Two DMA FIFO fill and asynchronous playback?

Similar Questions

• Using TWO DMA FIFO target host

  Hi, I use a unit 9004 cRIO for the purpose of data acquisition. I try to get signals analog (NI 9221) and CAN (NI 9853) modules and transfer them from the fpga VI VI in real time using DMA FIFO, that is to say, target to the host. I thought I should use 2 DMA FIFO, one for analog and CAN signals. When I tried to put them, I put the two FIFO in the same loop. I don't know if that's the only reason, but it would generate overflow error 50400. When I removed a FIFO and put the single FIFO, it worked fine. So, I thought they must perform independent while loops and implemented two while loops that would be run in parallel. And it worked. I don't know if this is how it is supposed to work? Is this that FIFO DMA should be run in parallel loops? Are they written in such a way that the FIFO function cannot be called for more of an instance in the same loop? This seems odd.

  I should also mention that I reconnected it the wires on analog, digital and CAN modules before putting them in separate loops. And I know for a fact that the FIFO CAN show underflow error if the entry of the threads on the port CAN are not connected. Also, I don't use mode Scan Engine (I think that its not available in 9004)

  You are right about the 9004 and Scan Engine - or the lack of support, I say...  Regarding the error of underflow, this behavior is expected when you have two Scriptures DMA FIFO the same loop.  What is happeing, it is that the two FIFOs write in true parallel.  Thus, when you write to FIFO1, FIFO2 expected FIFO1 finish. Then, when you write to FIFO2, FIFO 1 must wait on FIFO2 and the while loop to reiterate before he can write more data.  So, on your VI in real time, if you try to access one of these FIFO that has no data in there because it is waiting to write the next block of data, then an overflow error will happen.

  When you separate the two FIFOs, both can run independently from each other and write at a much faster pace.  In this way, your FIFO reads on the realtime VI will not be stuck in a State of negative overflow where a FIFO is waiting for data queued.

  I hope this helps!

• Read and write in two separate FIFOs DMA on RT host

  Hello

  I have two parallel loops running on a host of RT on a CRio9022. There are two DMA FIFO: one of the FIFO is written for in the upper loop, another FIFO is read from the bottom loop. I've attached a screenshot of my code.

  The problem I encounter is that I seem to be only able to run one of these loops - for example if I disable the first loop, I can see data through the second loop. Trying to run them in parallel because I think I've coded it means only one of the tracks-it loops is always the top loop in the code that I have attached the screenshot. Each of the FPGA VIs that run only use a FIFO so I think they should be able to run independently... I was wondering if anyone could shed some light on this?

  Thank you

  Hello.

  You can only have a single call to the FPGA in HOST mode, the algorithm that you post, that is make two calls to the VI 'Reference of VI FPGA open', this is not allowed, this is why the program works only with one of these cycles.

  Kind regards.

• Reading of the zeros of a DMA FIFO empty

  Hello

  I'm having a problem using the DMA FIFO to communicate between my real-time system and my FPGA. I use two DMA FIFO, one-way to the FPGA of the RT system and then vice versa. I can successfully get data to and from each system; However, before, after and sometimes inbetween my data, I'll get a seemingly random assortment of zeros. In the latest version of my code (which I have provided) I read the number of items stored in the DMA FIFO and read only this number to my RT system to try to get the data that I want and no zeros, but this does not work either.

  I'm new to both in real-time and FPGA and so it certainly feels like I'm missing something very basic. I tried dealing with this problem by myself well and have had absolutely no success and would appreciate any help.

  My equipment includes:
  A Dell laptop (used as a host of the user interface)
  1082 chassis
  Controller of 8133 (running the LabView RT operating system)
  7965R FPGA
  5781 module for FPGA (not currently in use)

  LabView 2014 SP1

  On the side of RT, you need to use a structure of the case so that you have not read of the FIFO if no data is available.

  I would also like to change the flow of network to be inside of your time in a loop until the user interface capturing each data point, as it comes (connect before the loop, close after the loop and write it as the data come from the inside of the loop).  This will save memory (which is VERY important in a RT system) since you don't have to set up the table.

  You write only as an element in the FPGA.  So any sous-suite readings should give you an array of 0 s, causing 0s back upward.  You must maintain the Timed Out in the FPGA so that the data will only until to the RT when there is real data to send.  Your FPGA code could be reduced to this:

  

• How to choose the maximum number of items for DMA FIFO to the R series FPGA

  Greetings!

