GH-45331: [C++] Use xsimd for CPU feature detection (#49940)

### Rationale for this change
Use xsimd cpu feature instead of maitaining them here.
Stacked on GH-49922.
Should also handle issue from GH-30368.

### What changes are included in this PR?
- Use xsimd for CPu feature detection, cache sizes are still using the same implementation
- We are loosing CPU model name (which was unused)
- Simplify CpuInfo (remove Pimpl pattern)

### Are these changes tested?
in CI

### Are there any user-facing changes?
No

* GitHub Issue: #45331

Authored-by: AntoinePrv <AntoinePrv@users.noreply.github.com>
Signed-off-by: Antoine Pitrou <antoine@python.org>

Author: AntoinePrv

cpp/cmake_modules/ThirdpartyToolchain.cmake

@@ -2844,28 +2844,20 @@ macro(build_xsimd)
   set(XSIMD_VENDORED TRUE)
 endmacro()
 
-if((NOT ARROW_SIMD_LEVEL STREQUAL "NONE") OR (NOT ARROW_RUNTIME_SIMD_LEVEL STREQUAL "NONE"
-                                             ))
-  set(ARROW_USE_XSIMD TRUE)
+# Xsimd is mandatory as its CPU feature detection is the basis for Arrow CpuInfo
+resolve_dependency(xsimd
+                   FORCE_ANY_NEWER_VERSION
+                   TRUE
+                   IS_RUNTIME_DEPENDENCY
+                   FALSE
+                   REQUIRED_VERSION
+                   "14.2.0")
+
+if(xsimd_SOURCE STREQUAL "BUNDLED")
+  set(ARROW_XSIMD arrow::xsimd)
 else()
-  set(ARROW_USE_XSIMD FALSE)
-endif()
-
-if(ARROW_USE_XSIMD)
-  resolve_dependency(xsimd
-                     FORCE_ANY_NEWER_VERSION
-                     TRUE
-                     IS_RUNTIME_DEPENDENCY
-                     FALSE
-                     REQUIRED_VERSION
-                     "14.2.0")
-
-  if(xsimd_SOURCE STREQUAL "BUNDLED")
-    set(ARROW_XSIMD arrow::xsimd)
-  else()
-    message(STATUS "xsimd found. Headers: ${xsimd_INCLUDE_DIRS}")
-    set(ARROW_XSIMD xsimd)
-  endif()
+  message(STATUS "xsimd found. Headers: ${xsimd_INCLUDE_DIRS}")
+  set(ARROW_XSIMD xsimd)
 endif()
 
 macro(build_zlib)

cpp/src/arrow/CMakeLists.txt

@@ -601,6 +601,9 @@ arrow_add_object_library(ARROW_UTIL ${ARROW_UTIL_SRCS})
 foreach(ARROW_UTIL_TARGET ${ARROW_UTIL_TARGETS})
   target_compile_definitions(${ARROW_UTIL_TARGET} PRIVATE URI_STATIC_BUILD)
 endforeach()
+foreach(ARROW_UTIL_TARGET ${ARROW_UTIL_TARGETS})
+  target_link_libraries(${ARROW_UTIL_TARGET} PRIVATE ${ARROW_XSIMD})
+endforeach()
 
 if(ARROW_USE_BOOST)
   foreach(ARROW_UTIL_TARGET ${ARROW_UTIL_TARGETS})
@@ -617,11 +620,6 @@ if(ARROW_USE_OPENSSL)
     target_link_libraries(${ARROW_UTIL_TARGET} PRIVATE ${ARROW_OPENSSL_LIBS})
   endforeach()
 endif()
-if(ARROW_USE_XSIMD)
-  foreach(ARROW_UTIL_TARGET ${ARROW_UTIL_TARGETS})
-    target_link_libraries(${ARROW_UTIL_TARGET} PRIVATE ${ARROW_XSIMD})
-  endforeach()
-endif()
 if(ARROW_WITH_BROTLI)
   foreach(ARROW_UTIL_TARGET ${ARROW_UTIL_TARGETS})
     target_link_libraries(${ARROW_UTIL_TARGET} PRIVATE ${ARROW_BROTLI_LIBS})
@@ -734,11 +732,9 @@ if(ARROW_CSV)
                            csv/parser.cc
                            csv/reader.cc
                            csv/writer.cc)
-  if(ARROW_USE_XSIMD)
-    foreach(ARROW_CSV_TARGET ${ARROW_CSV_TARGETS})
-      target_link_libraries(${ARROW_CSV_TARGET} PRIVATE ${ARROW_XSIMD})
-    endforeach()
-  endif()
+  foreach(ARROW_CSV_TARGET ${ARROW_CSV_TARGETS})
+    target_link_libraries(${ARROW_CSV_TARGET} PRIVATE ${ARROW_XSIMD})
+  endforeach()
 
