SIMD Acceleration of Truth-Table Operations with AVX2

CMakeLists.txt

@@ -16,6 +16,16 @@ if (HAS_FCOLOR_DIAGNOSTICS)
   add_compile_options(-fcolor-diagnostics)
 endif()
 
+# compiler options for SIMD operations
+if(MSVC)
+  add_compile_options(/arch:AVX2)
+elseif(UNIX)
+  check_cxx_compiler_flag("-mavx2" HAS_MAVX2)
+  if(HAS_MAVX2)
+    add_compile_options(-mavx2)
+  endif()
+endif()
+
 if (UNIX)
   add_compile_options(-W -Wall)
 else()

docs/Doxyfile

@@ -1103,7 +1103,7 @@ IGNORE_PREFIX          =
 # If the GENERATE_HTML tag is set to YES, doxygen will generate HTML output
 # The default value is: YES.
 
-GENERATE_HTML          = NO
+GENERATE_HTML          = YES
 
 # The HTML_OUTPUT tag is used to specify where the HTML docs will be put. If a
 # relative path is entered the value of OUTPUT_DIRECTORY will be put in front of

docs/changelog.rst

@@ -4,6 +4,9 @@ Change Log
 Next release
 ------------
 
+* SIMD operations: header ``simd_operations`` (contributed by Andrea Costamagna)
+  `#142 <https://github.com/msoeken/kitty/pull/142>`_
+
 v0.8 (September 9, 2022)
 ------------------------
 

docs/index.rst

@@ -21,6 +21,7 @@ Welcome to kitty's documentation!
 
    truth_table_data_structures
    bit_operations
+   simd_operations
    algorithm
    constructors
    operations

docs/simd_operations.rst

@@ -0,0 +1,20 @@
+Single Instruction Multiple Data Operations
+===========================================
+
+The header ``<kitty/simd_operations.hpp>`` implements bitwise operations
+that are optimized through vectorization, if supported by the machine on
+which the code is compiled. It is suggested to replace the traditional 
+operations with these alternatives for truth table with more than `1024`
+bits. However, compiler dependent profiling is required to identify the
+performance cutoff.
+
+The header provides the following functions:
+
+.. doc_brief_table::
+   simd::bitwise_and
+   simd::bitwise_or
+   simd::bitwise_xor
+   simd::bitwise_lt
+   simd::unary_not
+   simd::set_zero
+   simd::set_ones

include/kitty/kitty.hpp

@@ -55,6 +55,7 @@
 #include "permutation.hpp"
 #include "print.hpp"
 #include "properties.hpp"
+#include "simd_operations.hpp"
 #include "spectral.hpp"
 #include "spp.hpp"
 #include "traits.hpp"