to a component,
594 \item every function argument is translated to an input port,
595 \item the result value of a function is translated to an output port,
597 \item function applications are translated to component instantiations.
599 The result value can have a composite type (such as a tuple), so the fact
600 that a function has just a single result value does not pose any
601 limitation. The actual arguments of a function application are assigned to
602 signals, which are then mapped to the corresponding input ports of the
603 component. The output port of the function is also mapped to a signal,
604 which is used as the result of the application itself. Since every top
605 level function generates its own component, the hierarchy of function
606 calls is reflected in the final netlist.
607 %, creating a hierarchical description of the hardware.
608 % The separation in different components makes it easier for a developer
609 % to understand and possibly hand-optimize the resulting \VHDL\ output of
610 % the \CLaSH\ compiler.
612 The short example below (\ref{code:mac}) gives a demonstration of
613 the conciseness that can be achieved with \CLaSH\ when compared with
614 other (more traditional) \acrop{HDL}. The example is a combinational
615 multiply-accumulate circuit that works for \emph{any} word length (this
616 type of polymorphism will be further elaborated in
617 \Cref{sec:polymorhpism}). The corresponding netlist is depicted in
621 \begin{minipage}{0.93\linewidth}
623 mac a b c = add (mul a b) c
626 \begin{minipage}{0.07\linewidth}
633 \centerline{\includegraphics{mac.svg}}
634 \caption{Combinational Multiply-Accumulate}
639 The use of a composite result value is demonstrated in the next example
640 (\ref{code:mac-composite}), where the multiply-accumulate circuit not only
641 returns the accumulation result, but also the intermediate multiplication
642 result (see \Cref{img:mac-comb-composite}, where the double arrow suggests
643 the composite output).
646 \begin{minipage}{0.93\linewidth}
648 mac a b c = (z, add z c)
653 \begin{minipage}{0.07\linewidth}
655 \label{code:mac-composite}
661 \centerline{\includegraphics{mac-nocurry.svg}}
662 \caption{Combinational Multiply-Accumulate (composite output)}
663 \label{img:mac-comb-composite}
668 In Haskell, choice can be achieved by a large set of syntactic elements,
669 consisting of: \hs{case} expressions, \hs{if-then-else} expressions,
670 pattern matching, and guards. The most general of these are the \hs{case}
671 expressions (\hs{if} expressions can be directly translated to
672 \hs{case} expressions). When transforming a \CLaSH\ description to a
673 netlist, a \hs{case} expression is translated to a multiplexer. The
674 control value of the \hs{case} expression is fed into a number of
675 comparators, and their combined output forms the selection port of the
676 multiplexer. The result of each alternative in the \hs{case} expression is
677 linked to the corresponding input port of the multiplexer.
678 % A \hs{case} expression can in turn simply be translated to a conditional
679 % assignment in \VHDL, where the conditions use equality comparisons
680 % against the constructors in the \hs{case} expressions.
682 % Two versions of a contrived example are displayed below, the first
683 % (\ref{lst:code3}) using a \hs{case} expression and the second
684 % (\ref{lst:code4}) using an \hs{if-then-else} expression. Both examples
685 % sum two values when they are equal or non-equal (depending on the given
686 % predicate, the \hs{pred} variable) and return 0 otherwise.
688 A code example (\ref{code:counter1}) that uses a \hs{case} expression and
689 \hs{if-then-else} expressions is shown below. The function counts up or
690 down depending on the \hs{direction} variable, and has a \hs{bound}
691 variable that determines both the upper bound and wrap-around point of the
692 counter. The \hs{direction} variable is of the following, user-defined,
693 enumeration datatype:
696 data Direction = Up | Down
699 The naive netlist corresponding to this example is depicted in
700 \Cref{img:counter}. Note that the \hs{direction} variable is only
701 compared to \hs{Up}, as an inequality immediately implies that
702 \hs{direction} is \hs{Down} (as derived by the compiler).
705 \begin{minipage}{0.93\linewidth}
707 counter bound direction x = case direction of
708 Up -> if x < bound then
711 Down -> if x > 0 then
716 \begin{minipage}{0.07\linewidth}
718 \label{code:counter1}
723 % \begin{minipage}{0.93\linewidth}
726 % if pred == Equal then
727 % if a == b then a + b else 0
729 % if a != b then a + b else 0
732 % \begin{minipage}{0.07\linewidth}
740 % \centerline{\includegraphics{choice-case.svg}}
741 % \caption{Choice - sumif}
747 \centerline{\includegraphics{counter.svg}}
748 \caption{Counter netlist}
753 A \emph{user-friendly} and also powerful form of choice that is not found
754 in the traditional hardware description languages is pattern matching. A
755 function can be defined in multiple clauses, where each clause corresponds
756 to a pattern. When an argument matches a pattern, the corresponding clause
757 will be used. Expressions can also contain guards, where the expression is
758 only executed if the guard evaluates to true, and continues with the next
759 clause if the guard evaluates to false. Like \hs{if-then-else}
760 expressions, pattern matching and guards have a (straightforward)
761 translation to \hs{case} expressions and can as such be mapped to
762 multiplexers. A second version (\ref{code:counter2}) of the earlier
763 example, now using both pattern matching and guards, can be seen below.
764 The guard is the expression that follows the vertical bar (\hs{|}) and
765 precedes the assignment operator (\hs{=}). The \hs{otherwise} guards
766 always evaluate to \hs{true}.
768 The second version corresponds to the same naive netlist representation
769 (\Cref{img:counter}) as the earlier example.
772 \begin{minipage}{0.93\linewidth}
774 counter bound Up x | x < bound = x + 1
777 counter bound Down x | x > 0 = x - 1
781 \begin{minipage}{0.07\linewidth}
783 \label{code:counter2}
788 % \centerline{\includegraphics{choice-ifthenelse}}
789 % \caption{Choice - \emph{if-then-else}}
794 Haskell is a statically-typed language, meaning that the type of a
795 variable or function is determined at compile-time. Not all of
796 Haskell's typing constructs have a clear translation to hardware, this
797 section therefor only deals with the types that do have a clear
798 correspondence to hardware. The translatable types are divided into two
799 categories: \emph{built-in} types and \emph{user-defined} types. Built-in
800 types are those types for which a fixed translation is defined within the
801 \CLaSH\ compiler. The \CLaSH\ compiler has generic translation rules to
802 translate the user-defined types, which are described later on.
