Understanding SGBM and parallelizing to get better performance

asked 2017-01-03 00:08:42 -0500


I am trying to understand the basic implementation of SGBM to parallelize the code. I thought that the cost calculation was one that can be separated from the main method and do parallel operations and give the cost input to the aggregation module.

Am I in the right direction ?

Will it be possible to do optimization as the aggregate cost calculation was dependent on different directions and some directions traverse bottom to top and vice verse.

Will there be any other way to do optimization ?

Thanks in advance

edit retag flag offensive close merge delete



I am working on the same problem. Did you find any optimization techniques or understanding of the Cost Aggregation module?


CoderUnknown gravatar imageCoderUnknown ( 2018-10-25 06:38:29 -0500 )edit