cmake_minimum_required(VERSION 3.18)
project(tvm C CXX)
# Utility functions
include(cmake/utils/Utils.cmake)
include(cmake/utils/Summary.cmake)
include(cmake/utils/Linker.cmake)
include(cmake/utils/FindCUDA.cmake)
include(cmake/utils/FindNCCL.cmake)
include(cmake/utils/FindOpenCL.cmake)
include(cmake/utils/FindVulkan.cmake)
include(cmake/utils/FindLLVM.cmake)
include(cmake/utils/FindROCM.cmake)
include(cmake/utils/FindRCCL.cmake)
include(cmake/utils/FindEthosN.cmake)
if(EXISTS ${CMAKE_BINARY_DIR}/config.cmake)
include(${CMAKE_BINARY_DIR}/config.cmake)
else()
if(EXISTS ${CMAKE_SOURCE_DIR}/config.cmake)
include(${CMAKE_SOURCE_DIR}/config.cmake)
endif()
endif()
# NOTE: do not modify this file to change option values.
# You can create a config.cmake at build folder
# and add set(OPTION VALUE) to override these build options.
# Alernatively, use cmake -DOPTION=VALUE through command-line.
tvm_option(USE_CUDA "Build with CUDA" OFF)
tvm_option(USE_NCCL "Build with NCCL" OFF)
tvm_option(USE_MSCCL "Build with MSCCL" OFF)
tvm_option(USE_OPENCL "Build with OpenCL" OFF)
tvm_option(USE_OPENCL_ENABLE_HOST_PTR "Enable OpenCL memory object access to host" OFF)
tvm_option(USE_OPENCL_GTEST "Path to OpenCL specific gtest version for runtime cpp tests." /path/to/opencl/gtest)
tvm_option(USE_VULKAN "Build with Vulkan" OFF)
# Whether to use spirv-tools.and SPIRV-Headers from Khronos github or gitlab.
#
# Possible values:
# - OFF: not to use
# - /path/to/install: path to your khronis spirv-tools and SPIRV-Headers installation directory
#
tvm_option(USE_KHRONOS_SPIRV "Whether to use spirv-tools.and SPIRV-Headers from Khronos github or gitlab" OFF)
tvm_option(USE_SPIRV_KHR_INTEGER_DOT_PRODUCT "whether enable SPIRV_KHR_DOT_PRODUCT" OFF)
tvm_option(USE_METAL "Build with Metal" OFF)
tvm_option(USE_ROCM "Build with ROCM" OFF)
tvm_option(USE_RCCL "Build with RCCL" OFF)
tvm_option(ROCM_PATH "The path to rocm" /opt/rocm)
tvm_option(USE_HEXAGON "Build with Hexagon support" OFF)
tvm_option(USE_HEXAGON_SDK "Path to the Hexagon SDK root (required for Hexagon support)" /path/to/sdk)
tvm_option(USE_HEXAGON_RPC "Enable Hexagon RPC using minRPC implementation over Android." OFF)
tvm_option(USE_HEXAGON_GTEST "Path to Hexagon specific gtest version for runtime cpp tests." /path/to/hexagon/gtest)
tvm_option(USE_HEXAGON_EXTERNAL_LIBS "Path to git repo containing external Hexagon runtime sources or libraries" OFF)
tvm_option(USE_RPC "Build with RPC" ON)
tvm_option(USE_THREADS "Build with thread support" ON)
tvm_option(USE_LLVM "Build with LLVM, can be set to specific llvm-config path" OFF)
tvm_option(USE_MLIR "Build with MLIR support" OFF)
tvm_option(USE_STACKVM_RUNTIME "Include stackvm into the runtime" OFF)
