My design consumes a lot more block RAM than I expected.
And I can not read the overhead from the OpenCL code. Is there any other way to get the RAM block overhead details so that I can which part consumes the block RAM?
By the way, I am using OpenCl 16.01
Consult the area report which includes line-by-line resource usage. I don't remember exactly but I think the area report for v16.0.x had to be generated manually. If you use newer versions of the compiler (16.1+), you will get a more detailed HTML report.