An Overview of Data Communication in LabVIEW Elijah Kerry – LabVIEW Product Manager Certified LabVIEW Architect (CLA) - презентация

1 An Overview of Data Communication in LabVIEW Elijah Kerry – LabVIEW Product Manager Certified LabVIEW Architect (CLA)

2 2 Data Communication Options in LabVIEW 1.TCP and UDP 2.Network Streams 3.Shared Variables 4.DMAs 5.Web Services 6.Peer-to-Peer Streaming 7.Queues 8.Dynamic Events 9.Functional Global Variables 10.RT FIFOs 11.Datasocket 12.Local Variables 13.Programmatic Front Panel Interface 14.Target-scoped FIFOs 15.Notifier 16.Simple TCP/IP Messaging (STM) 17.AMC 18.HTTP 19.FTP 20.Global variables … just to name a few …

3 3 WindowsReal-TimeFPGA Communication is Important

4 4 Agenda Introduction of Data Communication Define Communication Types Identify Scope of Communication Inter-process Inter-target Next Steps ni.com/largeapps

5 5 Demonstration The pitfalls of local variables

6 6 Common Pitfalls of Data Communication Race conditions - two requests made to the same shared resource Deadlock - two or more depended processes are waiting for each other to release the same resource Data loss - gaps or discontinuities when transferring data Performance degradation - poor processing speed due to dependencies on shared resources Buffer overflows - writing to a buffer faster than it is read from the buffer Stale data - reading the same data point more than once

7 7 The Dining Philosophers

8 8 Communication Types Message/Command Get me a soda! Update/Monitor The current time is… Stream/Buffer …the day the music died… Variable/Tag Set Point = 72F

9 9 Message/Command Commander (Host) and Worker (Target) Systems Must be lossless* (can be buffered) Minimal latency Typically processed one at a time Reads are destructive Example: stop button, alarm, error *some commands may need to pre-empt other commands based on priority

10 10 Update/Monitor Periodic transfer of latest value Often used for HMIs or GUIs N Targets: 1 Host Can be lossy Non-buffered Example: monitoring current engine temperature

11 11 Stream/Buffer Continuous transfer, but not deterministic High throughput No data loss, buffered 1 Target: 1 Host; Unidirectional Example: High speed acquisition on target, sent to host PC for data logging

12 12 Variable/Tag Set Points and PID Constants Initial configuration data Can be updated during run-time Only latest value is of interest 1 Host: N Targets Example: reading/writing the set-point of a thermostat,.ini configuration files

13 13 Choosing Transfer Types MessageUpdateStreamVariable (Tag) Examples Exec Action Error Heartbeat Movie Waveform Image Setpoint Fundamental Features Buffering Blocking (Timeout) Single-Read Nonhistorical Blocking (Timeout) Buffering Blocking (Timeout) Nonhistorical Optional Features Ack Broadcast Multi-layer Buffering Dynamic Lookup Group Mgmt Latching Performance Low-Latency High- Throughput Low-Latency High-Count Configuration N Targets: 1 Host 1 Target:1 Host Unidirectional N Targets: 1 Host

14 14 Scope of Communication Inter-process: the exchange of data takes place within a single application context Inter-target : communication between multiple physical targets, often over a network layer

15 15 Defining Inter-process Communication Communication on same PC or Target Communicate between parallel processes or loops Offload data logging or processing to another CPU/Core/Thread within same VI/executable Loops can vary in processing priority Used to communicate synchronously and asynchronously ACQ LOG High LowMed

16 16 Inter-process Communication Options Shared Variables Update GUI loop with latest value Queues Stream continuous data between loops on a non-deterministic target Dynamic Events Register Dynamic Events to execute sections of code Functional Global Variables (FGV) Use a non-reentrant subVI to protect critical data RT FIFOs Stream continuous data between time critical loops on a single RT target

17 17 Basic Actions Set the value of the shift register INITIALIZE

18 18 Basic Actions Get the value currently stored in the shift register GET

19 19 Action Engine Perform an operation upon stored value and save result You can also output the new value ACTION

20 20 How It Works 1.Functional Global Variable is a Non-Reentrant SubVI 2.Actions can be performed upon data 3.Enumerator selects action 4.Stores result in uninitialized shift register 5.Loop only executes once

21 21 Demonstration Introduction to Functional Global Variables

22 22 Benefits: Comparison Global and Local Variables Can cause race conditions Create copies of data in memory Cannot perform actions on data Cannot handle error wires Drag and drop Functional Global Variables Prevent race conditions No copies of data Can behave like action engines Can handle error wires Take time to make

23 23 Understanding Data Dependency Code in a VI is organized into diagrams Diagrams are executed in order based on data dependency Objects within the diagrams are executed in order based on data dependency Data dependency is dictated by the flow of wires