  I'm working on a project with card PCIe-7842R-R series FPGA of NOR. I use to achieve the fast data transfer target-to-host DMA FIFO. And to minimize overhead costs, I would make the size of the FIFO as large as possible. According to the manual, 7842R a 1728 KB (216KO) integrated block of RAM, 108 000 I16 FIFOs items available in theory (1 728 000 / 16). However the FPGA had compilation error when I asked this amount of items. I checked the manual and searched online but could not find the reason. Can someone please explain? And in general, what is the maximum size of the FIFO given the size of the block of RAM?

  Thank you!

  Hey iron_curtain,

  You are right that the movement of large blocks of data can lead to a more efficient use of the bus, but it certainly isn't the most important factor here. Assuming of course that the FIFO on the FPGA is large enough to avoid overflowing, I expect the dominant factor to the size of reading on the host. In general, larger and reads as follows on the host drive to improve throughput, up to the speed of the bus. This is because as FIFO. Read is a relatively expensive operation software, so it is advantageous to fewer calls for the same amount of data.

  Note that your call to the FIFO. Read the largest host buffer should be. Depending on your application, you may be several times larger than the size of reading. You can set the size of the buffer with the FIFO. Configure the node.

  http://zone.NI.com/reference/en-XX/help/371599H-01/lvfpgaconcepts/fpga_dma_how_it_works/ explains the different buffers involved. It is important to note that the DMA engine moves data asynchronously read/write on the host nodes and FPGAs.

  Let me know if you have any questions about all of this.

  Sebastian

• How can I send more than one signal to DMA FIFO?

  Hello

  I am trying to send more than one signal to DMA FIFO, but I don't know how to do. I have no problems when I send a signal. I try to use a DMA FIFO block for a single signal. For example if I have 3 signal I use · FIFO of DMA but when I want to wath them a table waveform signals have a delay.

  How can I do to send more than one signal to DMA FIFO? and if this is not possible, how do I do to syncronizate 3 signals?

  The data type of the signal is FXP<16,10>

  Kind regards.

  Pablo

  
  

• simple DMA FIFO reading two analog channels

  Hello

  I have a question on a method of data transfer between two analog inputs for a simple DMA FIFO in FPGA. The code is described here: http://decibel.ni.com/content/docs/DOC-6303. If I use this method, and I got out in a graph of my host VI, the calendar in the graph reflects the same schedule as the signals that have been entered? Or will they be phase shift between two signals?

  Thank you

  Grant

  Grant:

  Because it is not all information of timing with the signals in the FIFO, there will be no lag phase on the chart.

  Hope that helps. I would like to know if I forgot something, or who does not explain very well.

  Thank you!

• Number of DMA FIFO of items to read mismatch in the FPGA and RT

  Hi all

  I use myRIO, LV14 to run my application.

  Request: I have to continuously acquire data via FPGA and host RT process once every 2000 samples are taken. I use DMA FIFO (size 8191) to acquire data, use timeout property in the FPGA to eliminate the buffer overflow. I had followed cRIOdevguide to implement this part. An excerpt of what I put in place is attached. All code runs in the SCTL at 50 MHz.

  Question: Two or three times I met with this strange behavior, the FPGA FIFO gives continous timeout and the RT is unable to read the FIFO. The number of elements to set the property in the FPGA VI gives 0 showing that FIFO is full and no more can be written, but the RT, remaining items gives 0, so it is reading 0 (none) elements.

  Solution: I put a case where I'll write to FIFO (under the code) and if the number of elements to write is different from zero. It seems to work fine, from now.

  What confuses me, is that my FPGA VI said that FIFO is full (number of items to write 0 = FIFO) and gives a timeout error, but RT VI said that number of items remaining in the FIFO is 0 and therefore no data is read. No idea why this is so? My RT and FPGA VIs continues to run, but with no gains or to read data.

  A few minutes after you run the code, I've seen this behavior. No idea why this happens? I try to reproduce the behavior, and will update if I meet with her again. Sorry, I can't post my code here, but I guess the code snippets to explain some extend.

  Thank you

  Arya

  Edit: Even with the mentioned workaround solution, the problem persists, now that the FPGA written any of FIFO. And the RT VI is not able to read all the elements he sees 0 items in the FIFO. The FIFO continues to be in a State of timeout. So I guess that the problem is on the side of RT.

  Why it looks like you read from the FIFO even in two different places in the same VI, at the same time? If the lower reading throws the FIFO, it will never trigger the reset, which could lead to the situation you describe, I think (it's hard to tell from a few screenshots).

