This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

12MP CMOS with DM355

Dear All:

 

Can DM355 support CMOS module up to 12MP?

How many fps can we have?

Is there anyone have done it already?

If DM355 can not support 12MP CMOS,

can DM365 support 12MP CMOS?

Any information will be appreciated. [:)]

 

Best regards,

Ray

  • yes, both DM355 and DM365 can support 12M pixel CMOS sensors.  How many frame per second you can get will depend on the chosen hardware pixel clock.  DM355 can support up to 75 MHz pixel clock, hence 75/12 = 6.25 frames per second...

    VPFE User Guide for each of these two corresponding parts have the information on maximum resolution ad pixel clocks supported for these parts.

  • Dear Juan:

    Thanks for this information.

    I was just wondering if there is any other constraint rather than

    75 MHz pixel clock.

    When we worked on 5MP CMOS, we encountered DMA buffer size

    problem in the 2.6.10 linux kernel. It failed to allocate a single DMA

    buffer big enough to hold the data coming from 5MP CMOS.

    We were afraid that there would

    be something else we need to pay attention to.

     

    regards,

     

    Ray

     

  • Did you enconter this buffer size limitation in the user space or in the kernel space? I have ran into this in the kernel space before but was able to overcome it. 

    With regards to other constraints, please note that each individual block in the video pipeline (e.g. preview, resizer...) have their own set of features supported.  I belive the previewer specially may have limitations on the maximum line size supported and this varies from device to device... the fact that CCD can take in a large frame size into DDR2, does not necessarily mean all the other video pipeline blocks can process those large frames...though once in DDR2, one one break the frame down into smaller pieces and process the smaller pieces...just things to keep in mind.

     

  • Dear Juan:

     

    Yes. The problem we encountered is in kernel space.

    But I found the problem seems to be fixed in 2.6.18.

    It can allocate a single buffer for one 5MP frame.

    I am curious about how you fixed the problem in your case.

    Maybe it will be helpful in our project as well. Currently we

    try to adopt 2.6.31.

     

    We realize that there are constraints in other HW components.

    To process the image data in smaller pieces is also a solution

    we adopted now. IPIPE is OK. But will it cause other side-effect

    in CCDC? Currently we are working on three different CMOS modules

    ranging from 720p to 5MP. 12MP will be used in the near future.

    There seems to be a lot things we need to modify in the kernel

    space. Your suggestion will be very helpful.

     

    Thanks for your help.

     

    Best regards,

     

    Ray

     

  • The changes we made were against MV Pro 4 (kernel 2.6.10), hence they may or may not apply to newer kernel.  The files and location that changed are described in the following App Note

    http://focus.ti.com/general/docs/litabsmultiplefilelist.tsp?literatureNumber=spraan0

    In particular, we had to make one line chages to both mmzone.h and memory.h to overcome buffer limit size.  I do not recall exact changed but a quick diff will likely highlight which lines changed... (source code included with App Note)

  • Dear Juan:

     

    Thanks for this info. I have tested it and it works well for our 5MP cmos.

     

    Best regards,

     

    Ray