Late Breaking Result: FPGA-Based Emulation and Fault Injection for CNN Inference Accelerators
Fuente:
arXiv
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866908459812782080 |
|---|---|
| author | Masar, Filip Mrazek, Vojtech Sekanina, Lukas |
| author_facet | Masar, Filip Mrazek, Vojtech Sekanina, Lukas |
| contents | A new field programmable gate array (FPGA)-based emulation platform is proposed to accelerate fault tolerance analysis of inference accelerators of convolutional neural networks (CNN). For a given CNN model, hardware accelerator architecture, and FT analysis target, an FPGA-based CNN implementation is generated (with the help of the Tengine framework), and fault injection logic is added. In our first case study, we report how the classification accuracy drop depends on the faults injected into multipliers used in Multiply-and-Accumulate Units of NVDLA inference accelerator executing ResNet-18 CNN. The FT analysis emulated on Zynq UltraScale+ SoC is an order of magnitude faster than software emulation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_12818 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Late Breaking Result: FPGA-Based Emulation and Fault Injection for CNN Inference Accelerators Masar, Filip Mrazek, Vojtech Sekanina, Lukas Hardware Architecture A new field programmable gate array (FPGA)-based emulation platform is proposed to accelerate fault tolerance analysis of inference accelerators of convolutional neural networks (CNN). For a given CNN model, hardware accelerator architecture, and FT analysis target, an FPGA-based CNN implementation is generated (with the help of the Tengine framework), and fault injection logic is added. In our first case study, we report how the classification accuracy drop depends on the faults injected into multipliers used in Multiply-and-Accumulate Units of NVDLA inference accelerator executing ResNet-18 CNN. The FT analysis emulated on Zynq UltraScale+ SoC is an order of magnitude faster than software emulation. |
| title | Late Breaking Result: FPGA-Based Emulation and Fault Injection for CNN Inference Accelerators |
| topic | Hardware Architecture |
| url | https://arxiv.org/abs/2501.12818 |