CMake error

[M3LLI55X]

when I run cmake --build build --config RelWithDebInfo -j , I got errors said identifier "parallel_for_gpu" is undefined

/dtu/blackhole/11/180913/NeuS2/include/neural-graphics-primitives/nerf_network.h(1206): error: identifier "parallel_for_gpu" is undefined
    parallel_for_gpu(n_params, [params=accumulated_rotation->params(), params_hp=params_hp.data()] __attribute__((device)) (size_t i) {
    ^
          detected during instantiation of "void ngp::NerfNetwork<T>::load_global_movement(tcnn::json) [with T=ngp::precision_t]" at line 3241 of /dtu/blackhole/11/180913/NeuS2/src/testbed.cu

/dtu/blackhole/11/180913/NeuS2/include/neural-graphics-primitives/nerf_network.h(1248): error: identifier "parallel_for_gpu" is undefined
    parallel_for_gpu(n_params, [params=m_delta_network->rotation()->params(), params_hp=params_hp.data()] __attribute__((device)) (size_t i) {
    ^
          detected during instantiation of "void ngp::NerfNetwork<T>::load_local_movement(tcnn::json) [with T=ngp::precision_t]" at line 3242 of /dtu/blackhole/11/180913/NeuS2/src/testbed.cu

/dtu/blackhole/11/180913/NeuS2/include/neural-graphics-primitives/nerf_network.h(1088): error: identifier "parallel_for_gpu" is undefined
    parallel_for_gpu(n_params, [params_fp=params_full_precision, params=params] __attribute__((device)) (size_t i) {
    ^
          detected during:
            instantiation of "void ngp::NerfNetwork<T>::init_accumulation_movement() [with T=ngp::precision_t]" at line 82
            instantiation of "ngp::NerfNetwork<T>::NerfNetwork(uint32_t, uint32_t, uint32_t, uint32_t, const ngp::NerfNetwork<T>::json &, const ngp::NerfNetwork<T>::json &, const ngp::NerfNetwork<T>::json &, const ngp::NerfNetwork<T>::json &) [with T=ngp::precision_t]" at line 119 of /appl9/gcc/12.3.0-binutils-2.40/include/c++/12.3.0/bits/stl_construct.h

my environment:

 gcc/12.3.0-binutils-2.40   
 cmake/3.28.3    
cuda/12.4

[M3LLI55X]

I changed the version of cuda to 12.1 to solve it