804 Type annotations (entities in \VHDL) are optional, since the \CLaSH\
805 compiler can derive them, when the top-level function \emph{is} annotated
808 % Translation of two most basic functional concepts has been
809 % discussed: function application and choice. Before looking further
810 % into less obvious concepts like higher-order expressions and
811 % polymorphism, the possible types that can be used in hardware
812 % descriptions will be discussed.
814 % Some way is needed to translate every value used to its hardware
815 % equivalents. In particular, this means a hardware equivalent for
816 % every \emph{type} used in a hardware description is needed.
818 % The following types are \emph{built-in}, meaning that their hardware
819 % translation is fixed into the \CLaSH\ compiler. A designer can also
820 % define his own types, which will be translated into hardware types
821 % using translation rules that are discussed later on.
823 \subsubsection{Built-in types}
824 The following types have fixed translations defined within the \CLaSH\
828 the most basic type available. It can have two values:
829 \hs{Low} or \hs{High}.
830 % It is mapped directly onto the \texttt{std\_logic} \VHDL\ type.
832 this is a basic logic type. It can have two values: \hs{True}
834 % It is translated to \texttt{std\_logic} exactly like the \hs{Bit}
835 % type (where a value of \hs{True} corresponds to a value of
837 Supporting the Bool type is required in order to support the
838 \hs{if-then-else} expression.
839 \item[\bf{Signed}, \bf{Unsigned}]
840 these are types to represent integers, and both are parametrizable in
841 their size. The overflow behavior of the numeric operators defined for
842 these types is \emph{wrap-around}.
843 % , so you can define an unsigned word of 32 bits wide as follows:
846 % type Word32 = SizedWord D32
849 % Here, a type synonym \hs{Word32} is defined that is equal to the
850 % \hs{SizedWord} type constructor applied to the type \hs{D32}.
851 % \hs{D32} is the \emph{type level representation} of the decimal
852 % number 32, making the \hs{Word32} type a 32-bit unsigned word. These
853 % types are translated to the \VHDL\ \texttt{unsigned} and
854 % \texttt{signed} respectively.
856 this type can contain elements of any type and has a static length.
857 The \hs{Vector} type constructor takes two arguments: the length of
858 the vector and the type of the elements contained in it. The
859 short-hand notation used for the vector type in the rest of paper is:
860 \hs{[a|n]}, where \hs{a} is the element type, and \hs{n} is the length
862 % Note that this is a notation used in this paper only, vectors are
863 % slightly more verbose in real \CLaSH\ descriptions.
864 % The state type of an 8 element register bank would then for example
868 % type RegisterState = Vector D8 Word32
871 % Here, a type synonym \hs{RegisterState} is defined that is equal to
872 % the \hs{Vector} type constructor applied to the types \hs{D8} (The
873 % type level representation of the decimal number 8) and \hs{Word32}
874 % (The 32 bit word type as defined above). In other words, the
875 % \hs{RegisterState} type is a vector of 8 32-bit words. A fixed size
876 % vector is translated to a \VHDL\ array type.
878 the main purpose of the \hs{Index} type is to be used as an index into
879 a \hs{Vector}, and has an integer range from zero to a specified upper
881 % This means that its range is not limited to powers of two, but
883 If a value of this type exceeds either bounds, an error will be thrown
886 % \comment{TODO: Perhaps remove this example?} To define an index for
887 % the 8 element vector above, we would do:
890 % type RegisterIndex = RangedWord D7
893 % Here, a type synonym \hs{RegisterIndex} is defined that is equal to
894 % the \hs{RangedWord} type constructor applied to the type \hs{D7}. In
895 % other words, this defines an unsigned word with values from
896 % 0 to 7 (inclusive). This word can be be used to index the
897 % 8 element vector \hs{RegisterState} above. This type is translated
898 % to the \texttt{unsigned} \VHDL type.
901 \subsubsection{User-defined types}
902 % There are three ways to define new types in Haskell: algebraic
903 % data-types with the \hs{data} keyword, type synonyms with the \hs{type}
904 % keyword and datatype renaming constructs with the \hs{newtype} keyword.
905 % \GHC\ offers a few more advanced ways to introduce types (type families,
906 % existential typing, {\acro{GADT}}s, etc.) which are not standard
907 % Haskell. As it is currently unclear how these advanced type constructs
908 % correspond to hardware, they are for now unsupported by the \CLaSH\
910 A designer may define a completely new type by an algebraic datatype
911 declaration using the \hs{data} keyword. Type synonyms can be introduced
912 using the \hs{type} keyword.
913 % Only an algebraic datatype declaration actually introduces a
914 % completely new type. Type synonyms and type renaming only define new
915 % names for existing types, where synonyms are completely interchangeable
916 % and a type renaming requires an explicit conversion.
917 Type synonyms do not need any particular translation, as a synonym will
918 use the same representation as the original type.
920 Algebraic datatypes can be categorized as follows:
922 \item[\bf{Single constructor}]
923 datatypes with a single constructor with one or more fields allow
924 values to be packed together in a record-like structure. Haskell's
925 built-in tuple types are also defined as single constructor algebraic
926 types (using a bit of syntactic sugar). An example of a single
927 constructor type with multiple fields is the following pair of
930 data IntPair = IntPair Int Int
932 % These types are translated to \VHDL\ record types, with one field
933 % for every field in the constructor.
934 \item[\bf{Multiple constructors, No fields}]
935 datatypes with multiple constructors, but without any fields are
937 % Note that Haskell's \hs{Bool} type is also defined as an enumeration
938 % type, but that there is a fixed translation for that type within the
940 An example of an enumeration type definition is:
942 data TrafficLight = Red | Orange | Green
944 % These types are translated to \VHDL\ enumerations, with one
945 % value for each constructor. This allows references to these
946 % constructors to be translated to the corresponding enumeration
948 \item[\bf{Multiple constructors with fields}]
949 datatypes with multiple constructors, where at least
950 one of these constructors has one or more fields are currently not
951 supported. Additional research is required to optimize the overlap of
952 fields belonging to the different constructors.
