Merge pull request #28 from ingowald/main

ray traversal kernels will do distance tracking on stack

Author: iwald-nvidia

.github/workflows/Ubuntu.yml

@@ -35,7 +35,7 @@ jobs:
 
     steps:
 
-      - name: Checkout OWL
+      - name: Checkout cuBQL
         uses: actions/checkout@v5
         with:
           submodules: true

.github/workflows/Windows.yml

@@ -49,7 +49,7 @@ jobs:
       visual_studio: "${{ matrix.visual-studio }}"
       build_dir: "build"
     steps:
-      - name: Checkout owl
+      - name: Checkout cuBQL
         uses: actions/checkout@v5
         with:
           submodules: true

CMakeLists.txt

@@ -1,11 +1,17 @@
-# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA
+# CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: Apache-2.0
 
 cmake_minimum_required(VERSION 3.16)
 
+set(CUBQL_VERSION_MAJOR 1)
+set(CUBQL_VERSION_MINOR 3)
+set(CUBQL_VERSION_PATCH 0)
+set(CUBQL_VERSION ${CUBQL_VERSION_MAJOR}.${CUBQL_VERSION_MINOR}.${CUBQL_VERSION_PATCH})
+
 cmake_policy(SET CMP0048 NEW)
 set(CMAKE_BUILD_TYPE_INIT "Release")
-project(cuBQL VERSION 1.2.0 LANGUAGES C CXX)
+project(cuBQL VERSION ${CUBQL_VERSION} LANGUAGES C CXX)
 
 if (CUBQL_OMP)
   set(CUBQL_DISABLE_CUDA ON)

cuBQL/builder/cuda/wide_gpu_builder.h

@@ -137,10 +137,6 @@ namespace cuBQL {
       }
       while (numWritten < N) {
         target.children[numWritten].bounds.set_empty();
-        // lower
-        //   = make_float3(+INFINITY,+INFINITY,+INFINITY);
-        // target.children[numWritten].bounds.upper
-        //   = make_float3(-INFINITY,-INFINITY,-INFINITY);
         target.children[numWritten].offset = (uint32_t)-1;
         target.children[numWritten].count  = (uint32_t)-1;
         target.children[numWritten].valid  = 0;

cuBQL/math/affine.h

@@ -221,4 +221,8 @@ namespace cuBQL {
   using affine3f = AffineSpace3f;
   using affine2d = AffineSpace2d;
   using affine3d = AffineSpace3d;
+
+  template<typename L>
+  inline __cubql_both dbgout operator<<(dbgout o, const AffineSpaceT<L> &v)
+  { o << "Affine(" << v.l << "+" << v.p << ")"; return o; }
 } // ::cuBQL

cuBQL/math/conservativeDistances.h

@@ -62,7 +62,7 @@ namespace cuBQL {
   inline __cubql_both float fSqrLength_rd(vec_t<T,D> v)
   {
     float sum = 0.f;
-#pragma unroll
+// #pragma unroll
     for (int i=0;i<D;i++)
       sum += fSquare_rd(v[i]);
     return sum;

cuBQL/math/linear.h

@@ -328,6 +328,11 @@ namespace cuBQL {
     return cout << "{ vx = " << m.vx << ", vy = " << m.vy << ", vz = " << m.vz << "}";
   }
 
+  template<typename T>
+  inline __cubql_both dbgout operator<<(dbgout o, const LinearSpace3<T> &m)
+  { o  << "{ vx = " << m.vx << ", vy = " << m.vy << ", vz = " << m.vz << "}"; return o; }
+  
+
   /*! Shortcuts for common linear spaces. */
   using LinearSpace2f  = LinearSpace2<vec2f> ;
   using LinearSpace3f  = LinearSpace3<vec3f> ;

cuBQL/math/math.h

@@ -22,7 +22,8 @@ namespace cuBQL {
 #endif
 
 #ifdef __CUDA_ARCH__
-# define CUBQL_INF ::cuda::std::numeric_limits<float>::infinity()
+# define CUBQL_INF CUDART_INF_F
+// # define CUBQL_INF ::cuda::std::numeric_limits<float>::infinity()
 #else
 # define CUBQL_INF std::numeric_limits<float>::infinity()
 #endif
@@ -45,6 +46,7 @@ namespace cuBQL {
 
   /*! square of a value */
   inline __cubql_both float sqr(float f) { return f*f; }
+  inline __cubql_both double sqr(double f) { return f*f; }
 
   /*! unary functors on scalar types, so we can lift them to vector types later on */
   inline __cubql_both float  rcp(float f)     { return 1.f/f; }

cuBQL/math/vec.h

@@ -669,7 +669,7 @@ namespace cuBQL {
   inline __cubql_both bool operator==(const vec_t_data<T,D> &a,
                                       const vec_t_data<T,D> &b)
   {
-#pragma unroll
+    CUBQL_PRAGMA_UNROLL
     for (int i=0;i<D;i++)
       if (a[i] != b[i]) return false;
     return true;

cuBQL/queries/triangleData/Triangle.h

@@ -15,20 +15,6 @@ namespace cuBQL {
   // *** INTERFACE ***
   // =========================================================================
 
-  // /*! a simple triangle consisting of three vertices. In order to not
-  //   overload this class with too many functions the actual
-  //   operations on triangles - such as intersectin with a ray,
-  //   computing distance to a point, etc - will be defined in the
-  //   respective queries */
-  // struct Triangle {
-  //   /*! returns an axis aligned bounding box enclosing this triangle */
-  //   inline __cubql_both box3f bounds() const;
-  //   inline __cubql_both vec3f sample(float u, float v) const;
-  //   inline __cubql_both vec3f normal() const;
-    
-  //   vec3f a, b, c;
-  // };
-
   template<typename T>
   struct triangle_t
   {