Atom (programming language): Difference between revisions

Revision as of 02:26, 9 August 2012

Revision as of 23:11, 8 January 2013

Line 29:

and [[Bluespec]], Atom compiled circuit descriptions, that were based on guarded atomic operations, or [[Term Rewriting System|conditional term rewriting]], into [[Verilog]] [[netlist]]s for simulation and [[logic synthesis]].

As a hardware compiler, Atom's primary objective was to maximize the number of operations, or rules, that can execute in a given clock cycle without violating the semantics of atomic operation.

By employing the properties of conflict-free and sequentially-composable rules,<ref name="hoe1"/>, Atom reduced maximizing execution concurrency to a [[feedback arc set]] optimization of a rule-data dependency graph. This process was similar to James Hoe's original algorithm.<ref name="hoe1" />.

When Atom's author switched careers in late 2007 from logic design to [[Embedded systems|embedded software]] engineering, Atom was redesigned from an [[hardware description language|HDL]] to a [[___domain specific language]]

targeting [[Realtime computing|hard realtime]] embedded applications. As a result, Atom's compiler's primary objective changed from maximizing rule concurrency to balancing processing load and minimizing

[[Worst case execution time|worse case timing latency]]. In September 2008, Atom was presented at CUFP,<ref>[http://cufp.galois.com/2008/schedule.html Controlling Hybrid Vehicles with Haskell.]</ref>,

and in April 2009, was released as open-source in its new form.<ref>[http://www.haskell.org/pipermail/haskell-cafe/2009-April/060602.html ANN: atom-0.0.2]</ref>

Line 40:

Atom is a concurrent programming language intended for embedded applications. Atom features [[Compile time|compile-time]] [[Scheduling (computing)|task scheduling]] and generates code with deterministic execution time and memory consumption, simplifying [[worst case execution time]] analysis for applications that require [[Realtime computing|hard realtime]] performance. Atom's concurrency model is that of [[Atomic action|guarded atomic actions]], which eliminates the need for, and the [[Lock_(computer_science)#The_problems_with_locks|problems]] of using, [[Lock (computer science)|mutex locks]].

By removing [[Run time (program lifecycle phase)|run-time]] task scheduling and mutex locking—two services traditionally served by an [[Real-time operating system|RTOS]] -- Atom—Atom can eliminate the need and overhead of an [[Real-time operating system|RTOS]] in embedded applications.

== Examples ==

{{Empty section|date=July 2010}}

== Limitations ==