  Also, your logic seems too complicated. I immediately noticed that there is no reason to select the entry, the output of = 0 - simply use the 'equal to zero' output directly. On the side of FPGA, why you need check the number of items that you want to write? There's nothing wrong with writing in a FIFO that is already full. just the data won't get written.

• DMA FIFO and node VHDL

  Hi guys!

  I am writing here after many days of attempts without success...

  My request is 'simple', send data from the Panel of Labview RT within a DMA FIFO target host. Then the FPGA core receives data and imported through a knot of VHDL processes, and then after the data is pointing to the Labview RT through a target to host DMA FIFO.

  I tested my node VHDL simulation mode in Labview FPGA where data sent by a target scope THAT FIFO and just work fine.

  But when I try to run the node VHDL in the real target with data from the RT by DMA FIFO basis, it won't. I already do some checks:

  -Data are properly sent through the host target DMA FIFO;

  -The data are correctly received in the FPGA base;

  -The data are correctly sent to the node VHDL;

  -Result of the node VHDL are correctly sent to the heart of the RT through the target to host DMA FIFO;

  -Result are correcly received in the heart of the RT, , but the result is false and absurd. But I have proof that my node is semanticly correct with my mock test

  So my question: are there reasons to see my work VHDL nice knot in simulation mode and not in mode real target with data from the base RT by DMA FIFO taking into account data Transfer between DMA FIFO work well in both sides? Is there some sample available with data send RT FPGA-based via DMA and data processing with a knot of VHDL and returned to the RT kernel to inspire me?

  I can't post my screw here because I work for a company, I use a MyRIO with Labview FPGA 2014 target.

  Thank you guys! I am available for some details on my implementations.

  Afghow.

  Hey!

  Thank you for your answer but I solved my problem. Indeed, at first, I tried to make a knot of Combinatorics (without clock) pure, but the problem seemed to come from that.

  I modified my node in order to incorporate a clock, according to the prescriptions of this white paper: http://zone.ni.com/reference/en-XX/help/371599K-01/lvfpgaconcepts/ipin_prepare_ip/ . And now, every thing seems to work well.

  The question remains why the combinatorial node has worked in simulation mode en not in the actual target?...

  But for people with the same problem, I suggest add them a CLK and check an edge of entry with rising_edge (CLK) and if it does not, add an input signal to check if the input signals are valid or not.

  Afghow.

• Reading and arraya thro DMA FIFO (PCI7813R)

  Normal 0 21 false false false SL X-NONE X-NONE MicrosoftInternetExplorer4 / * Style Definitions * / table. MsoNormalTable {mso-style-name : « Navadna tabela » ; mso-tstyle-rowband-taille : 0 ; mso-tstyle-colband-taille : 0 ; mso-style-noshow:yes ; mso-style-priorité : 99 ; mso-style-qformat:yes ; mso-style-parent : » « ;" mso-rembourrage-alt : 0 cm 5.4pt cm 0 5.4pt ; mso-para-marge-haut : 0 cm ; mso-para-marge-droit : 0 cm ; mso-para-marge-bas : 10.0pt ; mso-para-marge-gauche : 0 cm ; ligne-hauteur : 115 % ; mso-pagination : widow-orphelin ; police-taille : 11.0pt ; famille de police : « Calibri », « sans-serif » ; mso-ascii-font-family : Calibri ; mso-ascii-theme-font : minor-latin ; mso-hansi-font-family : Calibri ; mso-hansi-theme-font : minor-latin ;}

  Hello

  Location:

  I read 5-position encoder with PCI7813R. The position of encoder data format is I32. Following is each position encoder converted to U32 and build in a table by using "build the array function. Finally this table guided loop where I write it in DMA FIFO uses "FIFO write function. FIFO depth is equal to 16383 and Timeout to 0.

  Side host I'm reading this table of DMA FIFO. The host FIFO depth is set to 1000000. I use 'Method Invoke with FIFO read' to read data from the FIFO. I read 10000 elements each time.  Data output is converted to I32 and guided an extension to "decimate the 1 d array function. I use 5 indicators to represent the positions of Encoder on the host side.

  And now the problem:

  Position 1 Encoder shell displayed on indicator 1. But every time I close my application and run it again this position appears on the other indicator. Sometimes is the position of the Encoder 1 shown on the 1, sometimes on the indicator 5 indicator, sometimes on indicator 3... Although this position of the encoder is shown every time on the other indicator it has correct values. What I'll do as I will always receive encoder position 1 on indicator 1, position of the Encoder 2 Indicator 2...

