部署环境：

平台：Windows 10

Python：3.10.13

Torch：1.12.1

Cuda：11.7

通过nvcc编译render_utils_kernel.cu的时候会返回以下错误信息

render_utils_kernel.cu(441): error: calling a __host__ function("std::conditional< ::std::is_same_v|| ::std::is_same_v, long double,    ::std::conditional< ::std::is_same_v&& ::std::is_same_v, float, double> ::type> ::type  ::pow<double, float, (int)0> (T1, T2)") from a __global__ function("raw2alpha_cuda_kernel<double> ") is not allowed

render_utils_kernel.cu(441): error: identifier "pow<double, float, (int)0> " is undefined in device code

render_utils_kernel.cu(515): error: calling a __host__ function("std::conditional< ::std::is_same_v|| ::std::is_same_v, long double,    ::std::conditional< ::std::is_same_v&& ::std::is_same_v, float, double> ::type> ::type  ::pow<double, float, (int)0> (T1, T2)") from a __global__ function("raw2alpha_backward_cuda_kernel<double> ") is not allowed

render_utils_kernel.cu(515): error: identifier "pow<double, float, (int)0> " is undefined in device code

4 errors detected in the compilation of "render_utils_kernel.cu".

 根据错误提示，修改render_utils_kernel.cu的441行和515行

# 原始代码
alpha[i_pt] = 1 - pow((1 + e), (-interval));# 修改为
alpha[i_pt] = 1 - pow((double)(1 + e), (double)(-interval));

 

# 原始代码
grad[i_pt] = min(exp_d[i_pt], 1e10) * pow((1+exp_d[i_pt]), (-interval-1)) * interval * grad_back[i_pt];# 修改为grad[i_pt] = min(exp_d[i_pt], 1e10) * pow((double)(1+exp_d[i_pt]), (double)(-interval-1)) * interval * grad_back[i_pt];

修改完这两行代码后，编译就可以顺利通过了