Bug: Compilation failure with CUDA support on Windows: 1 error detected in the compilation of ggml-cuda/sum.cu.

[EverchangerL]

Commit 202084d introduced compillation issue on Windows with CUDA: 1 error detected in the compilation of "llama.cpp/ggml/src/ggml-cuda/sum.cu".

Windows 11 26100.1591
CUDA Toolkit 12.6.1

Visual Studio 2022 version 17.8.13 (Desktop development with C++) - the latest working version with AVX1 support
Also the same issue with Visual Studio 2019 version 16.11.39

Logs:

cmake -B build -DGGML_CUDA=ON

-- Building for: Visual Studio 16 2019
-- Selecting Windows SDK version 10.0.19041.0 to target Windows 10.0.26100.
-- The C compiler identification is MSVC 19.29.30154.0
-- The CXX compiler identification is MSVC 19.29.30154.0
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: C:/Program Files (x86)/Microsoft Visual Studio/2019/Enterprise/VC/Tools/MSVC/14.29.30133/bin/Hostx64/x64/cl.exe - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: C:/Program Files (x86)/Microsoft Visual Studio/2019/Enterprise/VC/Tools/MSVC/14.29.30133/bin/Hostx64/x64/cl.exe - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Found Git: C:/Program Files/Git/cmd/git.exe (found version "2.46.0.windows.1")
-- Looking for pthread.h
-- Looking for pthread.h - not found
-- Found Threads: TRUE
-- Found OpenMP_C: -openmp (found version "2.0")
-- Found OpenMP_CXX: -openmp (found version "2.0")
-- Found OpenMP: TRUE (found version "2.0")
-- OpenMP found
-- Using llamafile
-- Found CUDAToolkit: F:/LLM/Apps/Cuda/include (found version "12.6.68")
-- CUDA found
-- Using CUDA architectures: 52;61;70;75
-- The CUDA compiler identification is NVIDIA 12.6.68
-- Detecting CUDA compiler ABI info
-- Detecting CUDA compiler ABI info - done
-- Check for working CUDA compiler: F:/LLM/Apps/Cuda/bin/nvcc.exe - skipped
-- Detecting CUDA compile features
-- Detecting CUDA compile features - done
-- ccache found, compilation results will be cached. Disable with GGML_CCACHE=OFF.
-- CMAKE_SYSTEM_PROCESSOR: AMD64
-- CMAKE_GENERATOR_PLATFORM:
-- x86 detected
-- Performing Test HAS_AVX_1
-- Performing Test HAS_AVX_1 - Success
-- Performing Test HAS_AVX2_1
-- Performing Test HAS_AVX2_1 - Failed
-- Performing Test HAS_AVX2_2
-- Performing Test HAS_AVX2_2 - Failed
-- Performing Test HAS_FMA_1
-- Performing Test HAS_FMA_1 - Failed
-- Performing Test HAS_FMA_2
-- Performing Test HAS_FMA_2 - Failed
-- Performing Test HAS_AVX512_1
-- Performing Test HAS_AVX512_1 - Failed
-- Performing Test HAS_AVX512_2
-- Performing Test HAS_AVX512_2 - Failed
-- Configuring done
-- Generating done
-- Build files have been written to: F:/LLM/llama.cpp/build

cmake --build build --config Release

Microsoft (R) Build Engine version 16.11.2+f32259642 for .NET Framework
Copyright (C) Microsoft Corporation. All rights reserved.

  Checking Build System
  Building Custom Rule F:/LLM/llama.cpp/examples/gguf-hash/CMakeLists.txt
  Building Custom Rule F:/LLM/llama.cpp/examples/gguf-hash/CMakeLists.txt
  sha256.c
  xxhash.c
  sha256.vcxproj -> F:\LLM\llama.cpp\build\examples\gguf-hash\sha256.dir\Release\sha256.lib
  Generating build details from Git
  xxhash.vcxproj -> F:\LLM\llama.cpp\build\examples\gguf-hash\xxhash.dir\Release\xxhash.lib
  Building Custom Rule F:/LLM/llama.cpp/ggml/src/CMakeLists.txt
  -- Found Git: C:/Program Files/Git/cmd/git.exe (found version "2.46.0.windows.1")
  Building Custom Rule F:/LLM/llama.cpp/examples/gguf-hash/CMakeLists.txt
  Building Custom Rule F:/LLM/llama.cpp/common/CMakeLists.txt
  sha1.c
  build-info.cpp
  build_info.vcxproj -> F:\LLM\llama.cpp\build\common\build_info.dir\Release\build_info.lib
  sha1.vcxproj -> F:\LLM\llama.cpp\build\examples\gguf-hash\sha1.dir\Release\sha1.lib
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\argsort.cu...
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\binbcast.cu...
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\arange.cu...
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\acc.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\argsort.obj "F:\L
  LM\llama.cpp\ggml\src\ggml-cuda\argsort.cu"

