Fix CUDA version info in NVML application clock warning

Report driver and runtime versions, not the compute capability
of the card (since we already detected the card is recent enough).

Fixes #1706.

Change-Id: Ida8b36203f56dcda20dc367c69dc02586edf40e4

diff --git a/src/gromacs/gmxlib/gpu_utils/gpu_utils.cu b/src/gromacs/gmxlib/gpu_utils/gpu_utils.cu
index 3d93445f8d7e4762ea4467215ff1498d7a6bfe28..5614f673e121856bf1f4a947d44b0f39eeb27836 100644 (file)
--- a/src/gromacs/gmxlib/gpu_utils/gpu_utils.cu
+++ b/src/gromacs/gmxlib/gpu_utils/gpu_utils.cu
@@ -298,9 +298,14 @@ static gmx_bool init_gpu_application_clocks(FILE gmx_unused *fplog, int gmx_unus
 #ifndef HAVE_NVML
     if (bGpuCanUseApplicationClocks)
     {
-        md_print_warn( fplog, "NVML support was not found in CUDA library %d.%d, so your GPU of type %s cannot use application clock support to improve performance.\n",
-                       prop->major,
-                       prop->minor,
+        int cuda_driver  = 0;
+        int cuda_runtime = 0;
+        cudaDriverGetVersion(&cuda_driver);
+        cudaRuntimeGetVersion(&cuda_runtime);
+        md_print_warn( fplog, "Note: NVML support was not found (CUDA runtime %d.%d, driver %d.%d), so your\n"
+                       "      %s GPU cannot use application clock support to improve performance.\n",
+                       cuda_runtime/1000, cuda_runtime%100,
+                       cuda_driver/1000, cuda_driver%100,
                        prop->name );
     }
     return true;