diff --git a/configure.ac b/configure.ac
index e39ea39d..4b83d799 100644
--- a/configure.ac
+++ b/configure.ac
@@ -222,13 +222,46 @@ AS_IF([test "x$with_cuda" != xno],
         AS_IF([test "x$with_cuda" = xyes],
             [AS_IF([test "x$CUDA_CPPFLAGS" = x],
                 [dnl generate CUDA code for broad spectrum of devices
-                 dnl Note: cc 13 for Tesla
-                 dnl Note: cc 20 for Fermi
-                 dnl Note: cc 30 for Kepler
-                 CUDA_CPPFLAGS=["-gencode=arch=compute_12,code=sm_12 \
- -gencode=arch=compute_13,code=sm_13 \
- -gencode=arch=compute_20,code=sm_20 \
- -gencode=arch=compute_20,code=compute_20"]
+
+		 nvcc_version=[`nvcc --version 2>&1 |sed -e '/release/s/.* \(@<:@0-9@:>@\+\.@<:@0-9@:>@\+\).*/\1/p' -n`]
+		 nvcc_major_version=[`nvcc --version 2>&1 |sed -e '/release/s/.*release \(@<:@0-9@:>@\+\)\.@<:@0-9@:>@\+.*/\1/p' -n`]
+		 nvcc_minor_version=[`nvcc --version 2>&1 |sed -e '/release/s/.*release @<:@0-9@:>@\+\.\(@<:@0-9@:>@\+\).*/\1/p' -n`]
+
+		 AC_MSG_NOTICE([detected CUDA release $nvcc_major_version.$nvcc_minor_version])
+
+                 dnl Note: cc 1x for Tesla
+                 dnl Note: cc 2x for Fermi
+                 dnl Note: cc 3x for Kepler
+		 dnl Note: cc 5x for Maxwell
+		 dnl Note: cc 6x for Pascal
+		 cuda_compute_1x=["-gencode arch=compute_10,code=sm_10 -gencode arch=compute_12,code=sm_12"]
+ 		 cuda_compute_2x=["-gencode arch=compute_20,code=sm_20"]
+		 cuda_compute_30=["-gencode arch=compute_30,code=sm_30 -gencode arch=compute_35,code=sm_35"]
+		 cuda_compute_37=["-gencode arch=compute_37,code=sm_37"]
+		 cuda_compute_3x=["$cuda_compute_30 $cuda_compute_37"]
+	 	 cuda_compute_50=["-gencode arch=compute_50,code=sm_50"]
+		 cuda_compute_5x=["-gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52"]
+		 cuda_compute_6x=["-gencode arch=compute_60,code=sm_60 -gencode arch=compute_61,code=sm_61 -gencode arch=compute_62,code=sm_62"]
+
+		 CUDA_GENCODE_FLAGS=[]		
+		AS_IF([test $nvcc_major_version -lt "6"],
+			    [CUDA_GENCODE_FLAGS="$cuda_compute_1x $cuda_compute_2x $cuda_compute_30"],
+			    []
+			    )
+		AS_IF([test $nvcc_major_version -ge "6"],
+			    [CUDA_GENCODE_FLAGS="$cuda_compute_2x $cuda_compute_3x $cuda_compute_50"],
+			    []
+			    )
+		AS_IF([test $nvcc_major_version -ge "7"],
+			    [CUDA_GENCODE_FLAGS="$cuda_compute_2x $cuda_compute_3x $cuda_compute_5x"],
+			    []
+			    )
+		AS_IF([test $nvcc_major_version -eq "8"],
+			    [CUDA_GENCODE_FLAGS="$cuda_compute_2x $cuda_compute_3x $cuda_compute_5x $cuda_compute_6x"],
+			    []
+			    )
+		CUDA_CPPFLAGS=$CUDA_GENCODE_FLAGS
+		
                  AC_MSG_NOTICE([using CUDA_CPPFLAGS=$CUDA_CPPFLAGS])
                  AC_MSG_WARN([*** SHOC programs may fail if these -gencode values do not apply to your GPU.  If they don't apply, or if you want to shorten SHOC's compile time, determine the -gencode options that apply for your GPU and re-configure SHOC explicitly specifying only those options in CUDA_CPPFLAGS during the configure. ***])
                 ]