[test-suite] Add CUDA printf test.

[jlebar]

Summary

Checks that printf() actually does something in CUDA device code.

[tra]

External/CUDA/printf.cu

2

Why do we need <iostream> or any include at all?
Files included for CUDA compilation by default should provide all you need.

[jlebar]

Don't include <iostream>.

External/CUDA/CMakeLists.txt

macro(create_one_local_test Name FileGlob)
   endif()
   list(APPEND VARIANT_CUDA_TESTS ${_target})
 endmacro()
 
 # Create targets for CUDA tests that are part of the test suite.
 macro(create_local_cuda_tests VariantSuffix)
   create_one_local_test(axpy axpy.cu)
   create_one_local_test(empty empty.cu)
+  create_one_local_test(printf printf.cu)
   create_one_local_test(future future.cu)
 endmacro()
 
 macro(thrust_make_test_name TestName TestSourcePath)
   string(REPLACE "${THRUST_PATH}/testing/" "" _tmp ${TestSourcePath})
   string(REPLACE "/" "-" _tmp ${_tmp})
   string(REPLACE "<" "_" _tmp ${_tmp})
   string(REPLACE ">" "_" _tmp ${_tmp})

External/CUDA/printf.cu

+__global__ void kernel(char a, short b, int c, float d, double e) {
+  const char* f = "abcdefg";
+  printf("kernel_printf %c %d %d %f %f %s\n", a, b, c, d, e, f);
+}
+
+int main(int argc, char* argv[]) {
+  kernel<<<2, 8>>>('a', 2, 3, 4, 5);
+  cudaDeviceSynchronize();
+  cudaDeviceReset();
+  return 0;
+}

External/CUDA/printf.reference_output

+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+kernel_printf a 2 3 4.000000 5.000000 abcdefg
+exit 0