24 24 Understanding Dataflow in LabVIEW Clump 1 Clump 2 Clump 0

25 25 While Loop AcquireAnalyzeLogPresent 10ms50ms250ms20ms Doing Everything in One Loop Can Cause Problems One cycle takes at least 330 ms If the acquisition is reading from a buffer, it may fill up User interface can only be updated every 330 ms

26 26 While Loop AcquireAnalyze Log Present 10ms50ms 250ms 20ms Doing Everything in One Loop Can Cause Problems One cycle still takes at least 310 ms If the acquisition is reading from a buffer, it may fill up User interface can only be updated every 310 ms

27 27 Inter-Process Communication: ensures tasks run asynchronously and efficiently How? Loops are running independently User interface can be updated every 20 ms Acquisition runs every 10ms, helping to not overflow the buffer All while loops run entirely parallel of each other While Loop Acquire Analyze 10ms 50ms While Loop 250ms Log While Loop 20ms Present

28 28 Producer Consumer Thread 1 Thread 2 Thread 3 Best Practices 1.One consumer per queue 2.Keep at least one reference to a named queue available at any time 3.Consumers can be their own producers 4.Do not use variables Considerations 1.How do you stop all loops? 2.What data should the queue send?

29 29 LabVIEW FIFOs Queues RT FIFOs Network Streams DMAs User Events In general, FIFOs are good if you need lossless communication that preserves historical information

30 30 Adding Elements to the Queue Dequeueing Elements Reference to existing queue in memory Select the data type the queue will hold Dequeue will wait for data or time-out (defaults to -1) Queues

31 31 Demonstration Introduction to LabVIEW Queues

32 32 The Anatomy of Dynamic Events Dynamic Events Terminal Defines Data Type VI Gets Run on Event Data Sent Multiple Loops Can Register for Same Event

33 33 LabVIEW API for Managing User Events Using User Events Register User Events with Listeners

34 34 Choosing Transfer Types for Inter-process MessageUpdateStreamVariable (Tag) Windows Queue Shared Variable (Blocking, Buffered) SE Queue Notifier Shared Variable (Blocking) Queue Shared Variable (Blocking, Buffered) Local/Global Variable SE Queue FGV Shared Variable DVR RT Same as Windows RT FIFO Same as Windows SE RT FIFO Same as Windows RT FIFO Same as Windows FPGA FIFO (2009) SE FIFO (2009) FIFO Local/Global Variable FGV

35 35 RT FIFOs vs. Queues Queues can handle string, variant, and other variable size data types, while RT FIFOs can not RT FIFOs are pre-determined in size, queues can grow as elements are added to them Queues use blocking calls when reading/writing to a shared resource, RT FIFOs do not RT FIFOs do not handle errors, but can produce and propagate them Key Takeaway : RT FIFOs are more deterministic for the above reasons

36 36 What is Determinism? Determinism: An application (or critical piece of an application) that runs on a hard real-time operating system is referred to as deterministic if its timing can be guaranteed within a certain margin of error.

37 37 LabVIEW Real-Time Hardware Targets Single-Board RIOPXIDesktop or Industrial PC LabVIEW Real-Time CompactRIOVision Systems

38 38 Write Data to the RT FIFO Read Data from the RT FIFO Reference to existing RT FIFO in memory Select the data type the RT FIFO will hold Read/Write wait for data or time-out (defaults to 0) Write can overwrite data on a timeout condition RT FIFOs

39 39 Demonstration Inter-process Communication Using RT FIFOs

40 40 Defining Inter-target Communication PC, RT, FPGA, Mobile Device Offload data logging and data processing to another target Multi-target/device application Network based

41 Common Network Transfer Policies Latest Value or Network Publishing Making the current value of a data item available on the network to one or many clients Examples – I/O variables publishing to an HMI for monitoring – Logging temperature values on a remote PC Values persist until over written by a new value Lossy – client only cares about the latest value 1-11-N

42 Latest Value Communication APITypePerformance Ease of Use Supported Configurations 3 rd Party APIs? Shared Variable* LabVIEW Feature1:1, 1:N, N:1 Measurement Studio CVI CCC (CVT) Ref. Arch. Publishes the CVT 1:1Yes (TCP/IP) UDP LabVIEW Prim.1:1, 1:N, N:1Yes *Network buffering should be disabled

43 43 Using Shared Variables Effectively Programming Best Practices: Initialize shared variables Serialize shared variable execution Avoid reading stale shared variable data

44 Common Network Transfer Policies Streaming Sending a lossless stream of information Examples – Offloading waveform data from cRIO to remote PC for intensive processing – Sending waveform data over the network for remote storage Values dont persist (reads are destructive) Lossless – client must receive all of the data High-throughput required (latency not important) 1-1

45 Streaming Lossless Data APITypePerformance Ease of Use Supported Configurations 3 rd Party APIs? Network Streams NEW! LabVIEW Feature 1:1Not this year STM Ref. Arch.1:1Yes (TCP/IP) What about the shared variable with buffering enabled? NO!

