#---------------------------------------------------------------------------------------- # INFO: # - How to use: change the flags in the 'Set one of these ON and all others OFF' section # - CUDA: it will use automatically the CUDA SDK version installed # #---------------------------------------------------------------------------------------- cmake_minimum_required(VERSION 3.8) project(LLamaSharpCpp VERSION 0.10.0 LANGUAGES CXX CSharp) if(NOT MSVC) message(FATAL_ERROR "This CMake file only works with MSVC.") endif(NOT MSVC) #--------- Set one of these ON and all others OFF -------------------> option(LLAMA_CUDA_AVX2 "CUDA + AVX2" ON) option(LLAMA_AVX2 "AVX2 (no CUDA)" OFF) option(LLAMA_CUDA "CUDA (no AVX)" OFF) #etc... add other setups #<--------- Set one of these ON and all others OFF ------------------- # --------------- Don't change below this line ----------------------- # Variable Settings if(LLAMA_CUDA_AVX2) option(LLAMA_AVX "llama: enable AVX" ON) option(LLAMA_AVX2 "llama: enable AVX2" ON) option(LLAMA_CUBLAS "llama: use CUDA" ON) elseif(LLAMA_AVX2) option(LLAMA_AVX "llama: enable AVX" ON) option(LLAMA_AVX2 "llama: enable AVX2" ON) option(LLAMA_CUBLAS "llama: use CUDA" OFF) elseif(LLAMA_CUDA) option(LLAMA_AVX "llama: enable AVX" OFF) option(LLAMA_AVX2 "llama: enable AVX2" OFF) option(LLAMA_CUBLAS "llama: use CUDA" ON) elseif(OTHER_SETUPS) #etc... endif() # Fixed Settings # general option(BUILD_SHARED_LIBS "build shared libraries" ON) option(LLAMA_STATIC "llama: static link libraries" OFF) option(LLAMA_NATIVE "llama: enable -march=native flag" OFF) option(LLAMA_LTO "llama: enable link time optimization" OFF) option(LLAMA_CCACHE "llama: use ccache if available" ON) # debug option(LLAMA_ALL_WARNINGS "llama: enable all compiler warnings" ON) option(LLAMA_ALL_WARNINGS_3RD_PARTY "llama: enable all compiler warnings in 3rd party libs" OFF) option(LLAMA_GPROF "llama: enable gprof" OFF) # build option(LLAMA_FATAL_WARNINGS "llama: enable -Werror flag" OFF) # sanitizers option(LLAMA_SANITIZE_THREAD "llama: enable thread sanitizer" OFF) option(LLAMA_SANITIZE_ADDRESS "llama: enable address sanitizer" OFF) option(LLAMA_SANITIZE_UNDEFINED "llama: enable undefined sanitizer" OFF) option(LLAMA_AVX512 "llama: enable AVX512" OFF) option(LLAMA_AVX512_VBMI "llama: enable AVX512-VBMI" OFF) option(LLAMA_AVX512_VNNI "llama: enable AVX512-VNNI" OFF) option(LLAMA_FMA "llama: enable FMA" OFF) # in MSVC F16C is implied with AVX2/AVX512 if (NOT MSVC) option(LLAMA_F16C "llama: enable F16C" OFF) endif() if (WIN32) set(LLAMA_WIN_VER "0x602" CACHE STRING "llama: Windows Version") endif() # 3rd party libs option(LLAMA_ACCELERATE "llama: enable Accelerate framework" ON) option(LLAMA_BLAS "llama: use BLAS" OFF) set(LLAMA_BLAS_VENDOR "Generic" CACHE STRING "llama: BLAS library vendor") #option(LLAMA_CUDA_CUBLAS "llama: use cuBLAS for prompt processing" OFF) option(LLAMA_CUDA_FORCE_DMMV "llama: use dmmv instead of mmvq CUDA kernels" OFF) option(LLAMA_CUDA_FORCE_MMQ "llama: