You are here: NI Home > Support > Product Reference > Manuals > LabVIEW Real-Time Module 8.5 Help

Single-CPU computers achieve multitasking by running one application for a short amount of time and then running other applications for short amounts of time. As long as the amount of processor time allocated for each application is small enough, computers appear to have multiple applications running simultaneously. Systems with multiple CPUs, on the other hand, can achieve true parallelism.

Multithreading applies the concept of multitasking to a single application by breaking it into smaller tasks that execute for short amounts of time in different execution system threads. A thread is a completely independent flow of execution for an application within the execution system. Multithreaded applications maximize the efficiency of the processor because the processor does not sit idle if there are other threads ready to run. An application that reads and writes from a file, performs I/O, or polls the user interface for activity can benefit from multithreading because it can use the processor to run other tasks during breaks in these activities.

The real-time operating system (RTOS) on NI RT targets uses a combination of round robin and preemptive scheduling to execute threads in the execution system. Round robin scheduling applies to threads of equal priority. Equal shares of processor time are allocated among equal priority threads. For example, each normal priority thread is allotted 10 ms to run. The processor executes all the tasks it can in 10 ms and whatever is incomplete at the end of that period must wait to complete during the next allocation of time. Conversely, preemptive scheduling means that any higher priority thread that needs to execute immediately pauses execution of all lower priority threads and begins to execute. A time-critical priority thread is the highest priority and preempts all priorities.

LabVIEW executes timed structures threads below time-critical priority and above high priority. You can specify the priority level of Timed Loops relative to other timed structures within a VI by setting the priority of the Timed Loop. Use the Configure Timed Loop Dialog Box to configure a timing source, period, priority, and other advanced options for the execution of the Timed Loop.

Dividing Tasks to Create Deterministic Multithreaded Applications

Deterministic applications depend on deterministic tasks to complete on time, every time. Therefore, deterministic tasks need dedicated processor resources to ensure timely completion. Dividing tasks helps to ensure that each task receives the processor resources it needs to execute on time.

Separate deterministic tasks from all other tasks to ensure deterministic tasks receive enough processor resources. For example, if a control application acquires data at regular intervals and stores the data on disk, you must handle the timing and control of the data acquisition deterministically. However, storing the data on disk is inherently a non-deterministic task because file I/O operations have unpredictable response times that depend on the hardware and the availability of the hardware resource. You can use Timed Loops or VIs with different priorities to control the execution and timing of deterministic tasks.

Note  Within deterministic tasks, ensure that each operation receives dedicated processor resources by avoiding unnecessary parallelism. In a multiple CPU system, avoid creating more parallel operations than the number of available CPUs. Because it is impossible to determine the execution order of parallel operations, unnecessary parallelism can impede determinism.

Using the Analysis Workspace to Reduce Jitter in Mathematics and Signal Processing VIs (ETS)

Many of the analysis VIs on the Mathematics and Signal Processing palettes require resource allocations for internal calculations. Because requesting resources from the operating system can introduce jitter, analysis VIs require a preallocated analysis workspace to minimize jitter. Use the Real-Time Analysis Utilities VIs to handle the resource requirements of Mathematics and Signal Processing VIs. Use the Real-Time Analysis Utilities VIs in the following order:

1. Initialize Analysis Workspace
2. Start time-critical VI or high-priority Timed Loop
3. Enable Analysis Workspace
4. Execute time-critical Mathematics and Signal Processing VIs
5. Disable Analysis Workspace
6. End time-critical VI or high-priority Timed Loop
7. Uninitialize Analysis Workspace

Caution  You must place the Enable Analysis Workspace and Disable Analysis Workspace VIs in the same time-critical thread as the Mathematics and Signal Processing VIs that use the analysis workspace. Between the Enable Analysis Workspace and Disable Analysis Workspace VIs, do not include any functions that involve wait or timeout intervals. You can use only one analysis workspace in an RT application, and you cannot use the analysis workspace to execute multiple Mathematics and Signal Processing VIs in parallel, even on systems with multiple CPUs. Failure to follow these guidelines can cause race conditions, priority inversions, and other undefined behaviors.

Note  The Real-Time Analysis Utilities VIs reduce jitter only when used on RT targets running the ETS real-time operating system. Although you can run the Real-Time Analysis Utilities VIs without error on VxWorks targets, doing so does not reduce execution jitter.

The following figure shows a typical usage of the analysis workspace.

The analysis workspace benefits Mathematics and Signal Processing VIs that request memory for internal calculations. Some Mathematics and Signal Processing VIs do not benefit from the analysis workspace. The analysis workspace reduces execution jitter only for the following VIs:

• Autocorrelation
• Bohnman Window
• Chebyshev Window
• Cholesky Factorization
• Convolution
• Deconvolution
• Eigenvalues and Vectors
• Exponential Fit
• Gaussian Peak Fit
• General Cosine Window
• General LS Linear Fit
• Inverse Matrix
• Linear Fit
• Logarithm Fit
• Median Filter
• Modified Bartlett-Hanning Window
• Parzen Window
• Periodic Random Noise
• Power Fit
• PseudoInverse Matrix
• QR Decomposition
• Savitzky-Golay Filter
• Savitzky-Golay Filter Coefficients
• Scaled Time Domain Window
• Solve Linear Equations
• SVD Decomposition
• Symmetric Window
• Welch Window
• Zero Phase Filter

My Profile | Privacy | Legal | Contact NI   © 2009 National Instruments Corporation. All rights reserved.   |