46 Pitfalls of Streaming with Variables Machine 1 Machine 2 Server Client Writers Client Readers Lack of flow control can result in data loss Data may be lost if the TCP/IP connection is dropped Data loss does not result in an error, only a warning

47 Network Streams NEW! Machine A Machine B

48 Network Streams in Action Machine 1 Machine 2 Writer Endpoint Reader Endpoint Acknowledge (3) Acknowledge (2) Flow Control (2) Use Streams!

49 49 Demonstration Inter-target Communication Using Network Streams

50 Common Network Transfer Policies Command or Message Requesting an action from a worker Examples – Requesting an autonomous vehicle to move to a given position – Telling a process controller to begin its recipe Values dont persist (reads are destructive) Lossless – client must receive every command Low latency – deliver the command as fast as possible NN-1

51 Network Command Mechanisms APITypePerformance Ease of Use Supported Configurations 3 rd Party APIs? Network Streams LabVIEW Feature1:1No Shared Variable LabVIEW Feature1:1, 1:N, N:1 Measurement Studio CVI AMC Ref. Arch.1:NYes (UDP) Web Services Web Standard (New VIs in 2010) 1:1, 1:N, N:1Yes

52 52 Writing Elements to the Stream Reading Elements from Stream Reference to reader URL Select the data type the queue will hold Read will wait for data or time-out (defaults to -1) Network Streams

53 53 Network Streams Lossless transfer, even in connection loss* Can be tuned for high-throughput (streaming) or low-latency (messaging) Unidirectional, P2P, LabVIEW only Not deterministic Acquire/Control Log Data/Process

54 Sending Commands with Streams

55 55 Using Shared Variables Effectively Limit Shared Variable Usage Good for small applications with low channel counts Single-process (local) and Networked Combine identically-typed channels into an array Utilize RT FIFO functions Scalable architecture for large applications (local-only) Polling and blocking modes (high performance or low CPU, respectively)

56 56 Using Shared Variables Effectively Avoid unnecessary buffering: Does not guarantee lossless data transfer Consumes more CPU and memory On by default with Network-Published Shared Variables Select an appropriate host for shared variables: RT hosting simplifies multiple host access and increases stability RT hosting also increases CPU and memory overhead Use to retrieve/monitor latest value Use in cases where latency is not a priority

57 57 When to Use Network Streams No data can be lost from the acquisition during transfer between RT target and host computer Transfer data from RT target to host computer for logging data to file Transfer data from RT target to host computer for data processing and analysis that requires more memory than the RT target has available

58 58 DMA (Direct Memory Access) Use for Host to Target Communication (ie: RT to FPGA) Available for newer FPGAs Useful for transferring chunks of data High latency

59 59 Demonstration Introduction to Direct Memory Access

60 60 Data element Target to Host Transfer – Continuous FPGA FIFO RT Buffer DMA Engine Total Samples to Read = ??? Read Size = 4 RT Buffer Size = ~5x Read Size

61 61 Continuous Transfer - Buffer Overflow Total Samples to Read = ??? Read Size = 4 RT Buffer Size = ~5x Read Size FPGA FIFORT Data Buffer Data element DMA Engine

62 62 LabVIEW Application LabVIEW Web Service Client Response Request Sending Requests via URL: Physical Location of Server Name of Web Service Mapping to a VI Terminal Inputs (Optional) LabVIEW Web Services Application Architecture:

63 63 Web Server Web Services in LabVIEW Any Client Thin Client Windows and Real-Time Custom web clients No runtime engine needed Standard http protocol Firewall friendly

64 64 Demonstration Basic Web Services

65 65 ni.com/uibuilder

66 66 Demonstration Thin-Client Web Interfaces

67 67 Early Access Release Details Anyone can evaluate for free Fully functional except for Build and Deploy License for Build and Deploy is $1,499 per user License is sold as one-year software lease Not part of Developer Suite or Partner Lease

68 68 Inter-Target Communication Options TCP/IP and UDP Define low-level communication protocols to optimize throughput and latency Shared Variables Access latest value for a network published variable Network Streams Point to Point streaming in LabVIEW with high throughput and minimal coding Web UI Builder Create a thin client to communicate with a LabVIEW Web Service DMAs Direct memory access between to different components of a system

69 Certified LabVIEW Developer Exam Certified LabVIEW Architect Exam Certified LabVIEW Associate Developer Exam DeveloperSenior Developer Software Architect / Project Manager NI Certifications Align with Training "Certification is an absolute must for anyone serious about calling himself a LabVIEW expert... At our organization, we require that every LabVIEW developer be on a professional path to become a Certified LabVIEW Architect." - President, JKI Software, Inc. Managing Software Engineering in LabVIEW LabVIEW OOP Advanced Architecture LabVIEW Core 3 LabVIEW Core 2 LabVIEW Core 1 LabVIEW Core 1

70 Download Examples and Slides