955 \subsection{Polymorphism}\label{sec:polymorhpism}
956 A powerful feature of some programming languages is polymorphism, it
957 allows a function to handle values of different data types in a uniform
958 way. Haskell supports \emph{parametric polymorphism}, meaning that
959 functions can be written without mentioning specific types, and that they
960 can be used for arbitrary types.
962 As an example of a parametric polymorphic function, consider the type of
963 the following \hs{first} function, which returns the first element of a
964 tuple:\footnote{The \hs{::} operator is used to annotate a function
971 This type is parameterized in \hs{a} and \hs{b}, which can both
972 represent any type that is supported by the \CLaSH\ compiler. This means
973 that \hs{first} works for any tuple, regardless of what elements it
974 contains. This kind of polymorphism is extremely useful in hardware
975 designs, for example when routing signals without knowing their exact
976 type, or specifying vector operations that work on vectors of any length
977 and element type. Polymorphism also plays an important role in most higher
978 order functions, as will be shown in the next section.
980 % Another type of polymorphism is \emph{ad-hoc
981 % polymorphism}~\cite{polymorphism}, which refers to polymorphic
982 % functions which can be applied to arguments of different types, but
983 % which behave differently depending on the type of the argument to which
984 % they are applied. In Haskell, ad-hoc polymorphism is achieved through
985 % the use of \emph{type classes}, where a class definition provides the
986 % general interface of a function, and class \emph{instances} define the
987 % functionality for the specific types. An example of such a type class is
988 % the \hs{Num} class, which contains all of Haskell's numerical
989 % operations. A designer can make use of this ad-hoc polymorphism by
990 % adding a \emph{constraint} to a parametrically polymorphic type
991 % variable. Such a constraint indicates that the type variable can only be
992 % instantiated to a type whose members supports the overloaded functions
993 % associated with the type class.
995 Another type of polymorphism is \emph{ad-hoc polymorphism}, which refers
996 to functions that can be applied to arguments of a limited set to types.
997 Furthermore, how such functions work may depend on the type of their
998 arguments. For example, addition only works for numeric types, and it
999 works differently for e.g. integers and complex numbers.
1001 In Haskell, ad-hoc polymorphism is achieved through the use of \emph{type
1002 classes}, where a class definition provides the general interface of a
1003 function, and class \emph{instances} define the functionality for the
1004 specific types. For example, all numeric operators are gathered in the
1005 \hs{Num} class, so every type that wants to use those operators must be
1006 made an instance of \hs{Num}.
1008 By prefixing a type signature with class constraints, the constrained type
1009 parameters are forced to belong to that type class. For example, the
1010 arguments of the \hs{add} function must belong to the \hs{Num} type class
1011 because the \hs{add} function adds them with the (+) operator:
1014 add :: Num a => a -> a -> a
1018 % An example of a type signature that includes such a constraint if the
1019 % signature of the \hs{sum} function, which sums the values in a vector:
1021 % sum :: Num a => [a|n] -> a
1024 % This type is again parameterized by \hs{a}, but it can only contain
1025 % types that are \emph{instances} of the \emph{type class} \hs{Num}, so
1026 % that the compiler knows that the addition (+) operator is defined for
1029 % A place where class constraints also play a role is in the size and
1030 % range parameters of the \hs{Vector} and numeric types. The reason being
1031 % that these parameters have to be limited to types that can represent
1032 % \emph{natural} numbers. The complete type of for example the \hs{Vector}
1035 % Natural n => Vector n a
1038 % \CLaSH's built-in numerical types are also instances of the \hs{Num}
1040 % so we can use the addition operator (and thus the \hs{sum}
1041 % function) with \hs{Signed} as well as with \hs{Unsigned}.
1043 \CLaSH\ supports both parametric polymorphism and ad-hoc polymorphism. A
1044 circuit designer can specify his own type classes and corresponding
1045 instances. The \CLaSH\ compiler will infer the type of every polymorphic
1046 argument depending on how the function is applied. There is however one
1047 constraint: the top level function that is being translated can not have
1048 polymorphic arguments. The arguments of the top-level can not be
1049 polymorphic as there is no way to infer the \emph{specific} types of the
1052 With regard to the built-in types, it should be noted that members of
1053 some of the standard Haskell type classes are supported as built-in
1054 functions. These include: the numerial operators of \hs{Num}, the equality
1055 operators of \hs{Eq}, and the comparison (order) operators of \hs{Ord}.
1057 \subsection{Higher-order functions \& values}
1058 Another powerful abstraction mechanism in functional languages, is
1059 the concept of \emph{functions as a first class value} and
1060 \emph{higher-order functions}. These concepts allows a function to be
1061 treated as a value and be passed around, even as the argument of another
1062 function. The following example clarifies this concept:
1065 \begin{minipage}{0.93\linewidth}
1066 %format not = "\mathit{not}"
1068 negate{-"\!\!\!"-}Vector xs = map not xs
1071 \begin{minipage}{0.07\linewidth}
1073 \label{code:negatevector}
1077 The code above defines the \hs{negate{-"\!\!\!"-}Vector} function, which
1078 takes a vector of booleans, \hs{xs}, and returns a vector where all the
1079 values are negated. It achieves this by calling the \hs{map} function, and
1080 passing it another \emph{function}, boolean negation, and the vector of
1081 booleans, \hs{xs}. The \hs{map} function applies the negation function to
1082 all the elements in the vector.
