[BUG] print tensor<half> error

[luliyucoordinate]

An error occurred when I printed a tensor of type half using the program below.

#include <cute/tensor.hpp>
using namespace cute;

void test_print_half() {
  auto x = make_tensor<half>(Int<1024>{});
  print_tensor(x);
}

/mnt/d/cuda/cutlass/include/cute/util/print.hpp(258): error: more than one instance of overloaded function "cute::print" matches the argument list:
            function "cute::print(char)" (declared at line 66)
            function "cute::print(signed char)" (declared at line 72)
            function "cute::print(unsigned char)" (declared at line 78)
            function "cute::print(short)" (declared at line 84)
            function "cute::print(unsigned short)" (declared at line 90)
            function "cute::print(int)" (declared at line 96)
            function "cute::print(cutlass::uint1b_t)" (declared at line 102)
            function "cute::print(cutlass::int2b_t)" (declared at line 108)
            function "cute::print(cutlass::uint2b_t)" (declared at line 114)
            function "cute::print(cutlass::int4b_t)" (declared at line 120)
            function "cute::print(cutlass::uint4b_t)" (declared at line 126)
            function "cute::print(cutlass::bin1_t)" (declared at line 132)
            function "cute::print(unsigned int)" (declared at line 138)
            function "cute::print(long)" (declared at line 144)
            function "cute::print(unsigned long)" (declared at line 150)
            function "cute::print(long long)" (declared at line 156)
            function "cute::print(unsigned long long)" (declared at line 162)
            function "cute::print(float)" (declared at line 168)
            argument types are: (half)
    printf("  "); print(t);
                  ^
/mnt/d/cuda/cutlass/include/cute/util/print.hpp(187): note #3326-D: function "cute::print(const char *)" does not match because argument #1 does not match parameter
  print(char const* format) {
  ^
/mnt/d/cuda/cutlass/include/cute/util/print.hpp(181): note #3326-D: function template "cute::print(const char *, const T &...)" does not match because argument #1 does not match parameter
  print(char const* format, T const&... t) {
  ^
/mnt/d/cuda/cutlass/include/cute/numeric/numeric_types.hpp(106): note #3326-D: function "cute::print(cutlass::float_e5m2_t)" does not match because argument #1 does not match parameter
  print(float_e5m2_t a) {
  ^
/mnt/d/cuda/cutlass/include/cute/numeric/numeric_types.hpp(100): note #3326-D: function "cute::print(cutlass::float_e4m3_t)" does not match because argument #1 does not match parameter
  print(float_e4m3_t a) {
  ^
/mnt/d/cuda/cutlass/include/cute/numeric/numeric_types.hpp(94): note #3326-D: function "cute::print(cutlass::tfloat32_t)" does not match because argument #1 does not match parameter
  print(tfloat32_t a) {
  ^
/mnt/d/cuda/cutlass/include/cute/numeric/numeric_types.hpp(87): note #3326-D: function "cute::print(cutlass::bfloat16_t)" does not match because argument #1 does not match parameter
  print(bfloat16_t a) {
  ^
/mnt/d/cuda/cutlass/include/cute/numeric/numeric_types.hpp(81): note #3326-D: function "cute::print(cutlass::half_t)" does not match because argument #1 does not match parameter
  print(half_t a) {
  ^
          detected during:
            instantiation of "void cute::pretty_print(T) [with T=half]" at line 1112 of /mnt/d/cuda/cutlass/include/cute/tensor_impl.hpp
            instantiation of "void cute::print_tensor(const cute::Tensor<Engine, Layout> &, __nv_bool) [with Engine=cute::ArrayEngine<half, 1024UL>, Layout=cute::Layout<cute::C<1024>, cute::_1>]" at line 6 of /mnt/d/cuda/mykernels/test_cute/test_print_half.cu

1 error detected in the compilation of "/mnt/d/cuda/mykernels/test_cute/test_print_half.cu".
ninja: build stopped: subcommand failed.

[ccecka]

Should probably be using half_t rather than half for functionality and compatibility.

[luliyucoordinate]

ok