Programs can define ASIO_ENABLE_SEQUENTIAL_STRAND_ALLOCATION to switch
the allocation of strand implementations to use a round-robin approach
rather than hashing.

asio/include/asio/detail/impl/strand_service.ipp

diff --git a/asio/include/asio/detail/impl/strand_service.ipp b/asio/include/asio/detail/impl/strand_service.ipp
index 29aade4..4d05ad5 100644
--- a/asio/include/asio/detail/impl/strand_service.ipp
+++ b/asio/include/asio/detail/impl/strand_service.ipp
@@ -61,14 +61,18 @@
 
 void strand_service::construct(strand_service::implementation_type& impl)
 {
+  asio::detail::mutex::scoped_lock lock(mutex_);
+
   std::size_t salt = salt_++;
+#if defined(ASIO_ENABLE_SEQUENTIAL_STRAND_ALLOCATION)
+  std::size_t index = salt;
+#else // defined(ASIO_ENABLE_SEQUENTIAL_STRAND_ALLOCATION)
   std::size_t index = reinterpret_cast<std::size_t>(&impl);
   index += (reinterpret_cast<std::size_t>(&impl) >> 3);
   index ^= salt + 0x9e3779b9 + (index << 6) + (index >> 2);
+#endif // defined(ASIO_ENABLE_SEQUENTIAL_STRAND_ALLOCATION)
   index = index % num_implementations;
 
-  asio::detail::mutex::scoped_lock lock(mutex_);
-
   if (!implementations_[index].get())
     implementations_[index].reset(new strand_impl);
   impl = implementations_[index].get();