fast_statistics 0.1.1 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 77ab0c69365318bdd1f68d562c0460ed573d12bf1301542e65fe7ff204f70118
-  data.tar.gz: ff64565f990ea2d6f77c52afbb63d7271f25737744be191146d5935d69727a30
+  metadata.gz: f74e28e9d460ef4ebcab5b0efa21ef0bd7aa1bc17ec398292dbc90fb2f759c10
+  data.tar.gz: 1ab07eeea3e7e139b0305b0d4109f780d2d7f372a0c3d8998ffc297a690f8eef
 SHA512:
-  metadata.gz: 898331717252505cff2e972ee2fb380709b070209197bf56374fe3c398aa906945c71a543f435d493aa5f5c5f7e7c65100979772f56bd5d28eb68967788fa6d3
-  data.tar.gz: e725e1d41c7749b8bc14eca78e46f574d477e12ddce3c8cd09817af82b4581ca3dcf24f2f2e2ea06128c2c65fe906ba365c38c7e911d20210b474b698065a01b
+  metadata.gz: 7ce73ccecf92df45a42789a45dab9d00ea0642d5b08904e07ecb43eb75148c33f4d645737cf1999fd2b9b7931e80b5efc9f3882756b62820464481ae8a954bfe
+  data.tar.gz: 8e9385b1776b36d9d45c402ccc3015ecc01c4c52a95fae4321f3f6409ebc050921192d2c049cb8b453c49a46defd899d25a84491dbb811f9e10ae3bc2caa2b60

data/ext/fast_statistics/array_2d.cpp

@@ -6,21 +6,32 @@ namespace array_2d
 
 DFloat::~DFloat()
 {
-  free(entries);
-  delete[] stats;
+  if (data_initialized) {
+    free(entries);
+    delete[] stats;
+  }
 }
 
-DFloat::DFloat(VALUE arrays)
+DFloat::DFloat(VALUE arrays, bool initialize_data)
 {
-  cols = rb_array_len(arrays);
-  rows = rb_array_len(rb_ary_entry(arrays, 0));
-  entries = (double*)malloc(cols * rows * sizeof(double));
-  stats = NULL;
-
-  for (int j = 0; j < cols; j++) {
-    for (int i = 0; i < rows; i++) {
-      entries[j * rows + i] = (double)NUM2DBL(rb_ary_entry(rb_ary_entry(arrays, j), i));
+  data_initialized = initialize_data;
+
+  if (initialize_data) {
+    cols = rb_array_len(arrays);
+    rows = rb_array_len(rb_ary_entry(arrays, 0));
+    entries = (double*)malloc(cols * rows * sizeof(double));
+    stats = new Stats[cols];
+
+    for (int j = 0; j < cols; j++) {
+      for (int i = 0; i < rows; i++) {
+        entries[j * rows + i] = (double)NUM2DBL(rb_ary_entry(rb_ary_entry(arrays, j), i));
+      }
     }
+  } else {
+    cols = 0;
+    rows = 0;
+    entries = NULL;
+    stats = NULL;
   }
 }
 
@@ -68,7 +79,7 @@ DFloat::standard_deviation(double* col, double mean)
 Stats*
 DFloat::descriptive_statistics()
 {
-  stats = new Stats[cols];
+  if (!data_initialized) return stats;
 
   for (int col = 0; col < cols; col++) {
     Stats var_stats;
@@ -137,6 +148,7 @@ Stats*
 DFloat::descriptive_statistics_packed()
 {
   stats = new Stats[cols];
+  if (!data_initialized) return stats;
   const int simd_pack_size = 2;
 
   __m128d lengths = _mm_set_pd1((double)rows);

data/ext/fast_statistics/array_2d.h

@@ -44,10 +44,11 @@ class DFloat
 public:
   int cols;
   int rows;
+  bool data_initialized;
   double* entries;
   Stats* stats;
 
-  DFloat(VALUE ruby_arr);
+  DFloat(VALUE ruby_arr, bool initialize_data);
   ~DFloat();
 
   Stats* descriptive_statistics();

data/ext/fast_statistics/fast_statistics.cpp

@@ -40,17 +40,16 @@ build_results_hashes(Stats* stats, int num_variables)
 
 // Common
 void
-free_wrapped_array(void* array)
+free_wrapped_array(void* dfloat)
 {
-  ((DFloat*)array)->~DFloat();
+  ((DFloat*)dfloat)->~DFloat();
+  free(dfloat);
 }
 
 size_t
 wrapped_array_size(const void* data)
 {
-  DFloat* array = (DFloat*)data;
-  size_t size = sizeof(array->entries) + sizeof(*array);
-  return size;
+  return sizeof(DFloat);
 }
 
 static rb_data_type_t dfloat_wrapper = [] {
@@ -65,21 +64,36 @@ static rb_data_type_t dfloat_wrapper = [] {
 VALUE
 cArray2D_alloc(VALUE self)
 {
-  void* dfloat = (void*)malloc(sizeof(void*));
+  void* dfloat = (void*)malloc(sizeof(DFloat));
 
