RXMesh v0.2.0 (#3)

**Highlights:**
- Eliminate global index space and identify mesh elements using the combination of their patch index and their local index within the patch
- All query operation no longer needs to map their output to global index space
- Allocating `Attributes` per-patch basis instead of a single large array 
- Introduce `Vertex/Edge/FaceHandle` to improve type safety which is used to identify different mesh element and index the mesh `Attributes`
- Introduce `LocalVertex/Edge/FaceT` to improve type safety for internal implementation of local index space 
- `Attributes` are now managed by `RXMeshStatic` using `add_vertex/edge/face_attribute()` API
- Introduce `for_each_vertex/edge/face()` API in `RXMeshStatic` for simple operations on the mesh that do not require query operations (i.e., map operations) for both CUDA and OpenMP backend
- Introduce `ReduceHandle` to do reduction operations on `Attribute` and so reduction operation temp memory is no longer handle by `Attribute` itself 
- Improve the documentation for most of the user-facing APIs
- Removing code related to shuffling and sorting the input mesh (it was only relevant to reproduce SIGGRAPH paper results) 
- Accurately report the register usage and static shared memory used by the kernel by using the (pointer to) kernel itself instead of the prototype function 
- Improve initializing the data structure by removing duplicate supporting structure that were created by both `RXMesh` and `Patcher` and by using OpenMP when possible in `RXMesh`

**Known Issues:**
- Some queries (VE, VF, and higher queries) now require more registers per thread and thus they might fail on some (less powerful) GPUs. 

Co-authored-by: Ahmed Mahmoud <[email protected]>

.clang-format

@@ -4,14 +4,16 @@ BasedOnStyle:  Chromium
 TabWidth: 4
 UseTab: Never
 
+AlignConsecutiveAssignments: true
 AllowShortFunctionsOnASingleLine: false
 AllowShortIfStatementsOnASingleLine: false
 AllowShortLoopsOnASingleLine: false
 AlwaysBreakTemplateDeclarations: true
 AlignTrailingComments: true
+BinPackArguments: false
 BinPackParameters: false
 BreakBeforeTernaryOperators: false
-ConstructorInitializerAllOnOneLineOrOnePerLine: false
+ConstructorInitializerAllOnOneLineOrOnePerLine: true
 Cpp11BracedListStyle: true
 IndentCaseLabels: true
 IndentWidth:     4

.github/workflows/Ubuntu.yml

@@ -1,5 +1,5 @@
 name: Ubuntu
-on: [push, pull_request]
+on: [push, pull_request, workflow_dispatch]
 jobs:
   UbuntuRun:
     runs-on: ubuntu-latest

.github/workflows/Windows.yml

@@ -1,5 +1,5 @@
 name: Windows
-on: [push, pull_request]
+on: [push, pull_request, workflow_dispatch]
 jobs:
   WindowsRun:
     runs-on: windows-latest

.gitignore

@@ -1,4 +1,5 @@
 output/
+input/
 build/
 include/rxmesh/util/git_sha1.cpp
 .vscode/

CMakeLists.txt

@@ -1,11 +1,11 @@
-cmake_minimum_required(VERSION 3.15 FATAL_ERROR)
+cmake_minimum_required(VERSION 3.18 FATAL_ERROR)
 
 if (CMAKE_VERSION VERSION_GREATER_EQUAL 3.18)
   cmake_policy(SET CMP0104 OLD)
 endif()
 
 project(RXMesh 
-        VERSION 0.1.0 
+        VERSION 0.2.0 
         LANGUAGES C CXX CUDA)
 
