Dataflow

Figure 4 Dataflow diagram- Single compound design...

Blocking assignments mirror the dataflow in a combinational circuit. Consider the following always statement. 

All the assignments are blocking assignments. Statements within the sequential block imply to compute the value of TM first, then execute the second statement, assign to 77V, then go to third statement, assign to TO, and so on. This mimics the dataflow through the combinational logic. 

The type of interim measure implemented should be appropriate to the computer functionality being addressed. Computer funetionality being addressed should be mapped so that appropriate interim measures ean be identified. The mapping should include both a workflow analysis and a dataflow analysis. Controls that are already in place may provide the basis for the interim measures. Critieal aetivities that should be given particular consideration for interim measures include ... 

File View Options Layout Malt) Scenario Edit Dataflow Transitions Evolution Session Resources Assignment Views Test Remote Unmatched... 

EHH] Private task O Input, Output o Monitored task — Control flow o Remote task —> Dataflow... 

I I Monitored task —Control flow I I Remote task —Dataflow... 

I I Remote task Output —> Dataflow 2 VVaifing Suspended... 

Some authors [WF1] have drawn a contrast between data-centric scientific workflows (or dataflows) and task-oriented business workflows. In a scientific workflow, the flow of control is tied to the flow of data. In a business workflow, control flows between tasks that operate upon workflow objects, but the flow of control is independent of the data flow. Consider the difference between a workflow designed to extract novel compounds from a candidate data set and one designed to help a customer order a concert ticket. 

The application program interface (API) to SClRun is the visual dataflow environment called the network editor. Within the network editor, programs can be visually assembled from the hbrary of available algorithms. The dataflow network for a sample bioelectric field simulation is shown in Figure 23.1. 

The boxes in the network are called modules, and the lines connecting them are called datapipes. The point of attachment, where a datapipe attaches to a module, is called a datapoit, the dataports on the tops of the modules are input ports, and the ports on the bottoms of the modules are output ports. In SClRun, the dataports are color-coded to indicate the type of the data. For example, the blue datapipes are for Matrices, and the yellow datapipes are for Fields. Fields are used to represent 3D geometry, as well as the data values that are defined over that geometry. Taken as a whole, the collection of modules and datapipes in a dataflow apphcation is called a network, or net Each module can have an optional user-interface (Ul) button on its module if the user presses the U1 button, a separate window appears, with controls for viewing and modifying the state of the module s parameters. 

While this interface shortcoming has long been identified, it has only recently been addressed. With the 1.20 release of BioPSE/SCIRun (in October 2003), we introduced PowerApps. A PowerApp is a customized interface built atop a dataflow application network. The dataflow network controls the execution and synchronization of the modules that comprise the application, but the generic UI windows are replaced with entries that are placed in the context of a single application-specific interface window. 

FIGURE 23.1 BioPSE dataflow network for modeling, simulating, and visualizing the bioelectric field generated in a realistic head model, due to a single dipole source. 

Dataflow organization the program is entered as a set of boxes coimected by wires, flowing from the input data on the left to the outputs orders on the right. The wires convey data aeeording to data types the boxes transform data by means of Boolean operators, numerie operators or by means of functions. All loops in the layout are precisely controlled using the previous operator. 

As stated above, the first step in high-level synthesis is the c( iq lation of the input specification to a dataflow type internal representation. This step has many similarities with dataflow CQnq)uting, for which, an AG formalism has been given... 

For the hardware compiler project, a tod has been developed that translates the internal reiaesentation into ddaflow st)le VHIX. that can be used as a preprocessor in modem CAD simdation or synthesis enviroiunents. Dataflow stjie has been chosen because it is a straightforward translation d the scheduled CDFG. 

Figure 5 Dataflow style VHDL description of differential equation solver...

Farrow, R (1983) Attribute Grammars and Dataflow Languages. ACMSIGPLAN Symposium on Programming Language Issues in Software Systems, pp 28-40. 

Naini, M. (1989) A Dedicated Dataflow Architecture for l dware Conqnlation. 22° Aimual Hawaii International Conference on System Sciences. 

Kim, H., Choi, T.-H., Jung, S.-C., Kim, H.-C., Lee, O. Doh, K.-G. 2008. Applying Dataflow Analysis to Detecting Software Vulnerability. In Proceedings ofICACT- the 10th International Conference on Advanced Communication Technology 255-258, IEEE. 

---