use mmq kernels instead of cuBLAS" OFF) set(LLAMA_CUDA_DMMV_X "32" CACHE STRING "llama: x stride for dmmv CUDA kernels") set(LLAMA_CUDA_MMV_Y "1" CACHE STRING "llama: y block size for mmv CUDA kernels") option(LLAMA_CUDA_F16 "llama: use 16 bit floats for some calculations" OFF) set(LLAMA_CUDA_KQUANTS_ITER "2" CACHE STRING "llama: iters./thread per block for Q2_K/Q6_K") set(LLAMA_CUDA_PEER_MAX_BATCH_SIZE "128" CACHE STRING "llama: max. batch size for using peer access") option(LLAMA_HIPBLAS "llama: use hipBLAS" OFF) option(LLAMA_HIP_UMA "llama: use HIP unified memory architecture" OFF) option(LLAMA_CLBLAST "llama: use CLBlast" OFF) option(LLAMA_VULKAN "llama: use Vulkan" OFF) option(LLAMA_VULKAN_CHECK_RESULTS "llama: run Vulkan op checks" OFF) option(LLAMA_VULKAN_DEBUG "llama: enable Vulkan debug output" OFF) option(LLAMA_VULKAN_VALIDATE "llama: enable Vulkan validation" OFF) option(LLAMA_VULKAN_RUN_TESTS "llama: run Vulkan tests" OFF) option(LLAMA_METAL "llama: use Metal" ${LLAMA_METAL_DEFAULT}) option(LLAMA_METAL_NDEBUG "llama: disable Metal debugging" OFF) option(LLAMA_METAL_SHADER_DEBUG "llama: compile Metal with -fno-fast-math" OFF) option(LLAMA_METAL_EMBED_LIBRARY "llama: embed Metal library" OFF) option(LLAMA_KOMPUTE "llama: use Kompute" OFF) option(LLAMA_MPI "llama: use MPI" OFF) option(LLAMA_QKK_64 "llama: use super-block size of 64 for k-quants" OFF) option(LLAMA_SYCL "llama: use SYCL" OFF) option(LLAMA_SYCL_F16 "llama: use 16 bit floats for sycl calculations" OFF) option(LLAMA_CPU_HBM "llama: use memkind for CPU HBM" OFF) option(LLAMA_BUILD_TESTS "llama: build tests" OFF) option(LLAMA_BUILD_EXAMPLES "llama: build examples" ON) option(LLAMA_BUILD_SERVER "llama: build server example" OFF) # add perf arguments option(LLAMA_PERF "llama: enable perf" OFF) include_external_msproject(LLama.Unittest ./LLama.Unittest/LLama.Unittest.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE BAC1CFA9-E6AC-4BD0-A548-A8066D3C467E) include_external_msproject(LLama.Examples ./LLama.Examples/LLama.Examples.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE BD1909AD-E1F8-476E-BC49-E394FF0470CE) include_external_msproject(LLamaSharp ./LLama/LLamaSharp.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE 01A12D68-DE95-425E-AEEE-2D099305036D) include_external_msproject(LLama.WebAPI ./LLama.WebAPI/LLama.WebAPI.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE D3CEC57A-9027-4DA4-AAAC-612A1EB50ADF) include_external_msproject(LLama.Web ./LLama.Web/LLama.Web.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE C3531DB2-1B2B-433C-8DE6-3541E3620DB1) include_external_msproject(LLamaSharp.SemanticKernel ./LLama.SemanticKernel/LLamaSharp.SemanticKernel.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE D98F93E3-B344-4F9D-86BB-FDBF6768B587) include_external_msproject(LLamaSharp.KernelMemory ./LLama.KernelMemory/LLamaSharp.KernelMemory.csproj GUID 9A19103F-16F7-4668-BE54-9A1E7A4F7556 TYPE E5589AE7-B86F-4343-A1CC-8E5D34596E52) add_subdirectory(./llama.cpp)