1084 The \hs{map} function is called a higher-order function, since it takes
1085 another function as an argument. Also note that \hs{map} is again a
1086 parametric polymorphic function: it does not pose any constraints on the
1087 type of the input vector, other than that its elements must have the same
1088 type as the first argument of the function passed to \hs{map}. The element
1089 type of the resulting vector is equal to the return type of the function
1090 passed, which need not necessarily be the same as the element type of the
1091 input vector. All of these characteristics can be inferred from the type
1092 signature of \hs{map}:
1095 map :: (a -> b) -> [a|n] -> [b|n]
1098 In Haskell, there are two more ways to obtain a function-typed value:
1099 partial application and lambda abstraction. Partial application means that
1100 a function that takes multiple arguments can be applied to a single
1101 argument, and the result will again be a function, but takes one argument
1102 less. As an example, consider the following expression, that adds one to
1103 every element of a vector:
1106 \begin{minipage}{0.93\linewidth}
1111 \begin{minipage}{0.07\linewidth}
1113 \label{code:partialapplication}
1117 Here, the expression \hs{(add 1)} is the partial application of the
1118 addition function to the value \hs{1}, which is again a function that
1119 adds 1 to its (next) argument.
1121 A lambda expression allows a designer to introduce an anonymous function
1122 in any expression. Consider the following expression, which again adds 1
1123 to every element of a vector:
1126 \begin{minipage}{0.93\linewidth}
1128 map (\x -> x + 1) xs
1131 \begin{minipage}{0.07\linewidth}
1133 \label{code:lambdaexpression}
1137 Finally, not only built-in functions can have higher order arguments (such
1138 as the \hs{map} function), but any function defined in \CLaSH\ may have
1139 functions as arguments. This allows the circuit designer to apply a
1140 large amount of code reuse. The only exception is again the top-level
1141 function: if a function-typed argument is not instantiated with an actual
1142 function, no hardware can be generated.
1144 An example of a common circuit where higher-order functions and partial
1145 application lead to a very concise and natural description is a crossbar.
1146 The code (\ref{code:crossbar}) for this example can be seen below:
1149 \begin{minipage}{0.93\linewidth}
1151 crossbar inputs selects = map (mux inputs) selects
1153 mux inp x = (inp ! x)
1156 \begin{minipage}{0.07\linewidth}
1158 \label{code:crossbar}
1162 The the \hs{crossbar} function selects those values from \hs{inputs} that
1163 are indicated by the indexes in the vector \hs{selects}. The crossbar is
1164 polymorphic in the width of the input (defined by the length of
1165 \hs{inputs}), the width of the output (defined by the length of
1166 \hs{selects}), and the signal type (defined by the element type of
1167 \hs{inputs}). The type-checker can also automatically infer that
1168 \hs{selects} is a vector of \hs{Index} values due to the use of the vector
1169 indexing operator (\hs{!}).
1172 In a stateful design, the outputs depend on the history of the inputs, or
1173 the state. State is usually stored in registers, which retain their value
1174 during a clock cycle.
1175 % As \CLaSH\ has to be able to describe more than plain combinational
1176 % designs, there is a need for an abstraction mechanism for state.
1178 An important property in Haskell, and in many other functional languages,
1179 is \emph{purity}. A function is said to be \emph{pure} if it satisfies two
1182 \item given the same arguments twice, it should return the same value in
1184 \item that the function has no observable side-effects.
1186 % This purity property is important for functional languages, since it
1187 % enables all kinds of mathematical reasoning that could not be guaranteed
1188 % correct for impure functions.
1189 Pure functions are a perfect match for combinational circuits, where the
1190 output solely depends on the inputs. When a circuit has state however, it
1191 can no longer be described by a pure function.
1192 % Simply removing the purity property is not a valid option, as the
1193 % language would then lose many of it mathematical properties.
1194 \CLaSH\ deals with the concept of state by making the current state an
1195 additional argument of the function, and the updated state part of the
1196 result. In this sense the descriptions made in \CLaSH\ are the
1197 combinational parts of a mealy machine.
1199 A simple example is adding an accumulator register to the earlier
1200 multiply-accumulate circuit, of which the resulting netlist can be seen in
1201 \Cref{img:mac-state}:
1204 \begin{minipage}{0.93\linewidth}
1206 macS (State c) (a, b) = (State c', c')
1211 \begin{minipage}{0.07\linewidth}
1213 \label{code:macstate}
1217 Note that the \hs{macS} function returns both the new state and the value
1218 of the output port. The \hs{State} wrapper indicates which arguments are
1219 part of the current state, and what part of the output is part of the
1220 updated state. This aspect will also be reflected in the type signature of
1221 the function. Abstracting the state of a circuit in this way makes it very
1222 explicit: which variables are part of the state is completely determined
1223 by the type signature. This approach to state is well suited to be used in
1224 combination with the existing code and language features, such as all the
1225 choice elements, as state values are just normal values from Haskell's
1226 point of view. Stateful descriptions are simulated using the recursive
1230 \begin{minipage}{0.93\linewidth}
1232 run f s (i : inps) = o : (run f s' inps)
1237 \begin{minipage}{0.07\linewidth}
1243 The \hs{(:)} operator is the list concatenation operator, where the
1244 left-hand side is the head of a list and the right-hand side is the
1245 remainder of the list. The \hs{run} function applies the function the
1246 developer wants to simulate, \hs{f}, to the current state, \hs{s}, and the
1247 first input value, \hs{i}. The result is the first output value, \hs{o},
1248 and the updated state \hs{s'}. The next iteration of the \hs{run} function
1249 is then called with the updated state, \hs{s'}, and the rest of the
1250 inputs, \hs{inps}. In the context of this paper, it is assumed that there
1251 is one input per clock cycle. Note that the order of \hs{s',o,s,i} in the
1252 where clause of the \hs{run} functions corresponds with the order of the
1253 input, output and state of the \hs{macS} function (\ref{code:macstate}).
1254 Thus, in Haskell the expression \hs{run macS 0 inputs} simulates \hs{macS}
1255 on \hs{inputs} starting with the value \hs{0}
1258 \centerline{\includegraphics{mac-state.svg}}
1259 \caption{Stateful Multiply-Accumulate}
1260 \label{img:mac-state}
1264 The complete simulation can be compiled to an executable binary by a
1265 Haskell compiler, or executed in an Haskell interpreter. Both
1266 simulation paths require less effort from a circuit designer than first
1267 translating the description to \VHDL\ and then running a \VHDL\
1268 simulation; it is also very likely that both simulation paths are much
1271 \section{The \CLaSH\ compiler}
1272 \label{sec:compiler}
1273 The prototype \CLaSH\ compiler translates descriptions made in the \CLaSH\
1274 language as described in the previous section to synthesizable \VHDL.
