[PATCH] D24715: [OpenCL] Block captured variables in dynamic parallelism - OpenCL 2.0

[bekket mcclane via cfe-commits]

mshockwave added a comment.

> From this picture I don't see how the flattening itself can help us to avoid using global memory. Surely in both cases the captures content will have to be copied into the memory accessible for the enqueued kernel (which is a global memory in a general case, but doesn't have to be in some cases I am guessing). Perhaps I am missing some extra step in the approach you are proposing. If you rely on the parameter passing using normal function call into the block_invoke then in both cases we can skip the memcpy of captures at all. Otherwise both appoaches will need to make a copy of the captures.

Now I agree with the necessity of global `accessable_memory`. After all, kernel enqueuing functions in the host side(clEnqueueNDRangeXXX) also require pre-allocated `__global` memory, we should follow the same fashion.

> What we can improve though is avoiding extra data copy using the copy helpers you are proposing (this can also be achieved by calling mempy passing the capture offset pointer into block_literal and captures size instead of the the whole block_literal as highlighted above). We can also potentially avoid reloading of the captures in the enqueued kernel though the capture flattening, but this depends on the calling convention (or rather enqueueing convension I would say).

It seems that the flattening approach leave little space for the implementation, which violate the generic property of llvm. Also, although cap_copy_helper looks helpful, I think there is little chance for one to copy individual captured variables - copy the entire block_literal is sufficient. Of course, we can reduce block_literal size by removing redundant fields for the sake of optimization but that's another discussion topic I think.

I would like to remove this code review. 
@Anastasia thanks for your patient discussing with newbie llvm/clang developer like me :)


https://reviews.llvm.org/D24715