tvm_option(USE_GRAPH_EXECUTOR "Build with tiny graph executor" ON)
tvm_option(USE_GRAPH_EXECUTOR_CUDA_GRAPH "Build with tiny graph executor with CUDA Graph for GPUs" OFF)
tvm_option(USE_AOT_EXECUTOR "Build with AOT executor" ON)
tvm_option(USE_PROFILER "Build profiler for the VM and graph executor" ON)
tvm_option(USE_OPENMP "Build with OpenMP thread pool implementation" OFF)
tvm_option(USE_RELAY_DEBUG "Building Relay in debug mode..." OFF)
tvm_option(TVM_DEBUG_WITH_ABI_CHANGE "Enable debug code that may cause ABI changes" OFF)
tvm_option(TVM_LOG_BEFORE_THROW "Whether log before throw, for debugging purposes" OFF)
tvm_option(USE_RTTI "Build with RTTI" ON)
tvm_option(USE_MSVC_MT "Build with MT" OFF)
tvm_option(USE_MICRO "Build with Micro TVM support" OFF)
tvm_option(INSTALL_DEV "Install compiler infrastructure" OFF)
tvm_option(HIDE_PRIVATE_SYMBOLS "Compile with -fvisibility=hidden." OFF)
tvm_option(USE_TF_TVMDSOOP "Build with TensorFlow TVMDSOOp" OFF)
tvm_option(USE_PT_TVMDSOOP "Build with PyTorch TVMDSOOp" OFF)
tvm_option(USE_FALLBACK_STL_MAP "Use TVM's POD compatible Map" OFF)
tvm_option(USE_ETHOSN "Build with Arm(R) Ethos(TM)-N" OFF)
tvm_option(USE_CMSISNN "Build with Arm CMSIS-NN" OFF)
tvm_option(INDEX_DEFAULT_I64 "Defaults the index datatype to int64" ON)
tvm_option(USE_LIBBACKTRACE "Use libbacktrace to supply linenumbers on stack traces" AUTO)
tvm_option(BACKTRACE_ON_SEGFAULT "Install a signal handler to print a backtrace on segfault" OFF)
tvm_option(BUILD_STATIC_RUNTIME "Build static version of libtvm_runtime" OFF)
tvm_option(BUILD_DUMMY_LIBTVM "Build a dummy version of libtvm" OFF)
tvm_option(USE_PAPI "Use Performance Application Programming Interface (PAPI) to read performance counters" OFF)
tvm_option(USE_GTEST "Use GoogleTest for C++ sanity tests" AUTO)
tvm_option(USE_CUSTOM_LOGGING "Use user-defined custom logging, tvm::runtime::detail::LogFatalImpl and tvm::runtime::detail::LogMessageImpl must be implemented" OFF)
tvm_option(USE_ALTERNATIVE_LINKER "Use 'mold' or 'lld' if found when invoking compiler to link artifact" AUTO)
tvm_option(USE_CCACHE "Use ccache if found when invoking compiler" AUTO)
# 3rdparty libraries
tvm_option(DLPACK_PATH "Path to DLPACK" "3rdparty/dlpack/include")
tvm_option(DMLC_PATH "Path to DMLC" "3rdparty/dmlc-core/include")
tvm_option(RANG_PATH "Path to RANG" "3rdparty/rang/include")
tvm_option(COMPILER_RT_PATH "Path to COMPILER-RT" "3rdparty/compiler-rt")
tvm_option(PICOJSON_PATH "Path to PicoJSON" "3rdparty/picojson")
# Contrib library options
tvm_option(USE_BYODT_POSIT "Build with BYODT software emulated posit custom datatype" OFF)
tvm_option(USE_BLAS "The blas library to be linked" none)
tvm_option(USE_AMX "Enable Intel AMX" OFF)
tvm_option(USE_MKL "MKL root path when use MKL blas" OFF)