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\binbcast.obj "F:\
  LLM\llama.cpp\ggml\src\ggml-cuda\binbcast.cu"

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\acc.obj "F:\LLM\l
  lama.cpp\ggml\src\ggml-cuda\acc.cu"

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\arange.obj "F:\LL
  M\llama.cpp\ggml\src\ggml-cuda\arange.cu"
  acc.cu
  arange.cu
  argsort.cu
  tmpxft_00000cb8_00000000-7_acc.compute_75.cudafe1.cpp
  tmpxft_000024e8_00000000-7_arange.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\clamp.cu...
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\concat.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\clamp.obj "F:\LLM
  \llama.cpp\ggml\src\ggml-cuda\clamp.cu"

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\concat.obj "F:\LL
  M\llama.cpp\ggml\src\ggml-cuda\concat.cu"
  tmpxft_000026d0_00000000-7_argsort.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\conv-transpose-1d.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\conv-transpose-1d
  .obj "F:\LLM\llama.cpp\ggml\src\ggml-cuda\conv-transpose-1d.cu"
  binbcast.cu
  tmpxft_00001f6c_00000000-7_binbcast.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\convert.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\convert.obj "F:\L
  LM\llama.cpp\ggml\src\ggml-cuda\convert.cu"
  clamp.cu
  concat.cu
  conv-transpose-1d.cu
  tmpxft_0000487c_00000000-7_clamp.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\cpy.cu...
  tmpxft_000017fc_00000000-7_concat.compute_75.cudafe1.cpp
  tmpxft_000047b0_00000000-7_conv-transpose-1d.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\cross-entropy-loss.cu...
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\diagmask.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\cpy.obj "F:\LLM\l
  lama.cpp\ggml\src\ggml-cuda\cpy.cu"

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\cross-entropy-los
  s.obj "F:\LLM\llama.cpp\ggml\src\ggml-cuda\cross-entropy-loss.cu"

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\diagmask.obj "F:\
  LLM\llama.cpp\ggml\src\ggml-cuda\diagmask.cu"
  convert.cu
  tmpxft_00004c50_00000000-7_convert.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\dmmv.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\dmmv.obj "F:\LLM\
  llama.cpp\ggml\src\ggml-cuda\dmmv.cu"
  diagmask.cu
  cross-entropy-loss.cu
  tmpxft_0000478c_00000000-7_diagmask.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\fattn-tile-f16.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\fattn-tile-f16.ob
  j "F:\LLM\llama.cpp\ggml\src\ggml-cuda\fattn-tile-f16.cu"
  tmpxft_000010b8_00000000-7_cross-entropy-loss.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\fattn-tile-f32.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\fattn-tile-f32.ob
  j "F:\LLM\llama.cpp\ggml\src\ggml-cuda\fattn-tile-f32.cu"
  cpy.cu
  tmpxft_000040c8_00000000-7_cpy.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\fattn.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\fattn.obj "F:\LLM
  \llama.cpp\ggml\src\ggml-cuda\fattn.cu"
  dmmv.cu
  tmpxft_000054c4_00000000-7_dmmv.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\getrows.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\getrows.obj "F:\L
  LM\llama.cpp\ggml\src\ggml-cuda\getrows.cu"
  getrows.cu
  tmpxft_00005734_00000000-7_getrows.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\im2col.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\im2col.obj "F:\LL
  M\llama.cpp\ggml\src\ggml-cuda\im2col.cu"
  im2col.cu
  tmpxft_0000464c_00000000-7_im2col.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\mmq.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\mmq.obj "F:\LLM\l
  lama.cpp\ggml\src\ggml-cuda\mmq.cu"
  mmq.cu
  fattn.cu
  tmpxft_000054ec_00000000-7_fattn.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\mmvq.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\mmvq.obj "F:\LLM\
  llama.cpp\ggml\src\ggml-cuda\mmvq.cu"
  fattn-tile-f16.cu
  tmpxft_000052c0_00000000-7_fattn-tile-f16.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\norm.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\norm.obj "F:\LLM\
  llama.cpp\ggml\src\ggml-cuda\norm.cu"
  fattn-tile-f32.cu
  tmpxft_000038b4_00000000-7_fattn-tile-f32.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\pad.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\pad.obj "F:\LLM\l
  lama.cpp\ggml\src\ggml-cuda\pad.cu"
  tmpxft_000056e0_00000000-7_mmq.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\pool2d.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\pool2d.obj "F:\LL
  M\llama.cpp\ggml\src\ggml-cuda\pool2d.cu"
  norm.cu
  tmpxft_00005840_00000000-7_norm.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\quantize.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\quantize.obj "F:\
  LLM\llama.cpp\ggml\src\ggml-cuda\quantize.cu"
  pad.cu
  tmpxft_00001990_00000000-7_pad.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\rope.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\rope.obj "F:\LLM\
  llama.cpp\ggml\src\ggml-cuda\rope.cu"
  pool2d.cu
  tmpxft_000020b8_00000000-7_pool2d.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\scale.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\scale.obj "F:\LLM
  \llama.cpp\ggml\src\ggml-cuda\scale.cu"
  quantize.cu
  rope.cu
  tmpxft_000056b8_00000000-7_rope.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\softmax.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\softmax.obj "F:\L
  LM\llama.cpp\ggml\src\ggml-cuda\softmax.cu"
  scale.cu
  tmpxft_000047c4_00000000-7_scale.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\sum.cu...

