Random? FIFO overflows

[treeman]

Hello,

I am using a ring buffer to grab images from a Basler404kc.  This works fine most of the time, but sometimes I get a Fifo overflow error.  The error can come after grabbing  a few frames or after grabbing 10's of thousands. I desired 832 x 480 but have found that 1024 x 600 is more stable (both at 180fps). I know that I can grab at 2336x1024 at 300fps without the Fifo appearing so I am well within the usable bandwidth at the lower resolutions. Maybe I just haven't run long enough at the higer rates to get the FIFO error yet.

Previous postings and my own test lead me to believe that it is not an issue with my code holding up the data transfer. In that case I would be losing frames but that should not affect the FIFO. 

Also based on previous postings I am thinking it may be a memory problem, but I have had the FIFO overflow even with nothing else running and over 2GB of memory to spare.

My question is where should I be looking to find the problem? and does it make sense that the 480 height would cause more FIFO errors somehow? also do some frame rates perform better than others (ie certain mutliples)?

Thanks

[s_a_s]

Hello,

I'll see what I can find out to see if we can get to the source of this problem.  There have been reports of overflows not being handled properly, however we'll need a bit more information from you.

1. What version of NI-IMAQ are you using?
2. What version of LabVIEW (or other development environment such as VC or .NET)?
3. Are you using a x1 or a x4 PCIe slot?
4. What camera file are you using?

Thanks

[treeman]

Great,

I have...

1.  NI-IMAQ ver 3.5.1

2. not using LabVIEW, using VisualStudio 2005 C/C++

3. using NI PCIe 1429 with a 4x Lane Width, installed in "One PCI Express x8 card slot wired as x4" -
 Dell Precision Workstation 470

4. Basler 404k(1429) .icd  (version 5.50?)

I found the file "clsernat.dll" in three different locations and with two different versions

 c:\i386 - ver 3.1.0.49153 created 1/22/06

 c:\WINDOWS\system32 - ver 3.5.1.49152 created 11/29/05

 c:\Program Files\cameralink\serial - ver 3.5.1.49152 11/29/05

Is that a problem?

Thanks

[s_a_s]

Thanks for the extra information.  I'll see what we can figure out here.  As for the clsernat.dll having multiple copies, I don't think it would be the source of this problem.  That dll is used for serial communication over CameraLink so the error from the dll would deal more with a serial communication problem than a FIFO overflow.  One more request . . . the next time you get that error would you capture a print screen and post a copy of it here.  I want to be sure I'm looking for the right thing.  Thanks.

-Arves

[treeman]

I'm a bit confused about what a screen capture would tell.  The error I am getting is from imgSessionExamineBuffer(...) which returns the error, according to imgShowError(...),  "FIFO overflow caused acquisition to halt." .  I am using imgRingSetup(...) to start the grabbing.

A screen capture would simply show the last image that I had succesfully displayed.

[Steffen Hoog]

I posted to a similar or equal question in the past. May be it helps you, too.

http://forums.ni.com/ni/board/message?board.id=200&message.id=9215#M9215

Could be necessary to replace the bit-stream name of the attached file in the icd-file of you 404k

[s_a_s]

treeman,

I was assuming that you had gotten a pop-up error message.  The message text however is sufficient.  Let me know if Steffen's recommendation helps.

[treeman]

The other files do not work with my camera. The 504 seems to be a 10 tap camera where the 404 is an 8 tap.

I have been filming for the last 4 days.  On day one I used  1024 x 512 at 180fps. I lost about 20 shots due to the Fifo shutting down acquisition. This means I had to restart the camera over 20 times. The next day, after an hour or so and at least 5 shut downs I changed the resolution to 1280 x 544 at 180fps. From then on the Fifo only appeared 3-5 times a day. Nothing but the resolution changed.  Again the Fifo errors seemed random. Sometimes occuring after a few frames, sometimes after 100's of thousands of frames.

I only have one chance to film my subjects so I must have a dependable system with no shutdowns.  I could redo the code so that the overflows are detected and the grabber restarted, but that would still cause me to miss some subjects.

Any ideas?   

[s_a_s]

The fact that increasing the image size decreases the frequency of the FIFO error is counter intuitive and has me baffled.  This is especially true given that you are acquiring at the same frame rate in every case.  I would think that increasing the frame size would increase the frequency of FIFO errors.  Would you mind if I contacted you off-line to talk about this in greater detail?  We might need to ask you if it would be possible to send in some of your hardware so we can test it.

Thanks,

[treeman]

That'd be great, how do I get you my phone or email without giving it to the world?