Compile error on windows10 in the latest version 1.3.18 of mmcv-full.

[DingYuan0118]

Describe the Issue
Run the python setup.py install will raise an error similar to #888 on **/mmcv/ops/csrc/common/cuda/voxelization_cuda_kernel.cuh

Environment

1. Please run python -c "from mmcv.utils import collect_env; print(collect_env())" to collect necessary environment information and paste it here.
   {'sys.platform': 'win32', 'Python': '3.7.8 (tags/v3.7.8:4b47a5b6ba, Jun 28 2020, 08:53:46) [MSC v.1916 64 bit (AMD64)]', 'CUDA available': True, 'GPU 0': 'GeForce GTX 850M', 'CUDA_HOME': 'C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2', 'NVCC': 'Cuda compilation tools, release 10.2, V10.2.89', 'GCC': 'gcc (x86_64-posix-seh-rev0, Built by MinGW-W64 project) 8.1.0', 'PyTorch': '1.7.1+cu101', 'PyTorch compiling details': 'PyTorch built with:\n - C++ Version: 199711\n - MSVC 192729112\n - Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191125 for Intel(R) 64 architecture applications\n - Intel(R) MKL-DNN v1.6.0 (Git Hash 5ef631a030a6f73131c77892041042805a06064f)\n - OpenMP 2019\n - CPU capability usage: AVX2\n - CUDA Runtime 10.1\n - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_61,code=sm_61;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_37,code=compute_37\n - CuDNN 7.6.4\n - Magma 2.5.4\n - Build settings: BLAS=MKL, BUILD_TYPE=Release, CXX_FLAGS=/DWIN32 /D_WINDOWS /GR
   /EHsc /w /bigobj -openmp:experimental -DNDEBUG -DUSE_FBGEMM -DUSE_VULKAN_WRAPPER, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, USE_CUDA=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=OFF, USE_NNPACK=OFF, USE_OPENMP=ON, \n', 'TorchVision': '0.8.2+cu110', 'OpenCV': '4.5.2', 'MMCV': '1.3.18', 'MMCV Compiler': 'n/a', 'MMCV CUDA Compiler': 'n/a'}

Bug fix
Maybe we should replace the floor to floorf on the file as #891 do using a script:

#!/bin/bash

SITE=E:/TaskAndAnswer/code/mmcv-full-1.3.18/mmcv
sed -i.bak -e 's/floor/floorf/g' "${SITE}/ops/csrc/common/cuda/voxelization_cuda_kernel.cuh"

[DingYuan0118]

After running the script described above, a new link issue raise up as described in #1555 .
The solution:

• 1. download the latested mmcv repo (or git clone),
• 2. use above script to fix the floor error(pay attention that you need to change the SITE for your own mmcv path)
• 3. set the environment variable MMCV_WITH_OPS=1 or replace the os.getenv('MMCV_WITH_OPS', '0') == '0' to os.getenv('MMCV_WITH_OPS', '1') == '0' in the setup.py Line178 and Line357 to build the .cpp and .cu file
• 4. run python setup.py install

The mmcv-full will successfully installed in the environment. Run check_installation.py outputs

Start checking the installation of mmcv-full ...
CPU ops were compiled successfully.
CUDA ops were compiled successfully.
mmcv-full has been installed successfully.

Environment information:
------------------------------------------------------------
sys.platform: win32
Python: 3.7.8 (tags/v3.7.8:4b47a5b6ba, Jun 28 2020, 08:53:46) [MSC v.1916 64 bit (AMD64)]
CUDA available: True
GPU 0: GeForce GTX 850M
CUDA_HOME: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2
NVCC: Cuda compilation tools, release 10.2, V10.2.89
GCC: gcc (x86_64-posix-seh-rev0, Built by MinGW-W64 project) 8.1.0
PyTorch: 1.7.1+cu101
PyTorch compiling details: PyTorch built with:
  - C++ Version: 199711
  - MSVC 192729112
  - Intel(R) Math Kernel Library Version 2020.0.0 Product Build 20191125 for Intel(R) 64 architecture applications
  - Intel(R) MKL-DNN v1.6.0 (Git Hash 5ef631a030a6f73131c77892041042805a06064f)
  - OpenMP 2019
  - CPU capability usage: AVX2
  - CUDA Runtime 10.1
  - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_61,code=sm_61;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_37,code=compute_37
  - CuDNN 7.6.4
  - Magma 2.5.4
  - Build settings: BLAS=MKL, BUILD_TYPE=Release, CXX_FLAGS=/DWIN32 /D_WINDOWS /GR /EHsc /w /bigobj -openmp:experimental -DNDEBUG -DUSE_FBGEMM -DUSE_VULKAN_WRAPPER, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, USE_CUDA=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=OFF, USE_NNPACK=OFF, USE_OPENMP=ON,

TorchVision: 0.8.2+cu110
OpenCV: 4.5.2
MMCV: 1.4.0
MMCV Compiler: MSVC 192829913
MMCV CUDA Compiler: 10.2
------------------------------------------------------------

We can find that the package has been installed successfully.