Skip to content

REPORT.md

Latest commit

 

History

History
122 lines (90 loc) · 3.52 KB

REPORT.md

File metadata and controls

122 lines (90 loc) · 3.52 KB

ExecuTorch AI Layer Build Report

Generated: 2026-02-26T16:48:09Z Git Commit: e313cbba50d5 on main Repository Status: 🔄 Modified Last Commit: 2026-01-28 13:39:17 +0100

📊 Build Summary

  • Libraries: 9 static libraries
  • Models: 1 model assets
  • Operators: 9 selected operators
  • Build Type: Release

📚 Library Assets

Total Size: 9.7 MiB

Library Size Percentage Modified Hash
libcortex_m_kernels.a 20.0 KiB 0.2% 2026-02-26 16:46:10 0f0d4153cee555cc
libcortex_m_ops_lib.a 12.7 KiB 0.1% 2026-02-26 16:46:10 48544da99226a894
libexecutorch.a 51.9 KiB 0.5% 2026-02-26 16:46:10 5322395c951790a3
libexecutorch_core.a 216.5 KiB 2.2% 2026-02-26 16:46:10 8f7469e33e108fb6
libexecutorch_delegate_ethos_u.a 18.6 KiB 0.2% 2026-02-26 16:46:10 e93e92a1a5e75e9f
libportable_kernels.a 9.1 MiB 94.0% 2026-02-26 16:46:10 c1c58c9a19838ae6
libportable_ops_lib.a 11.6 KiB 0.1% 2026-02-26 16:48:08 290fcc85b2f1293c
libquantized_kernels.a 238.6 KiB 2.4% 2026-02-26 16:46:10 105eb92cd0bdc77f
libquantized_ops_lib.a 28.7 KiB 0.3% 2026-02-26 16:46:10 9bb5169c6ef3624a

🤖 Model Assets

Asset Type Size Modified Hash
ethos_u_minimal_example.pte Model 3.8 KiB 2026-02-26 16:48:08 bd7a211160a18572

⚙️ Build Configuration

CMake Configuration

  • Build Type: Release
  • Toolchain File: /__w/CMSIS-Executorch/CMSIS-Executorch/model/arm-none-eabi-gcc.cmake
  • ARM Baremetal: ON
  • Cortex-M Support: ON
  • Portable Ops: ON
  • Quantized Kernels: ON

Selected Operators

Source: Model file: ethos_u_minimal_example.pte (inferred)

Count: 9 operators

aten::add.out
aten::conv2d.out
aten::linear.out
aten::mul.out
aten::permute_copy.out
aten::relu.out
aten::sigmoid.out
aten::softmax.out
aten::view_copy.out

🔄 Model Conversion Details

Ethos-U Compile Specification:

  • target: ethos-u55-128
  • system_config: Ethos_U55_High_End_Embedded
  • memory_mode: Shared_Sram
  • extra_flags: --output-format=raw, --debug-force-regor, --verbose-all

Quantization Configuration:

  • Using EthosUQuantizer with symmetric quantization
  • Post-training quantization enabled

Model Architecture:

  • Model class: Add

📦 Source Layer Export

Total Source Files: 226 files across 2 layers

Layer Description Groups Files
stage1 Generated from compile_commands.json (327 files) 4 37
stage2 Generated from compile_commands.json (240 files) 1 189

Group Details

STAGE1:

  • Runtime: 19 files
  • Schema: 1 files
  • Kernels/quantized: 11 files
  • Backends: 6 files

STAGE2:

  • Kernels: 189 files

🛠️ Build Environment

  • Platform: Linux 904960cb5fff 6.14.0-1017-azure #17~24.04.1-Ubuntu SMP Mon Dec 1 20:10:50 UTC 2025 x86_64 x86_64 x86_64 GNU/Linux
  • Python: Python 3.12.3
  • CMake: cmake version 4.2.3
  • ARM GCC: arm-none-eabi-gcc (Arm GNU Toolchain 13.3.Rel1 (Build arm-13.24)) 13.3.1 20240614

📁 Asset Locations

ai_layer/
├── engine/
│   ├── lib/           # Static libraries
│   ├── include/       # Header files
│   └── model/         # Model assets
└── REPORT.md          # This report

Report generated by ExecuTorch AI Layer build system at 2026-02-26T16:48:09Z