Refacto.

.gitignore

@@ -9,5 +9,6 @@ src_files.txt
 tests/build*
 tests/code_coverage_files/
 tests/code_coverage_report/
+src/gen/*
 
 build*

.gitlab-ci.yml

@@ -920,7 +920,7 @@ coverage-linux:
   artifacts:
      name: code-coverage-report
      paths:
-      - ./code_coverage_report/
+      - ./tests/code_coverage_report/
   script:
     - ./ci/coverage-linux.sh
 

CMakeLists.txt

@@ -19,32 +19,32 @@ message(STATUS "  * MIPP_TESTS_EXE: '${MIPP_TESTS_EXE}'")
 message(STATUS "  * MIPP_STATIC_LIB: '${MIPP_STATIC_LIB}'")
 
 set(MIPP_HEADERS
-  src/mipp.h
-  src/mipp_impl_AVX512.hxx
-  src/mipp_impl_AVX.hxx
-  src/mipp_impl_NEON.hxx
-  src/mipp_impl_SSE.hxx
-  src/mipp_object.hxx
-  src/mipp_scalar_op.h
-  src/mipp_scalar_op.hxx
+  include/mipp.h
+  include/mipp_impl_AVX512.hxx
+  include/mipp_impl_AVX.hxx
+  include/mipp_impl_NEON.hxx
+  include/mipp_impl_SSE.hxx
+  include/mipp_object.hxx
+  include/mipp_scalar_op.h
+  include/mipp_scalar_op.hxx
 )
 
 set(MIPP_MATH_HEADERS
-  src/math/avx512_mathfun.h
-  src/math/avx512_mathfun.hxx
-  src/math/avx_mathfun.h
-  src/math/avx_mathfun.hxx
-  src/math/neon_mathfun.h
-  src/math/neon_mathfun.hxx
-  src/math/sse_mathfun.h
-  src/math/sse_mathfun.hxx
+  include/math/avx512_mathfun.h
+  include/math/avx512_mathfun.hxx
+  include/math/avx_mathfun.h
+  include/math/avx_mathfun.hxx
+  include/math/neon_mathfun.h
+  include/math/neon_mathfun.hxx
+  include/math/sse_mathfun.h
+  include/math/sse_mathfun.hxx
 )
 
 if (MIPP_STATIC_LIB)
   set(MIPP_SRCS
-      src/mipp_SSE_LUT.cpp
-      src/mipp_AVX_LUT.cpp
-      src/mipp_NEON_LUT.cpp
+      src/gen/compress_LUT_AVX.cpp
+      src/gen/compress_LUT_SSE.cpp
+      src/gen/compress_LUT_NEON.cpp
   )
 endif()
 
@@ -112,7 +112,8 @@ if (MIPP_TESTS_EXE)
     set_target_properties(run_tests PROPERTIES OUTPUT_NAME run-tests POSITION_INDEPENDENT_CODE ON) # set -fpie
 
     # include MIPP headers
-	target_include_directories(run_tests PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/src/")
+	target_include_directories(run_tests PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/include/")
+
 	# include Catch2 header
     target_include_directories(run_tests PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/tests/lib/Catch2/include/")
 

gencode.py

@@ -1,90 +1,93 @@
-#!/usr/bin/python3
+#!/usr/bin/env python3
+
 from jinja2 import Environment, FileSystemLoader
+from pathlib import Path
+import sys
+
+if sys.version_info[0] != 3 or sys.version_info[1] < 5:
+    print("This script requires Python version 3.5")
+    sys.exit(1)
 
 env = Environment(loader = FileSystemLoader("codegen"))
 template_lut = env.get_template("vcompress-LUT.cpp.j2")
 template_file = env.get_template("mipp_LUT.cpp.j2")
 