   list(APPEND ARROW_TESTING_SRCS csv/test_common.cc)
 else()
@@ -855,15 +851,13 @@ if(ARROW_COMPUTE)
   list(APPEND ARROW_COMPUTE_SHARED_INSTALL_INTERFACE_LIBS Arrow::arrow_shared)
   list(APPEND ARROW_COMPUTE_STATIC_LINK_LIBS arrow_static)
   list(APPEND ARROW_COMPUTE_SHARED_LINK_LIBS arrow_shared)
+  list(APPEND ARROW_COMPUTE_STATIC_LINK_LIBS ${ARROW_XSIMD})
+  list(APPEND ARROW_COMPUTE_SHARED_PRIVATE_LINK_LIBS ${ARROW_XSIMD})
 
   if(ARROW_USE_BOOST)
     list(APPEND ARROW_COMPUTE_STATIC_LINK_LIBS Boost::headers)
     list(APPEND ARROW_COMPUTE_SHARED_PRIVATE_LINK_LIBS Boost::headers)
   endif()
-  if(ARROW_USE_XSIMD)
-    list(APPEND ARROW_COMPUTE_STATIC_LINK_LIBS ${ARROW_XSIMD})
-    list(APPEND ARROW_COMPUTE_SHARED_PRIVATE_LINK_LIBS ${ARROW_XSIMD})
-  endif()
   if(ARROW_WITH_OPENTELEMETRY)
     list(APPEND ARROW_COMPUTE_STATIC_LINK_LIBS ${ARROW_OPENTELEMETRY_LIBS})
     list(APPEND ARROW_COMPUTE_SHARED_PRIVATE_LINK_LIBS ${ARROW_OPENTELEMETRY_LIBS})
@@ -910,11 +904,9 @@ endif()
 
 arrow_add_object_library(ARROW_COMPUTE_CORE ${ARROW_COMPUTE_SRCS})
 
-if(ARROW_USE_XSIMD)
-  foreach(ARROW_COMPUTE_CORE_TARGET ${ARROW_COMPUTE_CORE_TARGETS})
-    target_link_libraries(${ARROW_COMPUTE_CORE_TARGET} PRIVATE ${ARROW_XSIMD})
-  endforeach()
-endif()
+foreach(ARROW_COMPUTE_CORE_TARGET ${ARROW_COMPUTE_CORE_TARGETS})
+  target_link_libraries(${ARROW_COMPUTE_CORE_TARGET} PRIVATE ${ARROW_XSIMD})
+endforeach()
 if(ARROW_WITH_OPENTELEMETRY)
   foreach(ARROW_COMPUTE_CORE_TARGET ${ARROW_COMPUTE_CORE_TARGETS})
     target_link_libraries(${ARROW_COMPUTE_CORE_TARGET}

cpp/src/arrow/util/CMakeLists.txt

@@ -39,9 +39,7 @@ else()
 endif()
 
 set(ARROW_UTILITY_TEST_LINK_LIBS Boost::headers)
-if(ARROW_USE_XSIMD)
-  list(APPEND ARROW_UTILITY_TEST_LINK_LIBS ${ARROW_XSIMD})
-endif()
+list(APPEND ARROW_UTILITY_TEST_LINK_LIBS ${ARROW_XSIMD})
 if(ARROW_WITH_OPENTELEMETRY)
   list(APPEND ARROW_UTILITY_TEST_LINK_LIBS ${ARROW_OPENTELEMETRY_LIBS})
 endif()
@@ -135,9 +133,7 @@ add_arrow_benchmark(tdigest_benchmark)
 add_arrow_benchmark(thread_pool_benchmark)
 add_arrow_benchmark(trie_benchmark)
 set(ARROW_BENCHMARK_UTF8_UTIL_LINK_LIBS)
-if(ARROW_USE_XSIMD)
-  list(APPEND ARROW_BENCHMARK_UTF8_UTIL_LINK_LIBS ${ARROW_XSIMD})
-endif()
+list(APPEND ARROW_BENCHMARK_UTF8_UTIL_LINK_LIBS ${ARROW_XSIMD})
 add_arrow_benchmark(utf8_util_benchmark EXTRA_LINK_LIBS
                     ${ARROW_BENCHMARK_UTF8_UTIL_LINK_LIBS})
 add_arrow_benchmark(value_parsing_benchmark)