First working version. Video of input and output is shown in a loop. Good enough edge detection (not great). Difference in speeds too. Not exploiting pipeline parallelism though

Added a version of pipeline example (visc_parallel) which computes gradient and compute max gradient in parallel

Pipeline demo: Fixed zero crossing node and changed the dilation and erosion kernel B for better edge detection

new compute max gradient kernel with global barrier. Now wokring though. barrier variable needs to be volatile which is not supported in hvisc

checkpoint commit for pipeline demo. All other nodes working except compute max gradient (reduction). Working for 3 configurations - all scalar, all vector, all gpu

Fixed pipeline benchmark to not specify return types of internal nodes. That is automatically inferred from bind out intrinsics

Changed GenVISC pass to automatically infer return type of internal nodes from bind output intrinsics

Fixed pipeline code. Allows multiple runs within the code. Fixed visc.mk to support SPIR and PTX both

Made visc-rt threadsafe by implementing locks across all ocl functions. Very crude. Fine grained locking possible if performance deteriotes