release 0.9.1

Author: unknown

CMakeLists.txt

@@ -18,7 +18,7 @@
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 # THE SOFTWARE.
 
-cmake_minimum_required (VERSION 3.1)
+cmake_minimum_required (VERSION 2.8)
 project(amdovx)
 
 add_subdirectory(openvx)

README.md

@@ -4,7 +4,7 @@ AMD OpenVX (beta preview) is a highly optimized open source implementation of th
 #### Features
 * The code is highly optimized for both x86 CPU and OpenCL for GPU
 * Supported hardware spans the range from low power embedded APUs (like the new G series) to laptop, desktop and workstation graphics
-* Supports Windows and Linux
+* Supports Windows, Linux, and OS X
 * Includes a “graph optimizer” that looks at the entire processing pipeline and removes/replaces/merges functions to improve performance and minimize bandwidth at runtime 
 * Scripting support allows for rapid prototyping, without re-compiling at production performance levels
 * Interoperates with the popular (open source library) OpenCV
@@ -18,13 +18,17 @@ Build this project to generate AMD OpenVX library and RUNVX executable.
 ## Build Instructions
 
 #### Pre-requisites
-* AMD APP SDK 3.0 [download](http://developer.amd.com/tools-and-sdks/opencl-zone/amd-accelerated-parallel-processing-app-sdk/).
+* CMake 2.8 or newer [download](http://cmake.org/download/).
 * OpenCV 3.0 [download](http://opencv.org/downloads.html).
-* CMake 3.1 or newer [download](http://cmake.org/download/).
 * OpenCV_DIR environment variable should point to OpenCV/build folder
+* CPU: SSE4.1 or above CPU, 64-bit.
+* GPU: Radeon R7 Series or above (Kaveri+ APU), Radeon 3xx Series or above (optional)
+  * DRIVER: AMD Catalyst 15.7 or higher (version 15.20) with OpenCL 2.0 runtimes
+  * AMD APP SDK 3.0 [download](http://developer.amd.com/tools-and-sdks/opencl-zone/amd-accelerated-parallel-processing-app-sdk/).
+* libssl-dev on linux (optional)
 
 #### Build using Visual Studio Professional 2013 on 64-bit Windows 10/8.1/7
 * Use amdovx-core/amdovx.sln to build for x64 platform
 
-#### Build using CMake on Linux (Ubuntu 15.10 64-bit)
+#### Build using CMake
 * Use CMake to configure and generate Makefile

openvx/CMakeLists.txt

@@ -18,12 +18,12 @@
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 # THE SOFTWARE.
 
-cmake_minimum_required (VERSION 3.1)
+cmake_minimum_required (VERSION 2.8)
 project (openvx)
 
 set (CMAKE_CXX_STANDARD 11)
 
-find_package(OpenCL)
+find_package(OpenCL QUIET)
 
 include_directories(include ago api)
 
@@ -63,7 +63,7 @@ list(APPEND SOURCES
 	api/vx_nodes.cpp
 )
 
-add_library(openvx STATIC ${SOURCES})
+add_library(openvx SHARED ${SOURCES})
 
 if (OpenCL_FOUND)
 	target_compile_definitions(openvx PUBLIC ENABLE_OPENCL=1)
@@ -73,10 +73,14 @@ else(OpenCL_FOUND)
 	target_compile_definitions(openvx PUBLIC ENABLE_OPENCL=0)
 endif(OpenCL_FOUND)
 
-if ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "GNU")
-	set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -msse4.2 -mxop")
+if( POLICY CMP0054 )
+  cmake_policy( SET CMP0054 OLD )
+endif()
+if ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "MSVC")
+	set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /MT /DVX_API_ENTRY=__declspec(dllexport)")
+	set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} /MTd /DVX_API_ENTRY=__declspec(dllexport)")
+else()
+	set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -msse4.2 -std=c++11")
 	target_link_libraries(openvx dl)
-elseif ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "MSVC")
-	set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /MT")
-	set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} /MTd")
 endif()
+

openvx/ago/ago_drama.cpp

@@ -309,12 +309,12 @@ int agoOptimizeDramaComputeGraphHierarchy(AgoGraph * graph)
 #endif
 				if (data->outputUsageCount > 1) {
 					vx_status status = VX_ERROR_MULTIPLE_WRITERS;
-					agoAddLogEntry(&graph->ref, status, "ERROR: vxVerifyGraph: kernel %s: multiple writers for argument#%d (%s)\n", node->akernel->name, arg, data->name);
+					agoAddLogEntry(&graph->ref, status, "ERROR: vxVerifyGraph: kernel %s: multiple writers for argument#%d (%s)\n", node->akernel->name, arg, data->name.c_str());
 					return status;
 				}
 				else if (data->isVirtual && data->outputUsageCount == 0 && !data->isInitialized) {
 					vx_status status = VX_ERROR_MULTIPLE_WRITERS;
-					agoAddLogEntry(&graph->ref, status, "ERROR: vxVerifyGraph: kernel %s: no writer/initializer for virtual buffer at argument#%d (%s)\n", node->akernel->name, arg, data->name);
+					agoAddLogEntry(&graph->ref, status, "ERROR: vxVerifyGraph: kernel %s: no writer/initializer for virtual buffer at argument#%d (%s)\n", node->akernel->name, arg, data->name.c_str());
 					return status;
 				}
 			}

openvx/ago/ago_drama_remove.cpp

@@ -1241,8 +1241,8 @@ int agoOptimizeDramaRemoveNodeMerge(AgoGraph * agraph)
 				}
 				// transfer configuration from rule to childnode
 				childnode->attr_affinity = attr_affinity;
-				childnode->attr_border_mode;
-				childnode->callback;
+				//childnode->attr_border_mode = attr_border_mode;
+				//childnode->callback = callback;
 				debug_printf("INFO: agoOptimizeDramaRemoveNodeMerge: added node %s\n", childnode->akernel->name);
 				// verify the node
 				if (agoVerifyNode(childnode)) {