  F:\LLM\llama.cpp\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=compute_52,code=\"comput
  e_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arch=compute_61,code=\"c
  ompute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencode=arch=compute_70,cod
  e=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gencode=arch=compute_7
  5,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" --use-local-env -c
  cbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\Hos
  tX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -IF:\LLM\Apps\C
  uda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --m
  achine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_
  CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP
  -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_
  PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml
  _EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNI
  NGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA
  _MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPE
  N_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nolo
  go /O2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\sum.obj "F:\LLM\l
  lama.cpp\ggml\src\ggml-cuda\sum.cu"
C:\Program Files (x86)\Windows Kits\10\Include\10.0.19041.0\um\oaidl.h(804): error : expected an identif
ier [F:\LLM\llama.cpp\build\ggml\src\ggml.vcxproj]
            600 = CC_MSCPASCAL,
            ^

  tmpxft_00000ab8_00000000-7_quantize.compute_75.cudafe1.cpp
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\sumrows.cu...
  1 error detected in the compilation of "F:/LLM/llama.cpp/ggml/src/ggml-cuda/sum.cu".
  sum.cu
C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\MSBuild\Microsoft\VC\v160\BuildCustomizat
ions\CUDA 12.6.targets(799,9): error MSB3721: The command ""F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch
=compute_52,code=\"compute_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=a
rch=compute_61,code=\"compute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencod
e=arch=compute_70,code=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -gen
code=arch=compute_75,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75\" -
-use-local-env -ccbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.2
9.30133\bin\HostX64\x64" -x cu   -IF:\LLM\llama.cpp\ggml\src\..\include -IF:\LLM\llama.cpp\ggml\src\. -I
F:\LLM\Apps\Cuda\include -IF:\LLM\Apps\Cuda\include     --keep-dir x64\Release -use_fast_math -maxrregco
unt=0   --machine 64 --compile -cudart static -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -D
GGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_
OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_
CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dg
gml_EXPORTS -DWIN32 -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARN
INGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_
MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_S
OURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nologo /O
2 /FS   /MD /GR" -Xcompiler "/Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\sum.obj "F:\LLM\llama.cp
p\ggml\src\ggml-cuda\sum.cu"" exited with code 1. [F:\LLM\llama.cpp\build\ggml\src\ggml.vcxproj]

I know nothing about coding, so I don't know how to fix this. I have tried different versions of the Windows SDK without success with Visual Studio 2022 version 17.8.13. Then I installed Visual Studio 19 in the hope of fixing this, but got the same problem.

Hope you can fix this, thanks.

[JohannesGaessler]

C:\Program Files (x86)\Windows Kits\10\Include\10.0.19041.0\um\oaidl.h(804): error : expected an identif
ier [F:\LLM\llama.cpp\build\ggml\src\ggml.vcxproj]
            600 = CC_MSCPASCAL,

What does that file say on line 804? Does it say CC_PASCAL = CC_MSCPASCAL?

[eskeletor97]

C:\Program Files (x86)\Windows Kits\10\Include\10.0.19041.0\um\oaidl.h(804): error : expected an identif
ier [F:\LLM\llama.cpp\build\ggml\src\ggml.vcxproj]
            600 = CC_MSCPASCAL,

What does that file say on line 804? Does it say CC_PASCAL = CC_MSCPASCAL?

Hi, I'm having identical issue and in my case, it does exactly say that.