 set(CMAKE_CXX_STANDARD 17)
@@ -86,15 +86,16 @@ target_sources(RXMesh_header_lib
 
 # CUDA and C++ compiler flags
 set(cxx_flags 
-	$<$<CXX_COMPILER_ID:MSVC>:-D_SCL_SECURE_NO_WARNINGS /openmp /std:c++17> #Add MSVC-specific compiler flags here
-	$<$<CXX_COMPILER_ID:GNU>:-Wall -m64 -fopenmp -O3 -std=c++17>            #Add GCC/Clang-specific compiler flags here
+	$<$<CXX_COMPILER_ID:MSVC>:-D_SCL_SECURE_NO_WARNINGS /openmp /std:c++17>           #Add MSVC-specific compiler flags here
+	$<$<CXX_COMPILER_ID:GNU>:-Wall -m64 -fopenmp -O3 -std=c++17 -Wno-unused-function> #Add GCC/Clang-specific compiler flags here
 	)
 set(cuda_flags
-    -Xcompiler=$<$<CXX_COMPILER_ID:GNU>:-Wall -fopenmp -O3>
+    -Xcompiler=$<$<CXX_COMPILER_ID:GNU>:-Wall -fopenmp -O3 -Wno-unused-function>
     #Disables warning
     #177-D "function XXX was declared but never referenced"
     -Xcudafe "--display_error_number --diag_suppress=177"
     ${CUDA_ARCHS}
+    -rdc=true
 	-lineinfo	
 	--expt-extended-lambda	
 	-use_fast_math	
@@ -111,7 +112,6 @@ target_compile_options(developer_flags INTERFACE
     $<$<COMPILE_LANGUAGE:CUDA>:${cuda_flags}>
 )
 
-
 target_link_libraries(RXMesh_header_lib INTERFACE $<BUILD_INTERFACE:developer_flags>)
 
 #OpenMP

LICENSE

@@ -1,6 +1,6 @@
 BSD 2-Clause License
 
-Copyright (c) 2021, owensgroup
+Copyright (c) 2022, owensgroup
 All rights reserved.
 
 Redistribution and use in source and binary forms, with or without

apps/Filtering/CMakeLists.txt

@@ -27,6 +27,8 @@ endif()
 
 set_target_properties(Filtering PROPERTIES FOLDER "apps")
 
+set_property(TARGET Filtering PROPERTY CUDA_SEPARABLE_COMPILATION ON)
+
 source_group(TREE ${CMAKE_CURRENT_LIST_DIR} PREFIX "Filtering" FILES ${SOURCE_LIST})
 
 target_link_libraries(Filtering 

apps/Filtering/benchmark.sh

@@ -1,5 +1,4 @@
 #!/bin/bash
-echo "This script re-generates RXMesh data in Figure 8(c) in the paper."
 echo "Please make sure to first compile the source code and then enter the input OBJ files directory."
 read -p "OBJ files directory (no trailing slash): " input_dir
 
@@ -16,7 +15,7 @@ device_id=0
 
 for file in $input_dir/*.obj; do 	 
     if [ -f "$file" ]; then
-		echo $exe -p -input "$file" -num_filter_iter 5 -device_id $device_id
-             $exe -p -input "$file" -num_filter_iter 5 -device_id $device_id
+		echo $exe -input "$file" -num_filter_iter 5 -device_id $device_id
+         $exe -input "$file" -num_filter_iter 5 -device_id $device_id
     fi 
 done

apps/Filtering/filtering.cu

@@ -13,34 +13,22 @@
 
 struct arg
 {
-    std::string obj_file_name = STRINGIFY(INPUT_DIR) "sphere3.obj";
-    std::string output_folder = STRINGIFY(OUTPUT_DIR);
-    uint32_t    device_id = 0;
+    std::string obj_file_name   = STRINGIFY(INPUT_DIR) "sphere3.obj";
+    std::string output_folder   = STRINGIFY(OUTPUT_DIR);
+    uint32_t    device_id       = 0;
     uint32_t    num_filter_iter = 5;
     char**      argv;
     int         argc;
-    bool        shuffle = false;
-    bool        sort = false;
-
 } Arg;
 
 #include "filtering_openmesh.h"
 #include "filtering_rxmesh.cuh"
 
 TEST(App, Filtering)
 {
-    using namespace RXMESH;
+    using namespace rxmesh;
     using dataT = float;
 