tvm_option(USE_DNNL "Enable DNNL codegen" OFF)
tvm_option(USE_CUDNN "Build with cuDNN" OFF)
tvm_option(USE_CUBLAS "Build with cuBLAS" OFF)
tvm_option(USE_NVTX "Build with NVTX" OFF)
tvm_option(USE_CUTLASS "Build with CUTLASS" OFF)
tvm_option(USE_THRUST "Build with Thrust" OFF)
tvm_option(USE_CURAND "Build with cuRAND" OFF)
tvm_option(USE_MIOPEN "Build with ROCM:MIOpen" OFF)
tvm_option(USE_ROCBLAS "Build with ROCM:RoCBLAS" OFF)
tvm_option(USE_SORT "Build with sort support" ON)
tvm_option(USE_NNPACK "Build with nnpack support" OFF)
tvm_option(USE_LIBTORCH "Build with libtorch support" OFF)
tvm_option(USE_RANDOM "Build with random support" ON)
tvm_option(USE_MICRO_STANDALONE_RUNTIME "Build with micro.standalone_runtime support" OFF)
tvm_option(USE_CPP_RPC "Build CPP RPC" OFF)
tvm_option(USE_IOS_RPC "Build iOS RPC" OFF)
tvm_option(USE_TFLITE "Build with tflite support" OFF)
tvm_option(USE_TENSORFLOW_PATH "TensorFlow root path when use TFLite" none)
tvm_option(USE_COREML "Build with coreml support" OFF)
tvm_option(USE_BNNS "Build with BNNS support" OFF)
tvm_option(USE_TARGET_ONNX "Build with ONNX Codegen support" OFF)
tvm_option(USE_ARM_COMPUTE_LIB "Build with Arm Compute Library" OFF)
tvm_option(USE_ARM_COMPUTE_LIB_GRAPH_EXECUTOR "Build with Arm Compute Library graph executor" OFF)
tvm_option(USE_TENSORRT_CODEGEN "Build with TensorRT Codegen support" OFF)
tvm_option(USE_TENSORRT_RUNTIME "Build with TensorRT runtime" OFF)
tvm_option(USE_RUST_EXT "Build with Rust based compiler extensions, STATIC, DYNAMIC, or OFF" OFF)
tvm_option(USE_VITIS_AI "Build with VITIS-AI Codegen support" OFF)
tvm_option(SUMMARIZE "Print CMake option summary after configuring" OFF)
tvm_option(USE_CLML "Build with CLML Codegen support" OFF)
tvm_option(USE_CLML_GRAPH_EXECUTOR "Build with CLML graph runtime" OFF)
tvm_option(USE_UMA "Build with UMA support" OFF)
tvm_option(USE_VERILATOR "Build with Verilator support" OFF)
tvm_option(USE_MSC "Enable Multi-System Compiler" OFF)
tvm_option(USE_MRVL "Build with MRVL TVM support" OFF)
# include directories
include_directories(${CMAKE_INCLUDE_PATH})
include_directories("include")
include_directories(SYSTEM ${DLPACK_PATH})
include_directories(SYSTEM ${DMLC_PATH})
include_directories(SYSTEM ${RANG_PATH})
include_directories(SYSTEM ${COMPILER_RT_PATH})
include_directories(SYSTEM ${PICOJSON_PATH})
# initial variables
set(TVM_LINKER_LIBS "")
set(TVM_RUNTIME_LINKER_LIBS "")
# Check if this is being run on its own or as a subdirectory for another p
没有合适的资源?快使用搜索试试~ 我知道了~
温馨提示