1275 % , allowing a designer to actually run a \CLaSH\ design on an \acro{FPGA}.
1277 The Glasgow Haskell Compiler (\GHC)~\cite{ghc} is an open source Haskell
1278 compiler that also provides a high level \acro{API} to most of its internals.
1279 Furthermore, it provides several parts of the prototype compiler for free,
1280 such as the parser, the semantics checker, and the type checker. These parts
1281 together form the front-end of the prototype compiler pipeline, as seen in
1282 \Cref{img:compilerpipeline}.
1286 \centerline{\includegraphics{compilerpipeline.svg}}
1287 \caption{\CLaSHtiny\ compiler pipeline}
1288 \label{img:compilerpipeline}
1292 The output of the \GHC\ front-end consists of the translation of the original
1293 Haskell description to \emph{Core}~\cite{Sulzmann2007}, which is a small
1294 typed functional language. This \emph{Core} language is relatively easy to
1295 process compared to the larger Haskell language. A description in \emph{Core}
1296 can still contain elements which have no direct translation to hardware, such
1297 as polymorphic types and function-valued arguments. Such a description needs
1298 to be transformed to a \emph{normal form}, which corresponds directly to
1299 hardware. The second stage of the compiler, the \emph{normalization} phase,
1300 exhaustively applies a set of \emph{meaning-preserving} transformations on the
1301 \emph{Core} description until this description is in a \emph{normal form}.
1302 This set of transformations includes transformations typically found in
1303 reduction systems and lambda calculus~\cite{lambdacalculus}, such as
1304 $\beta$-reduction and $\eta$-expansion. It also includes transformations that
1305 are responsible for the specialization of higher-order functions to `regular'
1306 first-order functions, and specializing polymorphic types to concrete types.
1308 The final step in the compiler pipeline is the translation to a \VHDL\
1309 \emph{netlist}, which is a straightforward process due to the resemblance of a
1310 normalized description and a set of concurrent signal assignments. The
1311 end-product of the \CLaSH\ compiler is called a \VHDL\ \emph{netlist} as the
1312 result resembles an actual netlist description, and the fact that it is \VHDL\
1313 is only an implementation detail; e.g., the output could have been Verilog or
1317 \label{sec:usecases}
1318 \subsection{FIR Filter}
1319 As an example of a common hardware design where the relation between
1320 functional languages and mathematical functions, combined with the use of
1321 higher-order functions leads to a very natural description is a \acro{FIR}
1325 y_t = \sum\nolimits_{i = 0}^{n - 1} {x_{t - i} \cdot h_i }
1328 A \acro{FIR} filter multiplies fixed constants ($h$) with the current
1329 and a few previous input samples ($x$). Each of these multiplications
1330 are summed, to produce the result at time $t$. The equation of a \acro{FIR}
1331 filter is equivalent to the equation of the dot-product of two vectors, which
1335 \mathbf{a}\bullet\mathbf{b} = \sum\nolimits_{i = 0}^{n - 1} {a_i \cdot b_i }
1338 The equation for the dot-product is easily and directly implemented using
1339 higher-order functions:
1342 \begin{minipage}{0.93\linewidth}
1344 as *+* bs = fold (+) (zipWith (*) as bs)
1347 \begin{minipage}{0.07\linewidth}
1349 \label{code:dotproduct}
1353 The \hs{zipWith} function is very similar to the \hs{map} function seen
1354 earlier: It takes a function, two vectors, and then applies the function to
1355 each of the elements in the two vectors pairwise (\emph{e.g.}, \hs{zipWith (*)
1356 [1, 2] [3, 4]} becomes \hs{[1 * 3, 2 * 4]}).
1358 The \hs{fold} function takes a binary function, a single vector, and applies
1359 the function to the first two elements of the vector. It then applies the
1360 function to the result of the first application and the next element in the
1361 vector. This continues until the end of the vector is reached. The result of
1362 the \hs{fold} function is the result of the last application. It is obvious
1363 that the \hs{zipWith (*)} function is pairwise multiplication and that the
1364 \hs{fold (+)} function is summation.
1365 % Returning to the actual \acro{FIR} filter, we will slightly change the
1366 % equation describing it, so as to make the translation to code more obvious and
1367 % concise. What we do is change the definition of the vector of input samples
1368 % and delay the computation by one sample. Instead of having the input sample
1369 % received at time $t$ stored in $x_t$, $x_0$ now always stores the newest
1370 % sample, and $x_i$ stores the $ith$ previous sample. This changes the equation
1371 % to the following (note that this is completely equivalent to the original
1372 % equation, just with a different definition of $x$ that will better suit the
1373 % transformation to code):
1376 % y_t = \sum\nolimits_{i = 0}^{n - 1} {x_i \cdot h_i }
1378 The complete definition of the \acro{FIR} filter in \CLaSH\ is:
1381 \begin{minipage}{0.93\linewidth}
1383 fir (State (xs,hs)) x =
1384 (State (shiftInto x xs,hs), (x +> xs) *+* hs)
1387 \begin{minipage}{0.07\linewidth}
1393 where the vector \hs{xs} contains the previous input samples, the vector
1394 \hs{hs} contains the \acro{FIR} coefficients, and \hs{x} is the current input
1395 sample. The concatenate operator (\hs{+>}) creates a new vector by placing the
1396 current sample (\hs{x}) in front of the previous samples vector (\hs{xs}). The
1397 code for the \hs{shiftInto} function, that adds the new input sample (\hs{x})
1398 to the list of previous input samples (\hs{xs}) and removes the oldest sample,
1402 \begin{minipage}{0.93\linewidth}
1404 shiftInto x xs = x +> init xs
1407 \begin{minipage}{0.07\linewidth}
1409 \label{code:shiftinto}
1413 where the \hs{init} function returns all but the last element of a vector.
