From 4486847984f7cd2d65be87bdf124663bc8798111 Mon Sep 17 00:00:00 2001
From: Kevin Huck <khuck@cs.uoregon.edu>
Date: Mon, 12 Apr 2021 14:17:42 -0700
Subject: [PATCH] Removing APEX counters (llvm won't link them?)

---
 src/unit_tests/CUDA/apex_cuda.cu       | 4 ++--
 src/unit_tests/CUDA/apex_multi_cuda.cu | 2 +-
 src/unit_tests/CUDA/apex_vector.cu     | 2 +-
 src/unit_tests/CUDA/multiGpuThread.cu  | 2 +-
 4 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/src/unit_tests/CUDA/apex_cuda.cu b/src/unit_tests/CUDA/apex_cuda.cu
index 9d9eb1c3..330acb18 100644
--- a/src/unit_tests/CUDA/apex_cuda.cu
+++ b/src/unit_tests/CUDA/apex_cuda.cu
@@ -11,7 +11,7 @@
 /* For user instrumentation */
 #include "nvToolsExt.h"
 
-#define ITERATIONS 4
+#define ITERATIONS 1
 
 #define DRIVER_API_CALL(apiFuncCall)                                           \
 do {                                                                           \
@@ -89,7 +89,7 @@ void do_marker_payload(const char * name, int payload, nvtxDomainHandle_t domain
 }
 
 void launch(DataElement *elem) {
-  APEX_SCOPED_TIMER;
+  //APEX_SCOPED_TIMER;
   nvtxDomainHandle_t domain = nvtxDomainCreateA("apex.example.loop.domain");
   do_marker_payload(elem->name, elem->value, domain);
   Kernel<<< 1, 1 >>>(elem);
diff --git a/src/unit_tests/CUDA/apex_multi_cuda.cu b/src/unit_tests/CUDA/apex_multi_cuda.cu
index 8ced7b5f..991a0693 100644
--- a/src/unit_tests/CUDA/apex_multi_cuda.cu
+++ b/src/unit_tests/CUDA/apex_multi_cuda.cu
@@ -28,7 +28,7 @@ void Kernel(DataElement *elem) {
 }
 
 void launch(DataElement *elem) {
-  APEX_SCOPED_TIMER;
+  //APEX_SCOPED_TIMER;
   Kernel<<< 1, 1 >>>(elem);
   RUNTIME_API_CALL(cudaDeviceSynchronize());
 }
diff --git a/src/unit_tests/CUDA/apex_vector.cu b/src/unit_tests/CUDA/apex_vector.cu
index 5c5d0ea1..d43d1019 100644
--- a/src/unit_tests/CUDA/apex_vector.cu
+++ b/src/unit_tests/CUDA/apex_vector.cu
@@ -54,7 +54,7 @@ VecSub(const int* A, const int* B, int* C, int N)
 static void
 do_pass(cudaStream_t stream)
 {
-  APEX_SCOPED_TIMER;
+  //APEX_SCOPED_TIMER;
   int *h_A, *h_B, *h_C;
   int *d_A, *d_B, *d_C;
   size_t size = COMPUTE_N * sizeof(int);
diff --git a/src/unit_tests/CUDA/multiGpuThread.cu b/src/unit_tests/CUDA/multiGpuThread.cu
index c98f6baa..aca9f2df 100644
--- a/src/unit_tests/CUDA/multiGpuThread.cu
+++ b/src/unit_tests/CUDA/multiGpuThread.cu
@@ -36,7 +36,7 @@ __global__ void kernel_fc(int *dev_arr, int *dev_result)
 
 void *thread_func(void* struc)
 {
-    APEX_SCOPED_TIMER;
+    //APEX_SCOPED_TIMER;
     cuda_st * data = (cuda_st*)struc;
     printf("thread %d func start\n", data->thr_num);
     int i;