Max buffer size VLIB_getblobIIBufSize

Daniel Garcia Llorente

I would like to avoid calling the function VLIB_getblobIIBufSize to get the buffer size every time I need to rerun the algorithm. Instead, I would like to know the formula behind the function to know the maximum theoretical size of the buffer, so I can use a fix buffer size. If there is any recommendation about its size (e.g., the maximum theoretical buffer size is 10 for X parameters, but the chances of it being bigger than 5 are really small) I also would like to know it.

I'm using vlib_c66x_3_2_1_0.

As a side note, I could not find the description of the returning values of the functions in the VLIB library. Are they documented? If they are not, it would be nice for them to be.

Regards,

Daniel

over 10 years ago

0 Jesse Villarreal over 10 years ago

TI__Expert 5655 points

Daniel,

The formula for the size is:

size = 12 + 8*(height + maxNumIntervals);

These constants are based on the size of internal data structures used to store computation information.

Regarding the return values for functions which do not have it defined in the API, you can reference the VLIB_STATUS enumeration in the ti/vlib/src/common/c66/VLIB_types.h file.

Regards,
Jesse

0 Daniel Garcia Llorente over 10 years ago in reply to Jesse Villarreal

TI__Prodigy 100 points

Jesse,

Is there any estimation of the maximum value for maxNumIntervals that I could calculate beforehand?

Regards,
Daniel

0 Jesse Villarreal over 10 years ago in reply to Daniel Garcia Llorente

TI__Expert 5655 points

Yes, the worst case is if you have one blob filling the entire image in a checkerboard pattern. In this case, the maxNumIntervals = width * height / 2 ... (half of toal pixles in frame). maxNumIntervals should always be <= this number.

Processors

Processors forum

Max buffer size VLIB_getblobIIBufSize