[EverchangerL]

yes, in Windows SDK ver. 19041 it's this

CC_PASCAL = CC_MSCPASCAL,

[JohannesGaessler]

Please confirm whether or not this fix works: #9382

[EverchangerL]

Looks like it worked, thanks!

log

  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\sum.cu...
  F:\LLM\llama.cpp-cuda-cub-fix-windows\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=com
  pute_52,code=\"compute_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arc
  h=compute_61,code=\"compute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencod
  e=arch=compute_70,code=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -g
  encode=arch=compute_75,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75
  \" --use-local-env -ccbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSV
  C\14.29.30133\bin\HostX64\x64" -x cu   -I"F:\LLM\llama.cpp-cuda-cub-fix-windows\ggml\src\..\include" -
  I"F:\LLM\llama.cpp-cuda-cub-fix-windows\ggml\src\." -IF:\LLM\Apps\Cuda\include -IF:\LLM\Apps\Cuda\incl
  ude     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --machine 64 --compile -cudart static
  -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_
  CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DM
  MV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA
  _USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -DWIN32 -D_WINDOWS -DNDEBU
  G -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGM
  L_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2
   -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Rele
  ase\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nologo /O2 /FS   /MD /GR" -Xcompiler "/
  Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\sum.obj "F:\LLM\llama.cpp-cuda-cub-fix-windows\ggml\
  src\ggml-cuda\sum.cu"
  scale.cu
  tmpxft_0000575c_00000000-7_scale.compute_75.cudafe1.cpp
 
  Compiling CUDA source file ..\..\..\ggml\src\ggml-cuda\sumrows.cu...
  F:\LLM\llama.cpp-cuda-cub-fix-windows\build\ggml\src>"F:\LLM\Apps\Cuda\bin\nvcc.exe" -gencode=arch=com
  pute_52,code=\"compute_52,compute_52\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" -gencode=arc
  h=compute_61,code=\"compute_61,compute_61\" -gencode=arch=compute_61,code=\"sm_61,compute_61\" -gencod
  e=arch=compute_70,code=\"compute_70,compute_70\" -gencode=arch=compute_70,code=\"sm_70,compute_70\" -g
  encode=arch=compute_75,code=\"compute_75,compute_75\" -gencode=arch=compute_75,code=\"sm_75,compute_75
  \" --use-local-env -ccbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSV
  C\14.29.30133\bin\HostX64\x64" -x cu   -I"F:\LLM\llama.cpp-cuda-cub-fix-windows\ggml\src\..\include" -
  I"F:\LLM\llama.cpp-cuda-cub-fix-windows\ggml\src\." -IF:\LLM\Apps\Cuda\include -IF:\LLM\Apps\Cuda\incl
  ude     --keep-dir x64\Release -use_fast_math -maxrregcount=0   --machine 64 --compile -cudart static
  -Xcompiler="/EHsc -Ob2 /arch:AVX"   -D_WINDOWS -DNDEBUG -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_
  CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DM
  MV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2 -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA
  _USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Release\"" -Dggml_EXPORTS -DWIN32 -D_WINDOWS -DNDEBU
  G -DGGML_USE_CUDA -DGGML_SHARED -DGGML_BUILD -D_CRT_SECURE_NO_WARNINGS -DGGML_SCHED_MAX_COPIES=4 -DGGM
  L_USE_OPENMP -DGGML_USE_LLAMAFILE -DGGML_CUDA_DMMV_X=32 -DGGML_CUDA_MMV_Y=1 -DK_QUANTS_PER_ITERATION=2
   -DGGML_CUDA_PEER_MAX_BATCH_SIZE=128 -DGGML_CUDA_USE_GRAPHS -D_XOPEN_SOURCE=600 -D"CMAKE_INTDIR=\"Rele
  ase\"" -Dggml_EXPORTS -D_WINDLL -D_MBCS -Xcompiler "/EHsc /W3 /nologo /O2 /FS   /MD /GR" -Xcompiler "/
  Fdggml.dir\Release\vc142.pdb" -o ggml.dir\Release\sumrows.obj "F:\LLM\llama.cpp-cuda-cub-fix-windows\g
  gml\src\ggml-cuda\sumrows.cu"
  softmax.cu
  tmpxft_00004dec_00000000-7_softmax.compute_75.cudafe1.cpp

[EverchangerL]

Can confirm that ggml.dll was compiled successfully, thanks again.

Auto build dll exports
     Creating library F:/LLM/llama.cpp-cuda-cub-fix-windows/build/ggml/src/Release/ggml.lib and object F
  :/LLM/llama.cpp-cuda-cub-fix-windows/build/ggml/src/Release/ggml.exp
  ggml.vcxproj -> F:\LLM\llama.cpp-cuda-cub-fix-windows\build\bin\Release\ggml.dll