alibaba / MNN - Conditional Complexity

Intro

Conditional complexity (also called cyclomatic complexity) is a term used to measure the complexity of software. The term refers to the number of possible paths through a program function. A higher value ofter means higher maintenance and testing costs (infosecinstitute.com).
Conditional complexity is calculated by counting all conditions in the program that can affect the execution path (e.g. if statement, loops, switches, and/or operators, try and catch blocks...).
Conditional complexity is measured at the unit level (methods, functions...).
Units are classified in four categories based on the measured McCabe index: 1-5 (simple units), 6-10 (medium complex units), 11-25 (complex units), 26+ (very complex units).

Learn more...

Conditional Complexity Overall

There are 10,211 units with 219,845 lines of code in units (41.4% of code).

40 very complex units (16,842 lines of code)
139 complex units (22,600 lines of code)
690 medium complex units (53,011 lines of code)
1,056 simple units (42,196 lines of code)
8,286 very simple units (85,196 lines of code)

Legend:

51+

26-50

11-25

6-10

1-5

Alternative Visuals

3D view of all units

Conditional Complexity per Extension

51+

26-50

11-25

6-10

1-5

Conditional Complexity per Logical Component

primary logical decomposition

51+

26-50

11-25

6-10

1-5

Most Complex Units

Top 50 most complex units

Unit	# lines	McCabe index	# params
LabelT LabelingGrana() in tools/cv/source/imgproc/structural.cpp	1129	196	4
void _AVX512_MNNGemmInt8AddBiasScale_16x4_Unit_VNNI() in source/backend/cpu/x86_x64/avx512/GemmInt8_VNNI.cpp	1367	157	8
inl void MATMULCOREFUNC_NAME() in source/backend/cpu/x86_x64/avx512/Matmul_4_4_64.inl	1368	157	8
void _AVX512_MNNGemmInt8AddBiasScale_16x4_w4_Unit_VNNI() in source/backend/cpu/x86_x64/avx512/GemmInt8_VNNI.cpp	1198	139	8
inl void MATMULCOREFUNC_NAME_W4() in source/backend/cpu/x86_x64/avx512/Matmul_4_4_64.inl	1199	139	8
std::string CUDATarget::codegen() in codegen/cuda/CUDATarget.cpp	623	132	3
void _AVX512_MNNPackedMatMulO16FullLoadKernel() in source/backend/cpu/x86_x64/avx512/Gemm31_16.h	200	131	6
def mha_mhca_detected() in tools/converter/source/onnx/onnx_model_graph_opt.py	222	121	3
ErrorCode DenseConvInt8TiledExecutor::onExecute() in source/backend/cpu/compute/ConvInt8TiledExecutor.cpp	521	112	2
void _AVX512_MNNPackedMatMulO32SwapedKernel() in source/backend/cpu/x86_x64/avx512/Gemm10_32.h	235	92	6
virtual bool onExecute() in tools/converter/source/optimizer/postconvert/RemoveInvalidCast.cpp	196	88	1
void _AVX_MNNGemmInt8AddBiasScale_16x4_Unit() in source/backend/cpu/x86_x64/avx/GemmInt8.cpp	540	81	8
void _AVX512_MNNPackedMatMulO32Swaped6Kernel() in source/backend/cpu/x86_x64/avx512/Gemm10_32.h	209	80	6
int main() in tools/cpp/ModuleBasic.cpp	364	79	2
virtual EXPRP onExecute() in tools/converter/source/optimizer/onnxextra/OnnxEinsum.cpp	346	78	1
static FormatSetType _getFormatType() in tools/converter/source/optimizer/postconvert/AddTensorFormatConverter.cpp	96	75	2
void _AVX_MNNGemmInt8AddBiasScale_16x4_w4() in source/backend/cpu/x86_x64/avx/GemmInt8.cpp	457	73	8