  My application has been built according to these instructions:

  http://zone.NI.com/DevZone/CDA/tut/p/ID/4534

  TNX,

  Gregor

  Normal 0 21 false false false SL X-NONE X-NONE / * Style Definitions * / table. MsoNormalTable {mso-style-name : « Navadna tabela » ; mso-tstyle-rowband-taille : 0 ; mso-tstyle-colband-taille : 0 ; mso-style-noshow:yes ; mso-style-priorité : 99 ; mso-style-qformat:yes ; mso-style-parent : » « ;" mso-rembourrage-alt : 0 cm 5.4pt cm 0 5.4pt ; mso-para-marge-haut : 0 cm ; mso-para-marge-droit : 0 cm ; mso-para-marge-bas : 10.0pt ; mso-para-marge-gauche : 0 cm ; ligne-hauteur : 115 % ; mso-pagination : widow-orphelin ; police-taille : 11.0pt ; famille de police : « Calibri », « sans-serif » ; mso-ascii-font-family : Calibri ; mso-ascii-theme-font : minor-latin ; mso-hansi-font-family : Calibri ; mso-hansi-theme-font : minor-latin ;}

  Sorry Jochen, I placed the function reset at the end of my routine. L I corrected (put it on the beginning of the routine) and now it works as it is supposed to be.

  Thanks a lot again!

  Kind regards

  Gregor

• SMU FlexRIO DMA FIFO host read the FIFO overflows broadband bandwidth/DMA issues

  I'm working on an application that uses 2 modules FlexRIO, and 2 LVDS digital I/O adapters.  I'm driving each of the SDC A/SDC B ports on LVDS 16-bit data at 50 MHz adapters.  The FlexRIOs are expected to receive the data and write down them on four targets-to-host DMA FIFO (one per connector SDC), or two by FlexRIO.  The host reads the FIFO and brings together a series of tables each FIFO output 2D.  Ultimately, the individual tables (we're each a quarter of single image) will be assembled in simple images, but I haven't gotten that far yet.

  The duty cycle for the data is about 80% (in other words, I'm only transmit data to the FlexRIOs 80% of the time, the rest of the time the transmitters are disabled), so the flow is about 80 Mbytes/sec/port total invasion, or 320 MB/s on the four FIFOs DMA.  I find that the acquired data gaps sometimes inside that line up along the length of the material part of the DMA FIFO in FlexRIO modules.  In other words, if my memory FIFO DMA are set to 65535 length, I'll see a break in the data acquired at the word of data 65536th.  Data is a waveform of sight, which is essentially just a counter, so it's easy to see the break in the model.  For the words of first 65535, adjoins the data, then from Word 65536 model is discontinuous and starts counting again from there, contiguously.  At the beginning of the acquisition, the FIFO is erased: the beginning data read from the FIFO is always aligned correctly, so I know that the process starts at a good point.

  The error is not always the case: sometimes I get continuous data through the point 65536.  In addition, the error occurs independently between the four FIFO: on a particular race, a FIFO could have data of interest and some bad.  Rarely, all four FIFOs have good data.

  The fact that the gap of the configuration is to the point even the depth of the FIFO DMA tells me that fills the FlexRIO FIFO, the FPGA hardware without the system managing to move to read, which means that the data gets dropped during the period that the FIFO is full.  Then transfer to the host comes into action, there again is the space in the FIFO, and the data is once more contiguous in FIFO memory for a large amount of data (I have not yet tried to locate a second gap in the data of a single acquisition).  It seems therefore that the host doesn't have enough bandwidth between the FlexRIOs and the host of RAM to prevent the filling FIFO, or comes along some software process on the host that is temporarily stop the ability to instantly transfer.

  Are at - it a specification for the SMU flow system that would indicate that we are trying to use too much bandwidth?  Or are there priority controls on DMA FIFO that would allow us to raise the priority of the FIFO transfers as they are guaranteed to go in preference to other system tasks?

  System Specs:

  SMU-1075 chassis

  SMU-8135 CPU

  2 SMU-7962R FlexRIO modules

  2 digital i/o modules of NOR-6585

  LabView 2012 32-bit SP1 version 12.0.1

  A suggestion of an applications engineer of NOR and some experimentation has solved the problem. It turns out that I was calling the method FIFO of DMA stop just before the outbreak of the transmission of the data via a control for the FPGA FPGA VI. I did this in order to clear the FIFO before you begin data acquisition, but I didn't know that this method disables also transfer data between the memory FlexRIO and host. Following this call, I trigger the FPGA code to start filling its FIFO and then begin reading. Calling the Read of FIFO of DMA apparently light up the transfer back, but it seems that the host VI has been randomly slow down enough to move to the bed such as filling the side FlexRIO FIFO and dat would be lost. I changed the host VI to insert a FIFO method call start before the trigger for the FPGA signal, and the problem is now gone.

