raise an error at RTX 5090

when I run pip install turbodiffusion --no-build-isolation , 报错了。 我的显卡是 RTX 4000，如何才能解决 nvcc fatal : Unsupported gpu architecture 'compute_120a' 这个报错。
[5/5] /usr/local/cuda-12.4/bin/nvcc --generate-dependencies-with-compile --dependency-output /tmp/pip-install-mdch8p6u/turbodiffusion_56f05382417943d5823fbee3ef0cf52e/build/temp.linux-x86_64-cpython-312/turbodiffusion/ops/quant/quant.o.d -I/tmp/pip-install-mdch8p6u/turbodiffusion_56f05382417943d5823fbee3ef0cf52e/turbodiffusion/ops/cutlass/include -I/tmp/pip-install-mdch8p6u/turbodiffusion_56f05382417943d5823fbee3ef0cf52e/turbodiffusion/ops/cutlass/tools/util/include -I/tmp/pip-install-mdch8p6u/turbodiffusion_56f05382417943d5823fbee3ef0cf52e/turbodiffusion/ops -I/home/tubor/lib/python3.12/site-packages/torch/include -I/home/tubor/lib/python3.12/site-packages/torch/include/torch/csrc/api/include -I/home/tubor/lib/python3.12/site-packages/torch/include/TH -I/home/tubor/lib/python3.12/site-packages/torch/include/THC -I/usr/local/cuda-12.4/include -I/home/tubor/include -I/usr/include/python3.12 -c -c /tmp/pip-install-mdch8p6u/turbodiffusion_56f05382417943d5823fbee3ef0cf52e/turbodiffusion/ops/quant/quant.cu -o /tmp/pip-install-mdch8p6u/turbodiffusion_56f05382417943d5823fbee3ef0cf52e/build/temp.linux-x86_64-cpython-312/turbodiffusion/ops/quant/quant.o -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_BFLOAT16_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr --compiler-options ''"'"'-fPIC'"'"'' -O3 -std=c++17 -U__CUDA_NO_HALF_OPERATORS__ -U__CUDA_NO_HALF_CONVERSIONS__ -U__CUDA_NO_BFLOAT16_OPERATORS__ -U__CUDA_NO_BFLOAT16_CONVERSIONS__ -U__CUDA_NO_BFLOAT162_OPERATORS__ -U__CUDA_NO_BFLOAT162_CONVERSIONS__ --expt-relaxed-constexpr --expt-extended-lambda --use_fast_math --ptxas-options=--verbose,--warn-on-local-memory-usage -lineinfo -DCUTLASS_DEBUG_TRACE_LEVEL=0 -DNDEBUG -Xcompiler -fPIC -DEXECMODE=0 -gencode arch=compute_120a,code=sm_120a -gencode arch=compute_90,code=sm_90 -gencode arch=compute_89,code=sm_89 -gencode arch=compute_80,code=sm_80 --threads 4 -DTORCH_API_INCLUDE_EXTENSION_H '-DPYBIND11_COMPILER_TYPE="gcc"' '-DPYBIND11_STDLIB="libstdcpp"' '-DPYBIND11_BUILD_ABI="cxxabi1011"' -DTORCH_EXTENSION_NAME=turbo_diffusion_ops -D_GLIBCXX_USE_CXX11_ABI=0
FAILED: [code=1] /tmp/pip-install-mdch8p6u/turbodiffusion_56f05382417943d5823fbee3ef0cf52e/build/temp.linux-x86_64-cpython-312/turbodiffusion/ops/quant/quant.o
/usr/local/cuda-12.4/bin/nvcc --generate-dependencies-with-compile --dependency-output /tmp/pip-install-mdch8p6u/turbodiffusion_56f05382417943d5823fbee3ef0cf52e/build/temp.linux-x86_64-cpython-312/turbodiffusion/ops/quant/quant.o.d -I/tmp/pip-install-mdch8p6u/turbodiffusion_56f05382417943d5823fbee3ef0cf52e/turbodiffusion/ops/cutlass/include -I/tmp/pip-install-mdch8p6u/turbodiffusion_56f05382417943d5823fbee3ef0cf52e/turbodiffusion/ops/cutlass/tools/util/include -I/tmp/pip-install-mdch8p6u/turbodiffusion_56f05382417943d5823fbee3ef0cf52e/turbodiffusion/ops -I/home/tubor/lib/python3.12/site-packages/torch/include -I/home/tubor/lib/python3.12/site-packages/torch/include/torch/csrc/api/include -I/home/tubor/lib/python3.12/site-packages/torch/include/TH -I/home/tubor/lib/python3.12/site-packages/torch/include/THC -I/usr/local/cuda-12.4/include -I/home/tubor/include -I/usr/include/python3.12 -c -c /tmp/pip-install-mdch8p6u/turbodiffusion_56f05382417943d5823fbee3ef0cf52e/turbodiffusion/ops/quant/quant.cu -o /tmp/pip-install-mdch8p6u/turbodiffusion_56f05382417943d5823fbee3ef0cf52e/build/temp.linux-x86_64-cpython-312/turbodiffusion/ops/quant/quant.o -D__CUDA_NO_HALF_OPERATORS -D__CUDA_NO_HALF_CONVERSIONS_ -D__CUDA_NO_BFLOAT16_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr --compiler-options ''"'"'-fPIC'"'"'' -O3 -std=c++17 -U__CUDA_NO_HALF_OPERATORS__ -U__CUDA_NO_HALF_CONVERSIONS__ -U__CUDA_NO_BFLOAT16_OPERATORS__ -U__CUDA_NO_BFLOAT16_CONVERSIONS__ -U__CUDA_NO_BFLOAT162_OPERATORS__ -U__CUDA_NO_BFLOAT162_CONVERSIONS__ --expt-relaxed-constexpr --expt-extended-lambda --use_fast_math --ptxas-options=--verbose,--warn-on-local-memory-usage -lineinfo -DCUTLASS_DEBUG_TRACE_LEVEL=0 -DNDEBUG -Xcompiler -fPIC -DEXECMODE=0 -gencode arch=compute_120a,code=sm_120a -gencode arch=compute_90,code=sm_90 -gencode arch=compute_89,code=sm_89 -gencode arch=compute_80,code=sm_80 --threads 4 -DTORCH_API_INCLUDE_EXTENSION_H '-DPYBIND11_COMPILER_TYPE="_gcc"' '-DPYBIND11_STDLIB="_libstdcpp"' '-DPYBIND11_BUILD_ABI="_cxxabi1011"' -DTORCH_EXTENSION_NAME=turbo_diffusion_ops -D_GLIBCXX_USE_CXX11_ABI=0
nvcc fatal : Unsupported gpu architecture 'compute_120a'
ninja: build stopped: subcommand failed.
Traceback (most recent call last):
File "/home/tubor/lib/python3.12/site-packages/torch/utils/cpp_extension.py", line 2209, in _run_ninja_build
subprocess.run(
File "/usr/lib/python3.12/subprocess.py", line 571, in run
raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['ninja', '-v']' returned non-zero exit status 1.

Start by reproducing the pip install command on the reported RTX 4000, Python 3.12, PyTorch, and CUDA 12.4 environment. Trace the build configuration that emits the architecture flags for turb​odiffusion/ops/quant/quant.cu, especially compute_120a. Done means the extension builds without requesting an architecture unsupported by the installed nvcc.