1414 The resulting netlist of a 4-taps \acro{FIR} filter, created by specializing
1415 the vectors of the \acro{FIR} code to a length of 4, is depicted in
1419 \centerline{\includegraphics{4tapfir.svg}}
1420 \caption{4-taps \acrotiny{FIR} Filter}
1425 \subsection{Higher-order CPU}
1426 %format fun x = "\textit{fu}_" x
1427 This section discusses a somewhat more serious example in which user-defined
1428 higher-order function, partial application, lambda expressions, and pattern
1429 matching are exploited. The example concerns a \acro{CPU} which consists of
1430 four function units, \hs{fun 0,{-"\ldots"-},fun 3}, (see \Cref{img:highordcpu}) that each perform some binary operation.
1433 \centerline{\includegraphics{highordcpu.svg}}
1434 \caption{CPU with higher-order Function Units}
1435 \label{img:highordcpu}
1439 Every function unit has seven data inputs (of type \hs{Signed 16}), and two
1440 address inputs (of type \hs{Index 6}) that indicate which data inputs have to
1441 be chosen as arguments for the binary operation that the unit performs.
1442 These data inputs consists of one external input \hs{x}, two fixed
1443 initialization values (0 and 1), and the previous outputs of the four function
1444 units. The output of the \acro{CPU} as a whole is the previous output of
1447 The function units \hs{fun 1, fun 2}, and \hs{fun 3} can perform a fixed binary operation, whereas \hs{fun 0} has an additional input for an opcode to choose a binary operation out of a few possibilities. Each function unit outputs its result into a register, i.e., the state of the \acro{CPU}. This state can e.g. be defined as follows:
1450 type CpuState = State [Signed 16 | 4]
1453 Every function unit can now be defined by the following higher-order function
1454 \hs{fu}, which takes three arguments: the operation \hs{op} that the function
1455 unit performs, the seven \hs{inputs}, and the address pair \hs{(a1,a2)}:
1458 \begin{minipage}{0.93\linewidth}
1460 fu op inputs (a1, a2) = regIn
1464 regIn = op arg1 arg2
1467 \begin{minipage}{0.07\linewidth}
1469 \label{code:functionunit}
1473 \noindent Using partial application we now define:
1476 \begin{minipage}{0.93\linewidth}
1483 \begin{minipage}{0.07\linewidth}
1485 \label{code:functionunits1to3}
1489 In order to define \hs{fun 0}, the \hs{Opcode} type, and the \hs{multiop} functions that chooses a specific operation given the opcode, are defined first. It is assumed that the functions \hs{shifts} (which shifts its first
1490 operand by the number of bits indicate in the second operand), \hs{xor} (for
1491 the bitwise \hs{xor}), and (==) (for equality) already exist.
1494 \begin{minipage}{0.93\linewidth}
1496 data Opcode = Shift | Xor | Equal
1498 multiop Shift = shift
1500 multiop Equal = \a b -> if a == b then 1 else 0
1503 \begin{minipage}{0.07\linewidth}
1505 \label{code:multiop}
1509 Note that the result of \hs{multiop} is a binary function; this is supported
1510 by \CLaSH. The complete definition of \hs{fun 0}, which takes an opcode as
1511 additional argument, is:
1514 \begin{minipage}{0.93\linewidth}
1516 fun 0 c = fu (multiop c)
1519 \begin{minipage}{0.07\linewidth}
1521 \label{code:functionunit0}
1525 \noindent Now comes the definition of the full \acro{CPU}. Its type is:
1529 -> (Word, Opcode, [(Index 6, Index 6) | 4])
1533 \noindent Note that this type fits the requirements of the \hs{run} function.
1534 The definition of the \hs{cpu} function now is:
1537 \begin{minipage}{0.93\linewidth}
1539 cpu (State s) (x,opc,addrs) = (State s', out)
1541 inputs = x +> (0 +> (1 +> s))
1542 s' = [{-"\;"-}fun 0 opc inputs (addrs!0)
1543 ,{-"\;"-}fun 1 inputs (addrs!1)
1544 ,{-"\;"-}fun 2 inputs (addrs!2)
1545 ,{-"\;"-}fun 3 inputs (addrs!3)
1550 \begin{minipage}{0.07\linewidth}
1556 While this is still a simple (and maybe not very useful) design, it
1557 illustrates some possibilities that \CLaSH\ offers and suggests how to write
1560 % Each of the function units has both its operands connected to all data
1561 % sources, and can be programmed to select any data source for either
1562 % operand. In addition, the leftmost function unit has an additional
1563 % opcode input to select the operation it performs. The previous output of the
1564 % rightmost function unit is the output of the entire \acro{CPU}.
1566 % The code of the function unit (\ref{code:functionunit}), which arranges the
1567 % operand selection for the function unit, is shown below. Note that the actual
1568 % operation that takes place inside the function unit is supplied as the
1569 % (higher-order) argument \hs{op}, which is a function that takes two arguments.
1573 % The \hs{multiop} function (\ref{code:multiop}) defines the operation that takes place in the leftmost function unit. It is essentially a simple three operation \acro{ALU} that makes good use of pattern matching and guards in its description. The \hs{shift} function used here shifts its first operand by the number of bits indicated in the second operand, the \hs{xor} function produces
1574 % the bitwise xor of its operands.
1577 % The \acro{CPU} function (\ref{code:cpu}) ties everything together. It applies
1578 % the function unit (\hs{fu}) to several operations, to create a different
1579 % function unit each time. The first application is interesting, as it does not
1580 % just pass a function to \hs{fu}, but a partial application of \hs{multiop}.
1581 % This demonstrates how one function unit can effectively get extra inputs
1582 % compared to the others.
1584 % The vector \hs{inputs} is the set of data sources, which is passed to
1585 % each function unit as a set of possible operants. The \acro{CPU} also receives
1586 % a vector of address pairs, which are used by each function unit to select
1588 % The application of the function units to the \hs{inputs} and
1589 % \hs{addrs} arguments seems quite repetitive and could be rewritten to use
1590 % a combination of the \hs{map} and \hs{zipwith} functions instead.
1591 % However, the prototype compiler does not currently support working with
1592 % lists of functions, so a more explicit version of the code is given instead.