• Transmission of data to the host of RT to the FPGA via DMA FIFO

  Hello

  I try to write data from a host of RT on target FPGA using DMA FIFO and then process these data and read then return of the FPGA target to the host of the CR through an another DMA FIFO. I'm working on the NI SMU chassis 1062 q, with the built-in NI SMU-8130 RT controller and target FPGA NI SMU-7965R.

  The problem I face is that I want to send three different tables, two of the same size and the third with different size, and I need one more small to be sent first to the FPGA. I tried to use encode dish with two executives in the FPGA VI. In the first image, I read and write the first table in a while loop which is finite (that is, a finite number of iterations). The second frame contains the process of reading and writing the second two tables (of the same size) in a while loop that can be finite or infinite (depending on a control). The problem is that it does not work. 2 arrays are displayed on the front panel of the RT VI host and works well, however, the table that should have been read in the first sequence does not appear on the front panel of the RT VI host. It is not sensible because if it is not passed from the host to the fpga and vice versa then the second image should not have been executed. Note that I'm wiring (-1) for the time-out period to block the while loop iterations until the passage of each item is completed. So the first while loop has only 3 iterations. Could someone help me undersdtand why this happens and how to fix this?

  I enclose a picture of the host and the fpga vi.

  Thank you.

  If you vote for my idea here and it is implemented, you can even omit the loop FOR fully.

  (I also propose the RE / IM divided inside the loop FOR and perform operations on complex table before the loop the transpose and reshape .) In this way, you only need one instance of these operations. You might even save some unnecessary allocations table in this way)

• What would be the effect of the adjustment of the depth of DMA FIFO before every read?

  I'm using a PXI-7813R FPGA board using a 3rd party API. I had a few problems during playback. I looked in their API (LabVIEW) code and realized that they were in DMA FIFO depth before each reading of the FIFO. This apparently does not cause a failure of catastrphic issues I have observed are only transitional in nature.

  What kind of problems, if everything can these operations cause or will be ignored because the FIFO is already running?

  Thank you...

  mgerceker,

  According to the help file , that should be OK to do even though I could see it being a problem if it is not set correctly before performing an operation. What kind of symptoms do you observe?

  Greetings from Austin,

• Target to host DMA FIFO not compensation when they are arrested

  I use a PXI-7841R (Virtex5) and 32-bit data to the host via DMA FIFO transfer. When you read the FIFO on the host for the first time, the data are "stale" (which means that it is not what is currently coming in the FPGA, but what came in a few seconds ago stale). I tried both a stop and a configuration for clear memory FIFO before I use them. Documentation on one or both of these so-called clears the target and host the FIFOs. Does not help in both cases. With readings of the second and the following, FIFO has then 'valid' (same data as it appears on the FPGA entry node). What is curious is that each reading exactly the same thing:

  1. Stop the FIFO (must erase all data)
  2. Elements of reading 375 of the FIFO. (repeated playback of the FIFO)

  Also interesting: FPGA FIFO is implemented for 255 elements. The first 255 items host-side contain the data "stale" on the first reading. It reminds me of the never erased FPGA FIFO.

  Answered by support OR. The documentation for the FIFO, stop and configure FIFO is in error. Cars of documentation will be written against them.

  Solution:

  These methods remove only the FIFO on the side host. Data FPGA FIFO must be read following until no element.

• For a DMA FIFO running from the host to the 7976 is a data type that is optimum for PEP?

  For transmission to the host 7976, I can pack my data in say U64s or break up in U8s.  Y at - it a data type that will give the best rate?  Maybe based on the bus (SMU) or the implementation of RIO drivers behind the scenes?

  My experience is different.  Once, I did a load of tests with different widths DMA FIFO for FPGA and tested the throughput and latency.  If sending data via U8, U16 or U32 DMAs, I saw the same total transfer of bps.  My explanation for this?  Given that the width of the DMA is 32 - bit, LV little packaging in order to ensure that each part of the 32 bits is used.  This means that if you have a DMA U8, it will transfer at a time, 4 to DMA 2 both U16 or U32 DMA one at a time.  64-bit is divided into two individual transfers.

  Do not use FXP.  Even a 1-bit FXP is represented internally as 64-bit and will require TWO DMA transfers to an FPGA.

  Side fromt hat, U8, U16 or U32 makes essentially no difference because they are all packed 32-bit internally.