C2H compiler streaming interface

[Altera_Forum]

Hi, 

 Is it possible to write a image preprocessing functions in streaming operations using C2H compiler.   E.g if i want to perform few operations on an image without storing it into memory buffer.  filter --> threshold --> ccl ... and so on..   I am kind of confused about the way i need to write these C functions such that the input output will act as streams and wont access the memory. And thus can be efficiently hardware accelerated using C2H compiler.   can somebody please give a small example .. e.g the prototype of such function and a small sample code.   thanks in advance.  

[Altera_Forum]

It doesn't support streaming but you might be able to use a ST-->MM FIFO which C2H could then access using a master that continuously accesses the FIFO slave port.

[Altera_Forum]

<div class='quotetop'>QUOTE (BadOmen @ Jul 17 2009, 03:30 AM) <{post_snapback}> (index.php?act=findpost&pid=23157)</div> 

--- Quote Start ---  It doesn&#39;t support streaming but you might be able to use a ST-->MM FIFO which C2H could then access using a master that continuously accesses the FIFO slave port.[/b]  --- Quote End ---    Oh Ok. Thank you BadOmen.   I needed a few suggestions regarding the following topic. I have to evaluate the ImpulseC c2h converter and Altera c2h compiler.  And I suppose the major difference between the two is, the ImpulseC uses its own API and functions and datatypes, and the C code needs to be written in a particular structure in order to get it converted to hardware efficiently.  As for the altera c2h compiler, the code structure is comparatively pretty straight forward using standard C variable types.  Apart from it, the ImpulseC converts streaming algorithms with its co_stream data type. while for Altera c2h compiler, to implement streaming algorithms, we need to use the ST--> MM FIFO as you mentioned.   Is there any major difference between the performances of these two c2h converters ?  I understand if the C code is not written as per the guidelines, the hardware generation might not be very efficient but assuming that, i would be able to write a C code following the proper guidelines, will there be major difference between the performances of the two.   thanks in advance. regards.