机器学习 机器学习使计算机能够从研究数据和统计信息中学习。 机器学习是迈向人工智能(AI)方向的其中一步。 机器学习是一种程序,可以分析数据并学习预测结果。 从何处开始? 在本教程中,我们将回到数学并研究统计学,以及如何根据数据集计算重要数值。 我们还将学习如何使用各种 Python 模块来获得所需的答案。 并且,我们将学习如何根据所学知识编写能够预测结果的函数。 数据集 在计算机中,数据集指的是任何数据集合。它可以是从数组到完整数据库的任何内容。 通过查看数据库,我们可以看到最受欢迎的颜色是白色,最老的车龄是 17 年,但是如果仅通过查看其他值就可以预测汽车是否具有 AutoPass,该怎么办? 这就是机器学习的目的!分析数据并预测结果! 在机器学习中,通常使用非常大的数据集。在本教程中,我们会尝试让您尽可能容易地理解机器学习的不同概念,并将使用一些易于理解的小型数据集。 数据类型 如需分析数据,了解我们要处理的数据类型非常重要。 我们可以将数据类型分为三种主要类别: 数值(Numerical) 分类(Categorical) 序数(Ordinal) 数值数据是数字,可以分为两种数值
资源推荐
资源详情
资源评论
收起资源包目录
Apache TVM 是一个开放源代码的机器学习编译器框架,用于 CPU,GPU 和机器学习加速器.zip (2000个子文件)
graph_executor.c 41KB
crt_runtime_api.c 22KB
load_json.c 14KB
main.c 13KB
silence.c 12KB
unknown.c 12KB
no.c 12KB
yes.c 12KB
page_allocator.c 12KB
ai_runtime_api.c 10KB
graph_executor_module.c 8KB
aot_executor.c 8KB
aot_executor_module.c 7KB
func_registry.c 6KB
ndarray.c 5KB
stack_allocator.c 4KB
packed_func.c 4KB
runtime.c 3KB
platform-template.c 3KB
crt_backend_api.c 2KB
tvm_ethosu_runtime.c 2KB
aprofile_extra_support_routines.c 1KB
array_test.go 14KB
function.go 12KB
value.go 12KB
ndarray.go 12KB
function_test.go 9KB
value_test.go 7KB
complex.go 5KB
module.go 4KB
device.go 4KB
bytearray.go 3KB
module_test.go 3KB
simple.go 2KB
type.go 2KB
pack_func_closure_return.go 2KB
pack_func_closure_arg.go 2KB
pack_func_register.go 2KB
pack_func_handle_arg.go 2KB
pack_func_convert.go 2KB
bytearray_test.go 1KB
error.go 1KB
gotvm.go 1KB
gotvm_test.go 1KB
error_test.go 1KB
utils.go 1KB
packed_func.h 78KB
transform.h 78KB
nn.h 67KB
stmt.h 55KB
map.h 52KB
pipeline_struct.h 46KB
schedule.h 43KB
transform_step.h 41KB
op.h 40KB
pooling.h 36KB
expr.h 36KB
analysis.h 36KB
primitive.h 36KB
schedule.h 35KB
code_stack.h 34KB
object.h 34KB
doc.h 32KB
dataflow_pattern.h 31KB
pattern_utils.h 31KB
expr.h 31KB
pattern_match.h 30KB
nn.h 30KB
graph.h 29KB
array.h 29KB
utils.h 28KB
dnnl_tensor_requisite.h 28KB
attrs.h 28KB
transform.h 27KB
builtin.h 27KB
expr.h 26KB
logging.h 26KB
nn.h 26KB
c_runtime_api.h 25KB
analyzer.h 25KB
operation.h 25KB
utils.h 25KB
transform.h 25KB
transform.h 25KB
codegen_llvm.h 25KB
minrpc_server.h 24KB
expr.h 24KB
ir_builder.h 24KB
op_common.h 24KB
control_flow_graph.h 23KB
expr_functor.h 23KB
base_codegen.h 23KB
transform.h 23KB
reduction.h 23KB
analysis.h 22KB
database.h 22KB
profiling.h 22KB
nested_msg.h 21KB
stmt_functor.h 21KB
utils.h 21KB
共 2000 条
- 1
- 2
- 3
- 4
- 5
- 6
- 20
资源评论
野生的狒狒
- 粉丝: 2449
- 资源: 2131
下载权益
C知道特权
VIP文章
课程特权
开通VIP
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
最新资源
- Python 程序语言设计模式思路-行为型模式:策略模式:将算法封装成独立的类,并使它们可以互相替换及支付模式数据压缩
- main.py
- Last Loaded Test.DBK
- Screenshot_20240520_163011.jpg
- ubuntu-python3-whisper-tornado docker镜像 Dockerfile
- ubuntu-python3-whisper-tornado docker镜像07
- 新录音 8.m4a
- ubuntu-python3-whisper-tornado docker镜像
- ubuntu-python3-whisper-tornado docker镜像
- ubuntu-python3-whisper-tornado docker镜像09
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功