Merge pull request #126 from r-devulap/perf-objsort

Use uint32_t instead of size_t for object sort

Author: Raghuveer Devulapalli

.github/workflows/build-numpy.yml

@@ -51,8 +51,8 @@ jobs:
     - name: Install NumPy dependencies
       working-directory: ${{ github.workspace }}/numpy
       run: |
-        pip install -r build_requirements.txt
-        pip install -r test_requirements.txt
+        pip install -r requirements/build_requirements.txt
+        pip install -r requirements/test_requirements.txt
 
     - name: Update x86-simd-sort
       working-directory: ${{ github.workspace }}/numpy
@@ -116,8 +116,8 @@ jobs:
     - name: Install NumPy dependencies
       working-directory: ${{ github.workspace }}/numpy
       run: |
-        pip install -r build_requirements.txt
-        pip install -r test_requirements.txt
+        pip install -r requirements/build_requirements.txt
+        pip install -r requirements/test_requirements.txt
 
     - name: Update x86-simd-sort
       working-directory: ${{ github.workspace }}/numpy

benchmarks/bench-objsort.hpp

@@ -26,7 +26,7 @@ struct Point3D {
             return std::sqrt(x * x + y * y + z * z);
         }
         else if constexpr (name == "taxicab") {
-            return abs(x) + abs(y) + abs(z);
+            return std::abs(x) + std::abs(y) + std::abs(z);
         }
         else if constexpr (name == "chebyshev") {
             return std::max(std::max(x, y), z);
@@ -98,13 +98,15 @@ static void simdobjsort(benchmark::State &state)
             ->Arg(10e5) \
             ->Arg(10e6);
 
-BENCHMARK_OBJSORT(simdobjsort, Point3D, double, x)
-BENCHMARK_OBJSORT(scalarobjsort, Point3D, double, x)
-BENCHMARK_OBJSORT(simdobjsort, Point3D, float, x)
-BENCHMARK_OBJSORT(scalarobjsort, Point3D, float, x)
-BENCHMARK_OBJSORT(simdobjsort, Point3D, double, taxicab )
-BENCHMARK_OBJSORT(scalarobjsort, Point3D, double, taxicab)
-BENCHMARK_OBJSORT(simdobjsort, Point3D, double, euclidean)
-BENCHMARK_OBJSORT(scalarobjsort, Point3D, double, euclidean)
-BENCHMARK_OBJSORT(simdobjsort, Point3D, double, chebyshev)
-BENCHMARK_OBJSORT(scalarobjsort, Point3D, double, chebyshev)
+#define BENCH_ALL(dtype) \
+    BENCHMARK_OBJSORT(simdobjsort, Point3D, dtype, x) \
+    BENCHMARK_OBJSORT(scalarobjsort, Point3D, dtype, x) \
+    BENCHMARK_OBJSORT(simdobjsort, Point3D, dtype, taxicab ) \
+    BENCHMARK_OBJSORT(scalarobjsort, Point3D, dtype, taxicab) \
+    BENCHMARK_OBJSORT(simdobjsort, Point3D, dtype, euclidean) \
+    BENCHMARK_OBJSORT(scalarobjsort, Point3D, dtype, euclidean) \
+    BENCHMARK_OBJSORT(simdobjsort, Point3D, dtype, chebyshev) \
+    BENCHMARK_OBJSORT(scalarobjsort, Point3D, dtype, chebyshev) \
+
+BENCH_ALL(double)
+BENCH_ALL(float)

lib/x86simdsort.h

@@ -43,7 +43,7 @@ keyvalue_qsort(T1 *key, T2* val, size_t arrsize, bool hasnan = false);
 
 // sort an object
 template <typename T, typename Func>
-XSS_EXPORT_SYMBOL void object_qsort(T *arr, size_t arrsize, Func key_func)
+XSS_EXPORT_SYMBOL void object_qsort(T *arr, uint32_t arrsize, Func key_func)
 {
     /* (1) Create a vector a keys */
     using return_type_of =
@@ -55,9 +55,9 @@ XSS_EXPORT_SYMBOL void object_qsort(T *arr, size_t arrsize, Func key_func)
     }
 
     /* (2) Call arg based on keys using the keyvalue sort */
-    std::vector<size_t> arg(arrsize);
+    std::vector<uint32_t> arg(arrsize);
     std::iota(arg.begin(), arg.end(), 0);
-    keyvalue_qsort(keys.data(), arg.data(), arrsize);
+    x86simdsort::keyvalue_qsort(keys.data(), arg.data(), arrsize);
 
     /* (3) Permute obj array in-place */
     std::vector<bool> done(arrsize);