openvx/ago/ago_haf_cpu_canny.cpp

@@ -589,12 +589,6 @@ int HafCpu_CannySuppThreshold_U8XY_U16_3x3
 		}
 	}
 	*pxyStackTop = (vx_uint32)(pxyStack - xyStack);
-//	printf("Number of points in XY stack: %d\n", *pxyStackTop);
-#if 0
-	FILE *fp = fopen("c:\\temp\\norm.yuv", "wb");
-	if (fp) fwrite(pDst, 1, 640 * 480, fp);
-	fclose(fp);
-#endif
 	return AGO_SUCCESS;
 }
 

openvx/ago/ago_haf_cpu_geometric.cpp

@@ -134,8 +134,8 @@ int HafCpu_Remap_U8_U8_Nearest
 		if (extra_pixels){
 			unsigned char *pd = (unsigned char *)pdst;
 			for (unsigned int i = 0; i < extra_pixels; i++, pMapY_X++){
-				int x = (pMapY_X->x != 0xFFFF) ? (pMapY_X->x >> 3) + ((pMapY_X->x&7)>>2): 0;
-				int y = (pMapY_X->y != 0xFFFF) ? (pMapY_X->y >> 3) + ((pMapY_X->y&7)>>2) : 0;
+				int x = (pMapY_X->x != -1) ? (pMapY_X->x >> 3) + ((pMapY_X->x&7)>>2): 0;
+				int y = (pMapY_X->y != -1) ? (pMapY_X->y >> 3) + ((pMapY_X->y&7)>>2) : 0;
 				pd[i] = pSrcImage[y*srcImageStrideInBytes + x];
 			}
 		}
@@ -234,8 +234,8 @@ int HafCpu_Remap_U8_U8_Nearest_Constant
 		if (extra_pixels){
 			unsigned char *pd = (unsigned char *)pdst;
 			for (unsigned int i = 0; i < extra_pixels; i++, pMapY_X++){
-				int x = (pMapY_X->x != 0xFFFF) ? (pMapY_X->x >> 3) + ((pMapY_X->x & 7) >> 2) : border;
-				int y = (pMapY_X->y != 0xFFFF) ? (pMapY_X->y >> 3) + ((pMapY_X->y & 7) >> 2) : border;
+				int x = (pMapY_X->x != -1) ? (pMapY_X->x >> 3) + ((pMapY_X->x & 7) >> 2) : border;
+				int y = (pMapY_X->y != -1) ? (pMapY_X->y >> 3) + ((pMapY_X->y & 7) >> 2) : border;
 				pd[i] = pSrcImage[y*srcImageStrideInBytes + x];
 			}
 		}
@@ -1939,8 +1939,8 @@ ago_scale_matrix_t * matrix
 					(pSrcImage[M128I(mapx3).m128i_i32[2]] << 16) | (pSrcImage[M128I(mapx3).m128i_i32[3]] << 24);
 
 			}
-			while (x < dstWidth)
-				pDstImage[x] = pSrcImage[Xmap[x++] + yadd];
+			for (; x < dstWidth; x++)
+				pDstImage[x] = pSrcImage[Xmap[x] + yadd];
 
 			pDstImage += dstImageStrideInBytes;
 
@@ -1952,8 +1952,8 @@ ago_scale_matrix_t * matrix
 		{
 			unsigned int yadd = Ymap[y] * srcImageStrideInBytes;
 			x = 0;
-			while (x < dstWidth)
-				pDstImage[x] = pSrcImage[Xmap[x++] + yadd];
+			for (; x < dstWidth; x++)
+				pDstImage[x] = pSrcImage[Xmap[x] + yadd];
 			pDstImage += dstImageStrideInBytes;
 		}
 	}
@@ -2782,4 +2782,4 @@ int HafCpu_ScaleGaussianHalf_U8_U8_3x3
 		pDstImage += dstImageStrideInBytes;
 	}
 	return AGO_SUCCESS;
-}
+}

openvx/ago/ago_haf_cpu_opticalflow.cpp

@@ -196,7 +196,6 @@ vx_int32		   winsz
 		// calculate sharr derivatives Ix and Iy
 		ComputeSharr(dataStrideInBytes, pScratch, oldPyramid[level].width, oldPyramid[level].height, oldPyramid[level].strideInBytes, oldPyramid[level].pImage, pScharrScratch);
 		float ptScale = (float)(pow(pyramidScale, level));
-		//	printf("\nLevel : %d***************\n", level);
 
 		// do the Lukas Kanade tracking for each feature point
 		for (unsigned int pt = 0; pt < keyPointCount; pt++){
@@ -535,13 +534,11 @@ vx_int32		   winsz
 			if (!level){
 				newKeyPoint[pt].x = (vx_int32)(nextPt.x + halfWin - delta_dx + 0.5f);
 				newKeyPoint[pt].y = (vx_int32)(nextPt.y + halfWin - delta_dy + 0.5f);
-				//printf("Level: %d Key Point: %d  x: %d y: %d\n", level, pt, newKeyPoint[pt].x, newKeyPoint[pt].y);
 			}
 			else
 			{
 				pNextPtArray[pt].x = (nextPt.x + halfWin - delta_dx);
 				pNextPtArray[pt].y = (nextPt.y + halfWin - delta_dy);
-				//printf("Level: %d Key Point: %d  x: %d y: %d\n", level, pt, (int)pNextPtArray[pt].x, (int)pNextPtArray[pt].y);
 			}
 		}
 	}