Corrected to avoid CUDA tests when cuda-enabled GPU is not present on Windows

platforms/cuda/CMakeLists.txt

@@ -26,10 +26,12 @@ try_run(RUN_RESULT_VAR COMPILE_RESULT_VAR
     COMPILE_OUTPUT_VARIABLE COMPILE_OUTPUT_VAR
     RUN_OUTPUT_VARIABLE RUN_OUTPUT_VAR
 )
+message("COMPILE_OUTPUT_VAR = ${COMPILE_OUTPUT_VAR}")
+message("RUN_OUTPUT_VAR = ${RUN_OUTPUT_VAR}")
 message("${RUN_OUTPUT_VAR}") # Display number of GPUs found
 # COMPILE_RESULT_VAR is TRUE when compile succeeds
-# RUN_RESULT_VAR is equal to the number of GPUs found
-if(COMPILE_RESULT_VAR AND RUN_RESULT_VAR)
+# RUN_RESULT_VAR is zero when a GPU is found
+if(COMPILE_RESULT_VAR AND NOT RUN_RESULT_VAR)
     set(CUDA_HAVE_GPU TRUE CACHE BOOL "Whether CUDA-capable GPU is present")
 else()
     set(CUDA_HAVE_GPU FALSE CACHE BOOL "Whether CUDA-capable GPU is present")

platforms/cuda/tests/has_cuda_gpu.c

@@ -8,6 +8,7 @@ int main() {
     cudaError_t cudaResultCode = cudaGetDeviceCount(&deviceCount);
     if (cudaResultCode != cudaSuccess) 
         deviceCount = 0;
+    /* machines with no GPUs can still report one emulation device */
     for (device = 0; device < deviceCount; ++device) {
         cudaGetDeviceProperties(&properties, device);
         if (properties.major != 9999) /* 9999 means emulation only */
@@ -15,6 +16,11 @@ int main() {
     }
     printf("%d GPU CUDA device(s) found\n", gpuDeviceCount);
 
-    return gpuDeviceCount;
+    /* don't just return the number of gpus, because other weird
+       errors can yield non-zero return values */
+    if (gpuDeviceCount > 0)
+        return 0; /* success */
+    else
+        return 1; /* failure */
 }