diff --git a/src/flit/flitHelpers.h b/src/flit/flitHelpers.h
index 1fc01091..5e1f9e23 100644
--- a/src/flit/flitHelpers.h
+++ b/src/flit/flitHelpers.h
@@ -93,6 +93,7 @@
 #include <algorithm>
 #include <fstream>
 #include <iostream>
+#include <limits>
 #include <mutex>
 #include <ostream>
 #include <random>
@@ -102,6 +103,7 @@
 #include <vector>
 
 #include <cfloat>
+#include <cmath>
 
 #ifndef FLIT_UNUSED
 #define FLIT_UNUSED(x) (void)x
@@ -254,6 +256,24 @@ as_int(long double val) {
   return temp & (~zero >> 48);
 }
 
+template <typename T>
+bool equal_with_nan_inf(T a, T b) {
+  if (std::fpclassify(a) == std::fpclassify(b)) {
+    switch (std::fpclassify(a)) {
+      case FP_INFINITE:
+      case FP_NAN:
+        return std::signbit(a) == std::signbit(b);
+
+      case FP_NORMAL:
+      case FP_SUBNORMAL:
+      case FP_ZERO:
+      default:
+        return a == b;
+    }
+  }
+  return false;
+}
+
 /**
  * Default comparison used by FLiT.  Similar to
  *
@@ -266,7 +286,12 @@ as_int(long double val) {
  */
 template <typename T>
 T abs_compare(T expected, T actual) {
-  // TODO: implement all other cases
+  if (equal_with_nan_inf(expected, actual)) {
+    return T(0.0);
+  }
+  if (std::isnan(expected) && std::isinf(actual)) {
+    return std::numeric_limits<T>::infinity();
+  }
   return std::abs(actual - expected);
 }
 
diff --git a/tests/flit_src/tst_flitHelpers_h.cpp b/tests/flit_src/tst_flitHelpers_h.cpp
index d90b0ee6..e9c0b964 100644
--- a/tests/flit_src/tst_flitHelpers_h.cpp
+++ b/tests/flit_src/tst_flitHelpers_h.cpp
@@ -280,24 +280,6 @@ TH_TEST(tst_as_int_128bit) {
 
 namespace tst_abs_compare {
 
-template <typename T>
-bool equal_with_nan_inf(T a, T b) {
-  if (std::fpclassify(a) == std::fpclassify(b)) {
-    switch (std::fpclassify(a)) {
-      case FP_INFINITE:
-      case FP_NAN:
-        return std::signbit(a) == std::signbit(b);
-
-      case FP_NORMAL:
-      case FP_SUBNORMAL:
-      case FP_ZERO:
-      default:
-        return a == b;
-    }
-  }
-  return false;
-}
-
 template <typename T>
 void tst_equal_with_nan_inf_impl() {
   using lim = std::numeric_limits<T>;
@@ -305,7 +287,7 @@ void tst_equal_with_nan_inf_impl() {
   static_assert(lim::has_quiet_NaN);
   static_assert(lim::has_infinity);
 
-  auto eq = equal_with_nan_inf<T>;
+  auto &eq = flit::equal_with_nan_inf<T>;
   T my_nan = lim::quiet_NaN();
   T my_inf = lim::infinity();
   T normal = -3.2;
@@ -370,8 +352,8 @@ void tst_abs_compare_impl() {
   T normal = -3.2;
   T zero = 0.0;
 
-  auto eq = equal_with_nan_inf<T>;
-  auto comp = flit::abs_compare<T>;
+  auto &eq = flit::equal_with_nan_inf<T>;
+  auto &comp = flit::abs_compare<T>;
 
   // we have 25 cases
   TH_VERIFY(eq(comp( my_nan,  my_nan),   zero ));