-
-    if (Arg.shuffle) {
-        ASSERT_FALSE(Arg.sort) << " cannot shuffle and sort at the same time!";
-    }
-    if (Arg.sort) {
-        ASSERT_FALSE(Arg.shuffle)
-            << " cannot shuffle and sort at the same time!";
-    }
-
     // Select device
     cuda_query(Arg.device_id);
 
@@ -50,42 +38,27 @@ TEST(App, Filtering)
     std::vector<std::vector<dataT>>    Verts;
     ASSERT_TRUE(import_obj(Arg.obj_file_name, Verts, Faces));
 
-    if (Arg.shuffle) {
-        shuffle_obj(Faces, Verts);
-    }
 
-    // Create RXMeshStatic instance. If Arg.sort is true, Faces and Verts will
-    // be sorted based on the patching happening inside RXMesh
-    RXMeshStatic<PATCH_SIZE> rxmesh_static(Faces, Verts, Arg.sort, false);
+    TriMesh input_mesh;
+    ASSERT_TRUE(OpenMesh::IO::read_mesh(input_mesh, Arg.obj_file_name));
 
+    ASSERT_EQ(input_mesh.n_vertices(), Verts.size());
 
-    // Since OpenMesh only accepts input as obj files, if the input mesh is
-    // shuffled or sorted, we have to write it to a temp file so that OpenMesh
-    // can pick it up
-    TriMesh input_mesh;
-    if (Arg.sort || Arg.shuffle) {
-        export_obj(Faces, Verts, "temp.obj", false);
-        ASSERT_TRUE(OpenMesh::IO::read_mesh(input_mesh, "temp.obj"));
-    } else {
-        ASSERT_TRUE(OpenMesh::IO::read_mesh(input_mesh, Arg.obj_file_name));
-    }
+    // OpenMesh Impl
+    std::vector<std::vector<dataT>> ground_truth(Verts);
+    size_t                          max_neighbour_size = 0;
+    filtering_openmesh(
+        omp_get_max_threads(), input_mesh, ground_truth, max_neighbour_size);
 
-    //*** OpenMesh Impl
-    RXMESH::RXMeshAttribute<dataT> ground_truth;
-    size_t max_neighbour_size = 0;
-    filtering_openmesh(omp_get_max_threads(), input_mesh, ground_truth, max_neighbour_size);
-
 
-    //*** RXMesh Impl
-    filtering_rxmesh(rxmesh_static, Verts, ground_truth, max_neighbour_size);
+    // RXMesh Impl
+    filtering_rxmesh(Faces, Verts, ground_truth, max_neighbour_size);
 
-    // Release allocation
-    ground_truth.release();
 }
 
 int main(int argc, char** argv)
 {
-    using namespace RXMESH;
+    using namespace rxmesh;
     Log::init();
 
     ::testing::InitGoogleTest(&argc, argv);
@@ -101,9 +74,7 @@ int main(int argc, char** argv)
                         "                    Default is {} \n"
                         "                    Hint: Only accepts OBJ files\n"
                         " -o:                JSON file output folder. Default is {} \n"
-                        " -num_filter_iter:  Iteration count. Default is {} \n"                        
-                        " -s:                Shuffle input. Default is false.\n"
-                        " -p:                Sort input using patching output. Default is false.\n"
+                        " -num_filter_iter:  Iteration count. Default is {} \n"
                         " -device_id:        GPU device ID. Default is {}",
              Arg.obj_file_name, Arg.output_folder ,Arg.num_filter_iter ,Arg.device_id);
             // clang-format on
@@ -123,12 +94,6 @@ int main(int argc, char** argv)
             Arg.output_folder =
                 std::string(get_cmd_option(argv, argv + argc, "-o"));
         }
-        if (cmd_option_exists(argv, argc + argv, "-s")) {
-            Arg.shuffle = true;
-        }
-        if (cmd_option_exists(argv, argc + argv, "-p")) {
-            Arg.sort = true;
-        }
         if (cmd_option_exists(argv, argc + argv, "-device_id")) {
             Arg.device_id =
                 atoi(get_cmd_option(argv, argv + argc, "-device_id"));