1594 % While this is still a simple example, it could form the basis of an actual
1595 % design, in which the same techniques can be reused.
1597 \section{Related work}
1598 \label{sec:relatedwork}
1599 This section describes the features of existing (functional) hardware
1600 description languages and highlights the advantages that \CLaSH\ has
1603 % Many functional hardware description languages have been developed over the
1604 % years. Early work includes such languages as $\mu$\acro{FP}~\cite{muFP}, an
1605 % extension of Backus' \acro{FP} language to synchronous streams, designed
1606 % particularly for describing and reasoning about regular circuits. The
1607 % Ruby~\cite{Ruby} language uses relations, instead of functions, to describe
1608 % circuits, and has a particular focus on layout.
1610 \acro{HML}~\cite{HML2} is a hardware modeling language based on the strict
1611 functional language \acro{ML}, and has support for polymorphic types and
1612 higher-order functions. There is no direct simulation support for \acro{HML},
1613 so a description in \acro{HML} has to be translated to \VHDL\ and the
1614 translated description can then be simulated in a \VHDL\ simulator. Certain
1615 aspects of HML, such as higher-order functions are however not supported by
1616 the \VHDL\ translator~\cite{HML3}. The \CLaSH\ compiler on the other hand can
1617 correctly translate all of its language constructs.
1619 Like the research presented in this paper, many functional hardware
1620 description languages have a foundation in the functional programming language
1621 Haskell. Hawk~\cite{Hawk1} is a hardware modeling language embedded in Haskell
1622 and has sequential environments that make it easier to specify stateful
1623 computation (by using the \acro{ST} monad). Hawk specifications can be
1624 simulated; to the best knowledge of the authors there is however no support
1625 for automated circuit synthesis.
1627 The ForSyDe~\cite{ForSyDe2} system uses Haskell to specify abstract system
1628 models. A designer can model systems using heterogeneous models of
1629 computation, which include continuous time, synchronous and untimed models of
1630 computation. Using so-called domain interfaces a designer can simulate
1631 electronic systems which have both analog and digital parts. ForSyDe has
1632 several backends including simulation and automated synthesis, though
1633 automated synthesis is restricted to the synchronous model of computation.
1634 Though ForSyDe offers higher-order functions and polymorphism, ForSyDe's
1635 choice elements are limited to \hs{if} and \hs{case} expressions. ForSyDe's
1636 explicit conversions, where functions have to be wrapped in processes and
1637 processes have to be wrapped in systems, combined with the explicit
1638 instantiations of components, also makes ForSyDe far more verbose than \CLaSH.
1640 Lava~\cite{Lava,kansaslava} is a hardware description language embedded in
1641 Haskell which focuses on the structural representation of hardware. Like
1642 \CLaSH, Lava has support for polymorphic types and higher-order functions.
1643 Besides support for simulation and circuit synthesis, Lava descriptions can be
1644 interfaced with formal method tools for formal verification. As discussed in
1645 the introduction, taking the embedded language approach does not allow for
1646 Haskell's choice elements to be captured within the circuit descriptions. In
1647 this respect \CLaSH\ differs from Lava, in that all of Haskell's choice
1648 elements, such as \hs{case}-expressions and pattern matching, are synthesized
1649 to choice elements in the eventual circuit. Consequently, descriptions
1650 containing rich control structures can be specified in a more user-friendly
1651 way in \CLaSH\ than possible within Lava, and hence are less error-prone.
1653 Bluespec~\cite{Bluespec} is a high-level synthesis language that features
1654 guarded atomic transactions and allows for the automated derivation of control
1655 structures based on these atomic transactions. Bluespec, like \CLaSH, supports
1656 polymorphic typing and function-valued arguments. Bluespec's syntax and
1657 language features \emph{had} their basis in Haskell. However, in order to
1658 appeal to the users of the traditional \acrop{HDL}, Bluespec has adapted
1659 imperative features and a syntax that resembles Verilog. As a result, Bluespec
1660 is (unnecessarily) verbose when compared to \CLaSH.
1662 The merits of polymorphic typing and function-valued arguments are now also
1663 recognized in the traditional \acrop{HDL}, exemplified by the new \VHDL-2008
1664 standard~\cite{VHDL2008}. \VHDL-2008 support for generics has been extended to
1665 types and subprograms, allowing a designer to describe components with
1666 polymorphic ports and function-valued arguments. Note that the types and
1667 subprograms still require an explicit generic map, while the \CLaSH\ compiler
1668 automatically infers types, and automatically propagates function-valued
1669 arguments. There are also no (generally available) \VHDL\ synthesis tools that
1670 currently support the \VHDL-2008 standard.
1672 % Wired~\cite{Wired},, T-Ruby~\cite{T-Ruby}, Hydra~\cite{Hydra}.
1674 % A functional language designed specifically for hardware design is
1675 % $re{\mathit{FL}}^{ect}$~\cite{reFLect}, which draws experience from earlier
1676 % language called \acro{FL}~\cite{FL} to la
1678 % An example of a floating figure using the graphicx package.
1679 % Note that \label must occur AFTER (or within) \caption.
1680 % For figures, \caption should occur after the \includegraphics.
1681 % Note that IEEEtran v1.7 and later has special internal code that
1682 % is designed to preserve the operation of \label within \caption
1683 % even when the captionsoff option is in effect. However, because
1684 % of issues like this, it may be the safest practice to put all your
1685 % \label just after \caption rather than within \caption{}.
1687 % Reminder: the "draftcls" or "draftclsnofoot", not "draft", class
1688 % option should be used if it is desired that the figures are to be
1689 % displayed while in draft mode.
1693 %\includegraphics[width=2.5in]{myfigure}
1694 % where an .eps filename suffix will be assumed under latex,
1695 % and a .pdf suffix will be assumed for pdflatex; or what has been declared
1696 % via \DeclareGraphicsExtensions.
1697 %\caption{Simulation Results}
1701 % Note that IEEE typically puts floats only at the top, even when this
1702 % results in a large percentage of a column being occupied by floats.
1705 % An example of a double column floating figure using two subfigures.
1706 % (The subfig.sty package must be loaded for this to work.)
