diff --git a/profiling/simple-kernel-timer/kp_shared.cpp b/profiling/simple-kernel-timer/kp_shared.cpp
index 43f306d81..7b7928046 100644
--- a/profiling/simple-kernel-timer/kp_shared.cpp
+++ b/profiling/simple-kernel-timer/kp_shared.cpp
@@ -27,5 +27,38 @@ char* outputDelimiter;
 int current_region_level = 0;
 KernelPerformanceInfo* regions[512];
 
+void increment_counter(const char* name, KernelExecutionType kType) {
+  std::string nameStr(name);
+
+  if (count_map.find(name) == count_map.end()) {
+    KernelPerformanceInfo* info = new KernelPerformanceInfo(nameStr, kType);
+    count_map.insert(
+        std::pair<std::string, KernelPerformanceInfo*>(nameStr, info));
+
+    currentEntry = info;
+  } else {
+    currentEntry = count_map[nameStr];
+  }
+
+  currentEntry->startTimer();
+}
+
+void increment_counter_region(const char* name, KernelExecutionType kType) {
+  std::string nameStr(name);
+
+  if (count_map.find(name) == count_map.end()) {
+    KernelPerformanceInfo* info = new KernelPerformanceInfo(nameStr, kType);
+    count_map.insert(
+        std::pair<std::string, KernelPerformanceInfo*>(nameStr, info));
+
+    regions[current_region_level] = info;
+  } else {
+    regions[current_region_level] = count_map[nameStr];
+  }
+
+  regions[current_region_level]->startTimer();
+  current_region_level++;
+}
+
 }  // namespace KernelTimer
 }  // namespace KokkosTools
diff --git a/profiling/simple-kernel-timer/kp_shared.h b/profiling/simple-kernel-timer/kp_shared.h
index 990fee2a7..f719f5b20 100644
--- a/profiling/simple-kernel-timer/kp_shared.h
+++ b/profiling/simple-kernel-timer/kp_shared.h
@@ -31,39 +31,8 @@ extern char* outputDelimiter;
 extern int current_region_level;
 extern KernelPerformanceInfo* regions[512];
 
-inline void increment_counter(const char* name, KernelExecutionType kType) {
-  std::string nameStr(name);
-
-  if (count_map.find(name) == count_map.end()) {
-    KernelPerformanceInfo* info = new KernelPerformanceInfo(nameStr, kType);
-    count_map.insert(
-        std::pair<std::string, KernelPerformanceInfo*>(nameStr, info));
-
-    currentEntry = info;
-  } else {
-    currentEntry = count_map[nameStr];
-  }
-
-  currentEntry->startTimer();
-}
-
-inline void increment_counter_region(const char* name,
-                                     KernelExecutionType kType) {
-  std::string nameStr(name);
-
-  if (count_map.find(name) == count_map.end()) {
-    KernelPerformanceInfo* info = new KernelPerformanceInfo(nameStr, kType);
-    count_map.insert(
-        std::pair<std::string, KernelPerformanceInfo*>(nameStr, info));
-
-    regions[current_region_level] = info;
-  } else {
-    regions[current_region_level] = count_map[nameStr];
-  }
-
-  regions[current_region_level]->startTimer();
-  current_region_level++;
-}
+void increment_counter(const char* name, KernelExecutionType kType);
+void increment_counter_region(const char* name, KernelExecutionType kType);
 
 inline bool compareKernelPerformanceInfo(KernelPerformanceInfo* left,
                                          KernelPerformanceInfo* right) {