diff --git a/include/boost/multiprecision/detail/default_ops.hpp b/include/boost/multiprecision/detail/default_ops.hpp
index 4b7e7f9b..0fd06b26 100644
--- a/include/boost/multiprecision/detail/default_ops.hpp
+++ b/include/boost/multiprecision/detail/default_ops.hpp
@@ -725,10 +725,16 @@ inline void eval_round(T& result, const T& a)
    }
 }
 
+template <class B>
+void eval_lcm(B& result, const B& a, const B& b);
+template <class B>
+void eval_gcd(B& result, const B& a, const B& b);
+
 template <class T, class Arithmetic>
 inline typename enable_if<is_integral<Arithmetic> >::type eval_gcd(T& result, const T& a, const Arithmetic& b)
 {
    typedef typename boost::multiprecision::detail::canonical<Arithmetic, T>::type si_type;
+   using default_ops::eval_gcd;
    T t;
    t = static_cast<si_type>(b);
    eval_gcd(result, a, t);
@@ -742,6 +748,7 @@ template <class T, class Arithmetic>
 inline typename enable_if<is_integral<Arithmetic> >::type eval_lcm(T& result, const T& a, const Arithmetic& b)
 {
    typedef typename boost::multiprecision::detail::canonical<Arithmetic, T>::type si_type;
+   using default_ops::eval_lcm;
    T t;
    t = static_cast<si_type>(b);
    eval_lcm(result, a, t);
@@ -813,11 +820,6 @@ inline void eval_bit_unset(T& val, unsigned index)
       eval_bitwise_xor(val, mask);
 }
 
-template <class B>
-void eval_lcm(B& result, const B& a, const B& b);
-template <class B>
-void eval_gcd(B& result, const B& a, const B& b);
-
 //
 // These have to implemented by the backend, declared here so that our macro generated code compiles OK.
 //
diff --git a/include/boost/multiprecision/detail/generic_interconvert.hpp b/include/boost/multiprecision/detail/generic_interconvert.hpp
index 2cda6a6e..526ec86a 100644
--- a/include/boost/multiprecision/detail/generic_interconvert.hpp
+++ b/include/boost/multiprecision/detail/generic_interconvert.hpp
@@ -66,6 +66,7 @@ void generic_interconvert(To& to, const From& from, const mpl::int_<number_kind_
    using default_ops::eval_right_shift;
    using default_ops::eval_left_shift;
    using default_ops::eval_bitwise_or;
+   using default_ops::eval_is_zero;
    // smallest unsigned type handled natively by "From" is likely to be it's limb_type:
    typedef typename canonical<unsigned char, From>::type   limb_type;
    // get the corresponding type that we can assign to "To":
diff --git a/include/boost/multiprecision/detail/integer_ops.hpp b/include/boost/multiprecision/detail/integer_ops.hpp
index 7cd7fb78..9e53e094 100644
--- a/include/boost/multiprecision/detail/integer_ops.hpp
+++ b/include/boost/multiprecision/detail/integer_ops.hpp
@@ -309,7 +309,7 @@ void eval_powm(Backend& result, const Backend& a, const Backend& p, Integer c)
 }
 
 template <class Backend, class Integer>
-void eval_powm(Backend& result, const Backend& a, Integer b, const Backend& c)
+typename enable_if<is_unsigned<Integer> >::type eval_powm(Backend& result, const Backend& a, Integer b, const Backend& c)
 {
    typedef typename canonical<unsigned char, Backend>::type ui_type;
    typedef typename canonical<Integer, Backend>::type i_type;
@@ -320,11 +320,6 @@ void eval_powm(Backend& result, const Backend& a, Integer b, const Backend& c)
    using default_ops::eval_modulus;
    using default_ops::eval_right_shift;
 
-   if(b < 0)
-   {
-      BOOST_THROW_EXCEPTION(std::runtime_error("powm requires a positive exponent."));
-   }
-
    Backend x, y(a);
    x = ui_type(1u);
    while(b > 0)
@@ -341,8 +336,18 @@ void eval_powm(Backend& result, const Backend& a, Integer b, const Backend& c)
    eval_modulus(result, x, c);
 }
 
+template <class Backend, class Integer>
+typename enable_if<is_signed<Integer> >::type eval_powm(Backend& result, const Backend& a, Integer b, const Backend& c)
+{
+   if(b < 0)
+   {
+      BOOST_THROW_EXCEPTION(std::runtime_error("powm requires a positive exponent."));
+   }
+   eval_powm(result, a, static_cast<typename make_unsigned<Integer>::type>(b), c);
+}
+
 template <class Backend, class Integer1, class Integer2>
-void eval_powm(Backend& result, const Backend& a, Integer1 b, Integer2 c)
+typename enable_if<is_unsigned<Integer1> >::type eval_powm(Backend& result, const Backend& a, Integer1 b, Integer2 c)
 {
    typedef typename canonical<unsigned char, Backend>::type ui_type;
    typedef typename canonical<Integer1, Backend>::type i1_type;
@@ -354,11 +359,6 @@ void eval_powm(Backend& result, const Backend& a, Integer1 b, Integer2 c)
    using default_ops::eval_modulus;
    using default_ops::eval_right_shift;
 
-   if(b < 0)
-   {
-      BOOST_THROW_EXCEPTION(std::runtime_error("powm requires a positive exponent."));
-   }
-
    Backend x, y(a);
    x = ui_type(1u);
    while(b > 0)
@@ -375,6 +375,16 @@ void eval_powm(Backend& result, const Backend& a, Integer1 b, Integer2 c)
    eval_modulus(result, x, static_cast<i2_type>(c));
 }
 
+template <class Backend, class Integer1, class Integer2>
+typename enable_if<is_signed<Integer1> >::type eval_powm(Backend& result, const Backend& a, Integer1 b, Integer2 c)
+{
+   if(b < 0)
+   {
+      BOOST_THROW_EXCEPTION(std::runtime_error("powm requires a positive exponent."));
+   }
+   eval_powm(result, a, static_cast<typename make_unsigned<Integer1>::type>(b), c);
+}
+
 struct powm_func
 {
    template <class T, class U, class V>