32: Set AVX and AVX2 flags using CMake checks

[Dantali0n]

Closes: 32

This add the checks to enable avx and avx2

[Dantali0n]

It would be nice if could run the linting checks locally easily. Especially given they fail on files I haven't touched.

[csbnw]

You can simply run pre-commit run -a in the top-level directory of the repository to run the linters. I am not sure what happened to that file, but this PR fixes it. I suggest to merge it first, then rebase your branch onto the new main.

[Dantali0n]

Seem icpx compiler doesn't understand CORE-AVX:

make[2]: *** Waiting for unfinished jobs....
icpx: error: language not recognized: 'CORE-AVX'
icpx: error: language not recognized: 'CORE-AVX'

[Dantali0n]

icpx compiler wants flag -mavx for AVX not axCORE-AVX that is only for AVX2.

[csbnw]

According to this documentation, it should be AVX for AVX and CORE-AVX2 for AVX2.
The Intel compiler doesn't like to have both flags, so i would suggest using:

diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index a152ef5..97ea8e1 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -2,11 +2,6 @@ include(FetchContent)
 include(FindAVX)
 add_library(trigdx reference.cpp lookup.cpp)
 
-if(HAVE_AVX)
-  target_compile_definitions(trigdx PUBLIC HAVE_AVX)
-  target_compile_options(trigdx PUBLIC -mavx)
-endif()
-
 if(HAVE_AVX2)
   target_compile_definitions(trigdx PUBLIC HAVE_AVX2)
   if(CMAKE_CXX_COMPILER_ID STREQUAL "Intel" OR CMAKE_CXX_COMPILER_ID STREQUAL
@@ -15,6 +10,14 @@ if(HAVE_AVX2)
   else()
     target_compile_options(trigdx PUBLIC -mavx2)
   endif()
+elseif(HAVE_AVX)
+  target_compile_definitions(trigdx PUBLIC HAVE_AVX)
+  if(CMAKE_CXX_COMPILER_ID STREQUAL "Intel" OR CMAKE_CXX_COMPILER_ID STREQUAL
+                                               "IntelLLVM")
+    target_compile_options(trigdx PUBLIC -xAVX)
+  else()
+    target_compile_options(trigdx PUBLIC -mavx)
+  endif()
 endif()
 
 target_include_directories(trigdx PUBLIC ${PROJECT_SOURCE_DIR}/include)