-
 def generate_lut(entries, simdwidth, words_per_simd):
     lut = [[0 for j in range(0, simdwidth)] for i in range(0, entries)]
 
     elem_bytes = simdwidth // words_per_simd
-    
+
     for i in range(0, entries):
         mask = i
         j = 0
-        
+
         for k in range(0, words_per_simd):
             for b in range(0, elem_bytes):
                 lut[i][j + b] = k * elem_bytes + b
-                
+
             if mask & 1 == 1:
                 j += elem_bytes
             mask >>= 1
 
         for k in range(j, simdwidth):
             lut[i][k] = -1
-            
+
     return lut
 
-def write_all_luts(filename, all_luts):    
-
+def write_all_luts(filename, all_luts):
     all_content = template_file.render(
-	luts = all_luts,
+        luts = all_luts,
     )
 
     with open(filename, "w+") as file:
         file.write(all_content)
 
 def generate_luts(filename, simdname, simdwidth, entrydef, lut_params_list):
-
     (entrytype, entrybytes) = entrydef
-    
+
     all_luts = []
     for entries, simd_words in lut_params_list:
 
         elem_bits = (simdwidth // simd_words) * entrybytes * 8
-        
+
         lut = template_lut.render(
             lutname = f"vcompress_LUT{elem_bits}x{simd_words}_{simdname}",
             entries = entries,
             simdwidth = simdwidth,
-	    entrytype = entrytype,
+            entrytype = entrytype,
             lut = generate_lut(entries, simdwidth, simd_words)
         )
 
         all_luts += [lut]
 
     write_all_luts(filename, all_luts)
-
-
 
 def generate_AVX_luts(filename):
-
     lut_AVX_32x8 = template_lut.render(
-	lutname = f"vcompress_LUT32x8_AVX",
-	entries = 256,
-	simdwidth = 8,
-	entrytype = "int32_t",
-	lut = generate_lut(256, 8, 8)
+        lutname = f"vcompress_LUT32x8_AVX",
+        entries = 256,
+        simdwidth = 8,
+        entrytype = "int32_t",
+        lut = generate_lut(256, 8, 8)
     )
-    
+
     lut_AVX_64x4 = template_lut.render(
-	lutname = "vcompress_LUT64x4_AVX",
-	entries = 16,
-	simdwidth = 8,
-	entrytype = "int32_t",
-	lut = generate_lut(16, 8, 4)
+        lutname = "vcompress_LUT64x4_AVX",
+        entries = 16,
+        simdwidth = 8,
+        entrytype = "int32_t",
+        lut = generate_lut(16, 8, 4)
     )
 
     all_luts = [lut_AVX_32x8, lut_AVX_64x4]
 
     write_all_luts(filename, all_luts)
     pass
-
-generate_luts("src/mipp_SSE_LUT.cpp",  "SSE",  16, ("int8_t", 1),  [(4, 2), (16, 4), (256, 8), (65536, 16)])
-generate_luts("src/mipp_NEON_LUT.cpp", "NEON", 16, ("int8_t", 1),  [(4, 2), (16, 4), (256, 8), (65536, 16)])
 
-#generate_luts("src/mipp_AVX_LUT.cpp",  "AVX",  8, ("int32_t", 4), [(256, 8)])
-generate_AVX_luts("src/mipp_AVX_LUT.cpp")
+Path("src/gen").mkdir(parents=True, exist_ok=True)
+
+generate_luts("src/gen/compress_LUT_SSE.cpp",  "SSE",  16, ("int8_t", 1),  [(4, 2), (16, 4), (256, 8), (65536, 16)])
+generate_luts("src/gen/compress_LUT_NEON.cpp", "NEON", 16, ("int8_t", 1),  [(4, 2), (16, 4), (256, 8), (65536, 16)])
+
+#generate_luts("src/gen/mipp_compress_LUT_AVX.cpp",  "AVX",  8, ("int32_t", 4), [(256, 8)])
+generate_AVX_luts("src/gen/compress_LUT_AVX.cpp")