[Bug] Error occurred during tvm compilation

Thanks for participating in the TVM community! We use https://discuss.tvm.ai for any general usage questions and discussions. The issue tracker is used for actionable items such as feature proposals discussion, roadmaps, and bug tracking. You are always welcomed to post on the forum first :smile_cat:

Issues that are inactive for a period of time may get closed. We adopt this policy so that we won't lose track of actionable issues that may fall at the bottom of the pile. Feel free to reopen a new one if you feel there is an additional problem that needs attention when an old one gets closed.

### Expected behavior

### Actual behavior

error MSB3721: “"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.1\bin\nvcc.exe" --use-local-env -cc
bin "D:\Program Files\Microsoft Visual Studio\2022\Community\VC\Tools\MSVC\14.40.33807\bin\HostX64\x64" -x cu -I"E:\s
tudy\tvm\apache-tvm-src-v0.20.0\include" -I"E:\study\tvm\apache-tvm-src-v0.20.0\3rdparty\cutlass_fpA_intB_gemm\cutlass_
kernels\..\cutlass_extensions\include" -I"E:\study\tvm\apache-tvm-src-v0.20.0\3rdparty\cutlass_fpA_intB_gemm\cutlass\in
clude" -I"E:\study\tvm\apache-tvm-src-v0.20.0\3rdparty\cutlass_fpA_intB_gemm\cutlass_kernels\..\utils" -I"E:\study\tvm\
apache-tvm-src-v0.20.0\3rdparty\cutlass_fpA_intB_gemm\cutlass_kernels\.." -I"E:\study\tvm\apache-tvm-src-v0.20.0\3rdpar
ty\cutlass_fpA_intB_gemm" -I"E:\study\tvm\apache-tvm-src-v0.20.0\3rdparty\dlpack\include" -I"E:\study\tvm\apache-tvm-sr
c-v0.20.0\3rdparty\dmlc-core\include" -I"E:\study\tvm\apache-tvm-src-v0.20.0\3rdparty\rang\include" -I"E:\study\tvm\apa
che-tvm-src-v0.20.0\3rdparty\compiler-rt" -I"E:\study\tvm\apache-tvm-src-v0.20.0\3rdparty\picojson" -I"C:\Program Files
\NVIDIA GPU Computing Toolkit\CUDA\v12.1\include" -I"D:\exProgram\miniforge3\envs\tvm-build-venv\Library\include" -I"C:
\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.1\include" --keep-dir fpA_intB_gemm\x64\Debug -maxrregcount=0
--machine 64 --compile -cudart static -std=c++17 --generate-code=arch=compute_52,code=[compute_52,sm_52] /bigobj /wd
4244 /wd4267 /wd4018 /wd4316 /wd4101 /wd4180 /wd4251 /wd4624 /wd4146 /wd4141 /wd4068 -expt-relaxed-constexpr -Xcompiler
="/EHsc -Zi -Ob0" -g -D_WINDOWS -DWIN32_LEAN_AND_MEAN -D_CRT_SECURE_NO_WARNINGS -D_SCL_SECURE_NO_WARNINGS -D_ENABLE_EX
TENDED_ALIGNED_STORAGE -DNOMINMAX -DTVM_INDEX_DEFAULT_I64=1 -DTVM_KALLOC_ALIGNMENT=64 -DTVM_THREADPOOL_USE_OPENMP=0 -DD
MLC_USE_FOPEN64=0 -DNDEBUG=1 -D_CRT_SECURE_NO_DEPRECATE -D_CRT_NONSTDC_NO_DEPRECATE -D_CRT_NONSTDC_NO_WARNINGS -D_SCL_S
ECURE_NO_DEPRECATE -DUNICODE -D_UNICODE -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS -DTVM_LLV
M_VERSION=201 -DTVM_LLVM_HAS_AARCH64_TARGET=1 -D"CMAKE_INTDIR=\"Debug\"" -DfpA_intB_gemm_EXPORTS -D_WINDLL -D_UNICODE -
DUNICODE -DWIN32 -D_WINDOWS -DWIN32_LEAN_AND_MEAN -D_CRT_SECURE_NO_WARNINGS -D_SCL_SECURE_NO_WARNINGS -D_ENABLE_EXTENDE
D_ALIGNED_STORAGE -DNOMINMAX -DTVM_INDEX_DEFAULT_I64=1 -DTVM_KALLOC_ALIGNMENT=64 -DTVM_THREADPOOL_USE_OPENMP=0 -DDMLC_U
SE_FOPEN64=0 -DNDEBUG=1 -D_CRT_SECURE_NO_DEPRECATE -D_CRT_NONSTDC_NO_DEPRECATE -D_CRT_NONSTDC_NO_WARNINGS -D_SCL_SECURE
_NO_DEPRECATE -DUNICODE -D_UNICODE -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS -DTVM_LLVM_VER
SION=201 -DTVM_LLVM_HAS_AARCH64_TARGET=1 -D"CMAKE_INTDIR=\"Debug\"" -DfpA_intB_gemm_EXPORTS -Xcompiler "/EHsc /W1 /nolo
go /Od /FS /Zi /RTC1 /MD /GR" -Xcompiler "/FdfpA_intB_gemm.dir\Debug\vc143.pdb" -o fpA_intB_gemm.dir\Debug\moe_gemm_ker
nels_fp16_fp16.obj "E:\study\tvm\apache-tvm-src-v0.20.0\3rdparty\cutlass_fpA_intB_gemm\cutlass_kernels\moe_gemm\moe_gem
m_kernels_fp16_fp16.cu"”return code 1。 [E:\study\tvm\apache-tvm-src-v0.20.0\build\3rdparty\cutlass_fpA_intB_gemm\cutlass_
kernels\fpA_intB_gemm.vcxproj]
nvcc fatal : A single input file is required for a non-link phase when an outputfile is specified

### Environment

win10,tvm 0.20.0

### Steps to reproduce

Reproduce“ https://tvm.apache.org/docs/install/from_source.html#step -The compilation steps for "1-install dependencies" have enabled CUDA, CUBLAS, CUDNN, CUTlass, and LLVM




Start with the CUDA build command and generated build/3rdparty/cutlass_fpA_intB_gemm/cutlass_kernels/fpA_intB_gemm.vcxproj referenced in the error, then review the CUTLASS kernel path and the Windows build instructions at the linked TVM installation page. Reproduce the configuration with CUDA, CUBLAS, CUDNN, CUTLASS, and LLVM enabled and identify why nvcc rejects the compilation; done means the reported TVM 0.20.0 Windows build completes successfully.

Written by the indexing model from the issue text.