std::vector getGemmParams() in source/backend/opencl/core/OpenCLGemmTune.cpp	326	72	5
ErrorCode GeometryComputerUtils::shapeComputeAndGeometryTransform() in source/geometry/GeometryComputerUtils.cpp	272	71	8
ErrorCode DenseConvInt8TiledExecutor::onResize() in source/backend/cpu/compute/ConvInt8TiledExecutor.cpp	273	70	2
void _AVX512_MNNPackedMatMulO32FullLoadKernel() in source/backend/cpu/x86_x64/avx512/Gemm10_32.h	185	68	6
ErrorCode CoreMLBinary::onResize() in source/backend/coreml/execution/CoreMLBinary.cpp	210	67	2
virtual bool onExecute() in tools/converter/source/optimizer/postconvert/AddTensorFormatConverter.cpp	259	66	1
std::shared_ptr ConvolutionCommon::load() in source/core/ConvolutionCommon.cpp	233	66	5
virtual EXPRP onExecute() in tools/converter/source/optimizer/onnxextra/OnnxConvolutionMerge.cpp	250	65	1
virtual bool onCompute() in source/geometry/GeometryStridedSlice.cpp	291	65	5
ErrorCode CoreMLUnary::onResize() in source/backend/coreml/execution/CoreMLUnary.cpp	230	63	2
def configure_extension_build() in pymnn/pip_package/setup.py	236	62	0
void _AVX512_MNNPackedMatMulO48Swaped2Kernel() in source/backend/cpu/x86_x64/avx512/Gemm9_48.h	228	62	6
void MNNPackedSparseQuantMatMulEpx4() in source/backend/cpu/compute/Int8FunctionsOpt.cpp	723	62	7
virtual ErrorCode onExecute() in source/backend/cpu/CPURaster.cpp	305	60	2
int main() in tools/train/source/exec/transformerExecution.cpp	292	58	2
ErrorCode Pipeline::encode() in source/core/Pipeline.cpp	216	58	2
void _AVX512_MNNPackedMatMulO48Swaped4Kernel() in source/backend/cpu/x86_x64/avx512/Gemm9_48.h	233	56	6
def export_tokenizer() in transformers/llm/export/llmexport.py	149	55	1
virtual bool onComputeSize() in source/shape/ShapeStridedSlice.cpp	238	54	3
ErrorCode MetalConvolution1x1::onResize() in source/backend/metal/MetalConvolution1x1.mm	279	54	2
static int test_main() in tools/cpp/MNNV2Basic.cpp	321	52	2
ErrorCode ConvolutionPackFreeWinograd::onExecute() in source/backend/cpu/compute/ConvolutionPackFreeWinograd.cpp	459	51	2
ErrorCode ConvBufExecution::onResize() in source/backend/opencl/execution/buffer/ConvBufExecution.cpp	364	51	2
bool Cli::initializeMNNConvertArgs() in tools/converter/source/common/cli.cpp	358	50	3
void _AVX512_MNNPackedSparseMatMulEpx4() in source/backend/cpu/x86_x64/avx512/SparseKernelFunctionEpx4.cpp	608	50	9
void _AVX512_MNNPackedSparseMatMulEpx8() in source/backend/cpu/x86_x64/avx512/SparseKernelFunctionEpx8.cpp	769	50	9
void _AVX512_MNNPackedMatMulO48FullLoadKernel() in source/backend/cpu/x86_x64/avx512/Gemm9_48.h	190	50	6
void MNNPackedSparseMatMulEpx4() in source/backend/cpu/compute/CommonOptFunction.cpp	567	48	9
void WeightQuantAndCoding() in tools/converter/source/common/WeightQuantAndCoding.cpp	146	47	3
bool fuseConstIntoSubgraph() in tools/converter/source/optimizer/PostConverter.cpp	199	47	2
int main() in tools/cpp/SequenceModuleTest.cpp	281	47	2
ErrorCode CPURaster::onResize() in source/backend/cpu/CPURaster.cpp	208	47	2
ErrorCode CPUTensorConverter::convert() in source/backend/cpu/CPUTensorConvert.cpp	192	47	11