   return TypedData_Wrap_Struct(self, &dfloat_wrapper, dfloat);
 }
 
-inline bool
-cArray2D_check_array_args(VALUE arrays)
+/*
+ * def initialize(arrays)
+ */
+VALUE
+cArray2D_initialize(VALUE self, VALUE arrays)
 {
-  if (TYPE(arrays) == T_ARRAY && TYPE(rb_ary_entry(arrays, 0)) == T_ARRAY) {
-    return true;
-  } else {
+  // Initialize dfloat structure to store Dfloat in type wrapper
+  void* dfloat;
+  UNWRAP_DFLOAT(self, dfloat);
+
+  // Type-check 2D array
+  if (TYPE(arrays) != T_ARRAY) {
+    new (dfloat) DFloat(arrays, false);
     Check_Type(arrays, T_ARRAY);
+    return false;
+  }
+
+  if (TYPE(rb_ary_entry(arrays, 0)) != T_ARRAY) {
+    new (dfloat) DFloat(arrays, false);
     Check_Type(rb_ary_entry(arrays, 0), T_ARRAY);
     return false;
   }
+
+  new (dfloat) DFloat(arrays, true);
+  return self;
 }
 
 //{{{ Unpacked
@@ -89,22 +103,6 @@ simd_disabled(VALUE self)
   return Qfalse;
 }
 
-/*
- * def initialize(arrays)
- */
-VALUE
-cArray2D_initialize_unpacked(VALUE self, VALUE arrays)
-{
-  // Typecheck 2d array
-  if (cArray2D_check_array_args(arrays)) {
-    // Initialize dfloat structure to store Dfloat in type wrapper
-    void* dfloat;
-    UNWRAP_DFLOAT(self, dfloat);
-    new (dfloat) DFloat(arrays);
-  }
-  return self;
-}
-
 /*
  * Unpacked descriptive statistics
  *
@@ -122,7 +120,7 @@ cArray2D_descriptive_statistics_unpacked(VALUE self)
 }
 //}}}
 
-// Packed
+//{{{ Packed
 #ifdef HAVE_XMMINTRIN_H
 extern "C" VALUE
 simd_enabled(VALUE self)
@@ -130,22 +128,6 @@ simd_enabled(VALUE self)
   return Qtrue;
 }
 
-/*
- * def initialize(arrays)
- */
-VALUE
-cArray2D_initialize_packed(VALUE self, VALUE arrays)
-{
-  // Typecheck 2d array
-  if (cArray2D_check_array_args(arrays)) {
-    // Initialize dfloat structure to store Dfloat in type wrapper
-    void* dfloat;
-    UNWRAP_DFLOAT(self, dfloat);
-    new (dfloat) DFloat(arrays);
-  }
-  return self;
-}
-
 /*
  * Packed descriptive statistics
  *
@@ -162,6 +144,7 @@ cArray2D_descriptive_statistics_packed(VALUE self)
   return build_results_hashes(stats, dfloat->cols);
 }
 #endif
+//}}}
 
 extern "C" void
 Init_fast_statistics(void)
@@ -169,10 +152,10 @@ Init_fast_statistics(void)
   mFastStatistics = rb_define_module("FastStatistics");
   cArray2D = rb_define_class_under(mFastStatistics, "Array2D", rb_cData);
   rb_define_alloc_func(cArray2D, cArray2D_alloc);
+  rb_define_method(cArray2D, "initialize", RUBY_METHOD_FUNC(cArray2D_initialize), 1);
 
 #ifdef HAVE_XMMINTRIN_H
   rb_define_singleton_method(mFastStatistics, "simd_enabled?", RUBY_METHOD_FUNC(simd_enabled), 0);
-  rb_define_method(cArray2D, "initialize", RUBY_METHOD_FUNC(cArray2D_initialize_packed), 1);
   rb_define_method(
     cArray2D,
     "descriptive_statistics",
@@ -180,7 +163,6 @@ Init_fast_statistics(void)
     0);
 #else
   rb_define_singleton_method(mFastStatistics, "simd_enabled?", RUBY_METHOD_FUNC(simd_disabled), 0);
-  rb_define_method(cArray2D, "initialize", RUBY_METHOD_FUNC(cArray2D_initialize_unpacked), 1);
   rb_define_method(
     cArray2D,
     "descriptive_statistics",

data/lib/fast_statistics/version.rb

@@ -1,3 +1,3 @@
 module FastStatistics
-  VERSION = "0.1.1"
+  VERSION = "0.2.0"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: fast_statistics
 version: !ruby/object:Gem::Version
-  version: 0.1.1
+  version: 0.2.0
 platform: ruby
 authors:
 - Martin Nyaga
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2021-03-23 00:00:00.000000000 Z
+date: 2021-03-24 00:00:00.000000000 Z
 dependencies: []
 description: Fast computation of descriptive statistics in ruby using native code
   and SIMD