- IntrinsicsVISC.td, visc.h, GenVISC.cpp
(2) Simplified GenVISC to easily add support for new intrinsics which have a 1
to 1 mapping instruction during code gen
(3) Added runtime api call to set ocl shared memory argument

on the kernel. This promotes the private memory allocation in each thread to
registers.

visc_sh version: manually unrolled the internal loops of kernel. This helps in promoting private array to registers. Commented out

(2) Fixed a bug in sgemm visc_sh version. Wrong grid size was used earlier,
resulting in the bug

request_mem failure
(2) modified visc_sh version of sgemm

edge to be from parent node, whereas an edge from parent node has Entry node as
its source. Hence the check SrcDFNode != PNode was always true. Changed to
!SrcDFNode->isDummyNode()

not producing empty struct as return type. Fixed it

:)

This one identifies allocation node and add s the property to the node. Can be
used by passes later

This field can be used to contain pointer to Node specific properties.
Currently used for Allocation Node property which is a list of Edge, Value pairs

__visc__barrier, __visc__getParentNode, __visc__getNumNodeInstances ...
(2) Also fixed bugs. bfs passing correctly through GenVISC pass

(2) EdgeDetection pipeline example working

Added singleNodeStream.ll and twoNodeStream.ll test cases

(1) Changes to DFGraph.h for identifying streaming graphs and edges
(2) Added/modified launch, wait, push, pop intrinsics for streaming graphs
(3) Changes to DFG2LLVM_X86.cpp and DFG2LLVM.h pass to support code gen for streaming
(4) BuildDFG.cpp modified to create streaming edges
(5) Modified visc-rt.cpp and visc-rt.h to add runtime support for streaming