AML-08 - Mastering Altera® AI Suite - 2 Days - Enroll Now This course provides comprehensive, hands-on expertise with the Altera® FPGA AI Suite — the end-to-end toolchain for generating AI accelerators targeting specific latency, throughput, and power requirements on Agilex® FPGA devices. Custom models using the OpenVINO™ Model Optimizer and custom extension framework is also explored. Course Content: 1. Altera® FPGA AI Suite component architecture 2. AI Suite supported model sources 3. End-to-end AI Suite workflow 4. AI Suite compiler input requirements 5. Supported operators 6. AI Suite GUI 7. Agilex device family overview 8. AI Suite example designs and reference flows 9. AI Suite Python API overview 10. Development environment setup 11. License management 12. Agilex® AI Tensor block internals 13. Compiler layer mapping 14. Layer fusion in the AI Suite compiler 15. Operator quantization in the AI Suite 16. INT8 PTQ calibration workflow 17. Per-layer quantization sensitivity 18. Mixed precision compilation 19. Structured pruning integration 20. AI Suite compiler directives 21. Compilation report analysis 22. Throughput and latency estimation 23. AI Suite compiler error messages and resolution 24. Deep learning inference IP core architecture 25. IP parameterization 26. Instantiating the deep learning inference IP in a Quartus® Prime Pro project: IP Catalog, generate HDL, add to project 27. Platform Designer integration 28. Memory interface requirements 29. HBM2e configuration for AI Suite IP: pseudo channel assignment, bandwidth allocation 30. DMA engine integration 31. Streaming interfaces for input and output data 32. Quartus Prime Pro compilation 33. Timing closure for inference IP 34. Resource utilization analysis Prerequisites: - FPGA Authority “FPGA Architecture and AI Acceleration Survey” course or equivalent Quartus Prime Pro and Altera FPGA experience - FPGA Authority “ML/AI Essentials” course or equivalent foundational machine learning knowledge - Proficiency in Python 3 scripting: virtual environments, pip, NumPy, and basic PyTorch or TensorFlow model loading - Working knowledge of Quartus Prime Pro: project creation, compilation, and basic Platform Designer system assembly - Familiarity with Linux command line: file operations, environment variables, executing shell scripts - Recommended: exposure to ONNX model format and OpenVINO toolkit Tools Required: - Quartus Prime Pro Edition - Altera FPGA AI Suite - Platform Designer - ModelSim-Altera FPGA Edition / Questa - Signal Tap Logic Analyzer - Python 3.11+ - PyTorch - TensorFlow / Keras - ONNX Runtime - CMake Course Code: AML-08. FA_AISUITE.