1707 % The subfigure \label commands are set within each subfloat command, the
1708 % \label for the overall figure must come after \caption.
1709 % \hfil must be used as a separator to get equal spacing.
1710 % The subfigure.sty package works much the same way, except \subfigure is
1711 % used instead of \subfloat.
1713 %\begin{figure*}[!t]
1714 %\centerline{\subfloat[Case I]\includegraphics[width=2.5in]{subfigcase1}%
1715 %\label{fig_first_case}}
1717 %\subfloat[Case II]{\includegraphics[width=2.5in]{subfigcase2}%
1718 %\label{fig_second_case}}}
1719 %\caption{Simulation results}
1723 % Note that often IEEE papers with subfigures do not employ subfigure
1724 % captions (using the optional argument to \subfloat), but instead will
1725 % reference/describe all of them (a), (b), etc., within the main caption.
1728 % An example of a floating table. Note that, for IEEE style tables, the
1729 % \caption command should come BEFORE the table. Table text will default to
1730 % \footnotesize as IEEE normally uses this smaller font for tables.
1731 % The \label must come after \caption as always.
1734 %% increase table row spacing, adjust to taste
1735 %\renewcommand{\arraystretch}{1.3}
1736 % if using array.sty, it might be a good idea to tweak the value of
1737 % \extrarowheight as needed to properly center the text within the cells
1738 %\caption{An Example of a Table}
1739 %\label{table_example}
1741 %% Some packages, such as MDW tools, offer better commands for making tables
1742 %% than the plain LaTeX2e tabular which is used here.
1743 %\begin{tabular}{|c||c|}
1753 % Note that IEEE does not put floats in the very first column - or typically
1754 % anywhere on the first page for that matter. Also, in-text middle ("here")
1755 % positioning is not used. Most IEEE journals/conferences use top floats
1756 % exclusively. Note that, LaTeX2e, unlike IEEE journals/conferences, places
1757 % footnotes above bottom floats. This can be corrected via the \fnbelowfloat
1758 % command of the stfloats package.
1762 \section{Conclusion}
1763 \label{sec:conclusion}
1764 This research demonstrates once more that functional languages are well suited
1765 for hardware descriptions: function applications provide an elegant notation
1766 for component instantiation. While circuit descriptions made in \CLaSH\ are
1767 very concise when compared to other (traditional) \acrop{HDL}, their intended
1768 functionality remains clear. \CLaSH\ goes beyond the existing (functional)
1769 hardware descriptions languages by including advanced choice elements, such as
1770 pattern matching and guards, which are well suited to describe the conditional
1771 assignments in control-oriented circuits. Besides being able to translate
1772 these basic constructs to synthesizable \VHDL, the prototype compiler can also
1773 translate descriptions that contain both polymorphic types and user-defined
1774 higher-order functions.
1776 % Where recent functional hardware description languages have mostly opted to
1777 % embed themselves in an existing functional language, this research features
1778 % a `true' compiler. As a result there is a clear distinction between
1779 % compile-time and run-time, which allows a myriad of choice constructs to be
1780 % part of the actual circuit description; a feature the embedded hardware
1781 % description languages do not offer.
1783 Besides simple circuits such as variants of both the \acro{FIR} filter and
1784 the higher-order \acro{CPU} shown in \Cref{sec:usecases}, the \CLaSH\ compiler
1785 has also been able to translate non-trivial functional descriptions such as a
1786 streaming reduction circuit~\cite{blindreview} %~\cite{reductioncircuit}
1787 for floating point numbers.
1789 \section{Future Work}
1790 \label{sec:futurework}
1791 The choice of describing state explicitly as and extra argument and result can
1792 be seen as a mixed blessing. Even though descriptions that use state are
1793 usually very clear, distributing and collecting substate can become tedious
1794 and even error-prone. Automating the required distribution and collection, or
1795 finding a more suitable abstraction mechanism for state would make \CLaSH\
1796 easier to use. Currently, one of the examined approaches to suppress state in
1797 the specification is by using Haskell's arrow-abstraction.
1799 The transformations in the normalization phase of the prototype compiler are
1800 developed in an ad-hoc manner, which makes the existence of many desirable
1801 properties unclear. Such properties include whether the complete set of
1802 transformations will always lead to a normal form or whether the normalization
1803 process always terminates. Though extensive use of the compiler suggests that
1804 these properties usually hold, they have not been formally proven. A
1805 systematic approach to defining the set of transformations allows one to proof
1806 that the earlier mentioned properties do indeed hold.
1808 % conference papers do not normally have an appendix
1811 % use section* for acknowledgement
1812 % \section*{Acknowledgment}
1814 % The authors would like to thank...
1816 % trigger a \newpage just before the given reference
1817 % number - used to balance the columns on the last page
1818 % adjust value as needed - may need to be readjusted if
1819 % the document is modified later
1820 % \IEEEtriggeratref{14}
1821 % The "triggered" command can be changed if desired:
1822 %\IEEEtriggercmd{\enlargethispage{-5in}}
1824 % references section
1826 % can use a bibliography generated by BibTeX as a .bbl file
1827 % BibTeX documentation can be easily obtained at:
1828 % http://www.ctan.org/tex-archive/biblio/bibtex/contrib/doc/
1829 % The IEEEtran BibTeX style support page is at:
1830 % http://www.michaelshell.org/tex/ieeetran/bibtex/
1831 \bibliographystyle{IEEEtran}
1832 % argument is your BibTeX string definitions and bibliography database(s)
1833 \bibliography{clash}
1835 % <OR> manually copy in the resultant .bbl file
1836 % set second argument of \begin to the number of references
1837 % (used to reserve space for the reference number labels box)
1838 % \begin{thebibliography}{1}
1840 % \bibitem{IEEEhowto:kopka}
1841 % H.~Kopka and P.~W. Daly, \emph{A Guide to \LaTeX}, 3rd~ed.\hskip 1em plus
1842 % 0.5em minus 0.4em\relax Harlow, England: Addison-Wesley, 1999.
1844 % \end{thebibliography}
1852 % vim: set ai sw=2 sts=2 expandtab: