Need to extend when we want to support more than one nodes providing outputs.   

generating multiplication of local and global dim statements even in case of 1
level node, when local dim is null
M    llvm/projects/visc-rt/visc-rt.cpp
M    llvm/lib/Transforms/DFG2LLVM_NVPTX/DFG2LLVM_NVPTX.cpp

(2) Changes to DFGraph.h and visc-rt.cpp were not committed before.

M    llvm/test/VISC/parboil/benchmarks/mri-q/src/opencl/kernels.cl
M    llvm/test/VISC/parboil/benchmarks/mri-q/src/opencl/main.c
M    llvm/test/VISC/parboil/benchmarks/mri-q/src/opencl/macros.h
M    llvm/test/VISC/parboil/benchmarks/mri-q/src/opencl/ocl.c
M    llvm/test/VISC/parboil/benchmarks/mri-q/Makefile
M    llvm/include/llvm/IR/DFGraph.h
M    llvm/projects/visc-rt/visc-rt.cpp

(2) PTX Pass changed to support parameters being used to pass grid and block
    size to kernel nodes.
M    Transforms/ClearDFG/ClearDFG.cpp
M    Transforms/DFG2LLVM_NVPTX/DFG2LLVM_NVPTX.cpp

    set. This is important so as not to fail in case of 2-level PTX kernel,
    where the intermediate node does not have a gen function, but would pass
    because its parent has one.
(2) Modified PTX pass to
    i.  Correct mapping of arguments between intermediate and kernel launch node
    ii. Allow both constants and arguments as node limit values
    iii.Generate get_group_id and get_global_id correctly

M    Transforms/DFG2LLVM_X86/DFG2LLVM_X86.cpp
M    Transforms/DFG2LLVM_NVPTX/DFG2LLVM_NVPTX.cpp

call of exec node. Useful for codegen of 2-level hierarchy kernels

parent instead of getInValAt() function

    in out attributes of arguments
(2) check-point commit in preparing mri-q benchmark

line. metadata.py is not required anymore

environment variable LLVM_SRC_ROOT is set, then running test cases should work.
The passes and llvm-lit now use this variable to get llvm source root and define
all paths relative to it

Changes to makefile and visc-rt to hide print statements by default and only enable when using make DEBUG=1

(2) Modified both MatrixMultiplication test cases to have empty return structs
    types
(3) visc-rt changed to reflect the same