diff --git a/libcxx/cmake/caches/Apple.cmake b/libcxx/cmake/caches/Apple.cmake
--- a/libcxx/cmake/caches/Apple.cmake
+++ b/libcxx/cmake/caches/Apple.cmake
@@ -7,7 +7,7 @@
 set(LIBCXX_ENABLE_SHARED ON CACHE BOOL "")
 set(LIBCXX_CXX_ABI libcxxabi CACHE STRING "")
 set(LIBCXX_ENABLE_VENDOR_AVAILABILITY_ANNOTATIONS ON CACHE BOOL "")
-set(LIBCXX_PSTL_CPU_BACKEND libdispatch)
+set(LIBCXX_PSTL_CPU_BACKEND libdispatch CACHE STRING "")
 
 set(LIBCXX_HERMETIC_STATIC_LIBRARY ON CACHE BOOL "")
 set(LIBCXXABI_HERMETIC_STATIC_LIBRARY ON CACHE BOOL "")
diff --git a/libcxx/include/__algorithm/pstl_backends/cpu_backends/libdispatch.h b/libcxx/include/__algorithm/pstl_backends/cpu_backends/libdispatch.h
--- a/libcxx/include/__algorithm/pstl_backends/cpu_backends/libdispatch.h
+++ b/libcxx/include/__algorithm/pstl_backends/cpu_backends/libdispatch.h
@@ -10,6 +10,7 @@
 #define _LIBCPP___ALGORITHM_PSTL_BACKENDS_CPU_BACKENDS_LIBDISPATCH_H
 
 #include <__algorithm/lower_bound.h>
+#include <__algorithm/max.h>
 #include <__algorithm/upper_bound.h>
 #include <__atomic/atomic.h>
 #include <__config>
@@ -25,7 +26,9 @@
 #include <__utility/terminate_on_exception.h>
 #include <cstddef>
 #include <new>
-#include <vector>
+
+_LIBCPP_PUSH_MACROS
+#include <__undef_macros>
 
 #if !defined(_LIBCPP_HAS_NO_INCOMPLETE_PSTL) && _LIBCPP_STD_VER >= 17
 
@@ -107,13 +110,20 @@
   }
 
   using __merge_range_t = __merge_range<_RandomAccessIterator1, _RandomAccessIterator2, _RandomAccessIterator3>;
+  auto const __n_ranges = __partitions.__chunk_count_ + 1;
 
-  vector<__merge_range_t> __ranges;
-  __ranges.reserve(__partitions.__chunk_count_ + 1);
+  // TODO: use __uninitialized_buffer
+  auto __destroy = [=](__merge_range_t* __ptr) {
+    std::destroy_n(__ptr, __n_ranges);
+    std::allocator<__merge_range_t>().deallocate(__ptr, __n_ranges);
+  };
+  unique_ptr<__merge_range_t[], decltype(__destroy)> __ranges(
+    std::allocator<__merge_range_t>().allocate(__n_ranges), __destroy);
 
   // TODO: Improve the case where the smaller range is merged into just a few (or even one) chunks of the larger case
   std::__terminate_on_exception([&] {
-    __ranges.emplace_back(__first1, __first2, __result);
+    __merge_range_t* __r = __ranges.get();
+    std::construct_at(__r++, __first1, __first2, __result);
 
     bool __iterate_first_range = __last1 - __first1 > __last2 - __first2;
 
@@ -137,14 +147,14 @@
     };
 
     // handle first chunk
-    __ranges.emplace_back(__compute_chunk(__partitions.__first_chunk_size_));
+    std::construct_at(__r++, __compute_chunk(__partitions.__first_chunk_size_));
 
     // handle 2 -> N - 1 chunks
     for (ptrdiff_t __i = 0; __i != __partitions.__chunk_count_ - 2; ++__i)
-      __ranges.emplace_back(__compute_chunk(__partitions.__chunk_size_));
+      std::construct_at(__r++, __compute_chunk(__partitions.__chunk_size_));
 
     // handle last chunk
-    __ranges.emplace_back(__last1, __last2, __result);
+    std::construct_at(__r, __last1, __last2, __result);
 
     __libdispatch::__dispatch_apply(__partitions.__chunk_count_, [&](size_t __index) {
       auto __first_iters = __ranges[__index];
@@ -223,4 +233,6 @@
 
 #endif // !defined(_LIBCPP_HAS_NO_INCOMPLETE_PSTL) && _LIBCPP_STD_VER >= 17
 
+_LIBCPP_POP_MACROS
+
 #endif // _LIBCPP___ALGORITHM_PSTL_BACKENDS_CPU_BACKENDS_LIBDISPATCH_H
diff --git a/libcxx/test/libcxx/vendor/apple/system-install-properties.sh.cpp b/libcxx/test/libcxx/vendor/apple/system-install-properties.sh.cpp
--- a/libcxx/test/libcxx/vendor/apple/system-install-properties.sh.cpp
+++ b/libcxx/test/libcxx/vendor/apple/system-install-properties.sh.cpp
@@ -42,3 +42,7 @@
 // when they are loaded by dyld, if the compatibility version was bumped.
 //
 // RUN: otool -L "%{lib}/libc++.1.dylib" | grep "libc++.1.dylib" | grep "compatibility version 1.0.0"
+
+// Make sure we use the libdispatch backend for the PSTL.
+//
+// RUN: grep "%{include}/__config_site" -e '#define _LIBCPP_PSTL_CPU_BACKEND_LIBDISPATCH'