Thank you for sending your enquiry! One of our team members will contact you shortly.
Thank you for sending your booking! One of our team members will contact you shortly.
Duration 21 hours
Course Outline
Introduction to the Chinese AI GPU Ecosystem
- Benchmarking Huawei Ascend, Biren, and Cambricon MLUs
- Contrasting CUDA with CANN, Biren SDK, and BANGPy architectures
- Market trends and vendor ecosystem dynamics
Preparation for Migration
- Evaluating the current CUDA codebase
- Selecting target platforms and specific SDK versions
- Setting up toolchains and development environments
Techniques for Code Translation
- Migrating CUDA memory access patterns and kernel logic
- Aligning compute grid and thread models
- Exploring automated versus manual translation methods
Platform-Specific Execution
- Leveraging Huawei CANN operators and custom kernels
- Utilizing the Biren SDK conversion pipeline
- Reconstructing models using BANGPy (Cambricon)
Testing and Optimization Across Platforms
- Profiling execution efficiency on each target architecture
- Comparing memory tuning and parallel execution strategies
- Monitoring performance and iterative improvements
Administering Mixed GPU Environments
- Managing hybrid deployments across multiple architectures
- Implementing fallback strategies and device detection
- Creating abstraction layers for long-term code maintainability
Case Studies and Industry Best Practices
- Migrating vision and NLP models to Ascend or Cambricon
- Adapting inference pipelines on Biren clusters
- Resolving version discrepancies and API gaps
Conclusion and Future Directions
Requirements
- Proficiency in programming with CUDA or GPU-based applications
- Knowledge of GPU memory models and compute kernels
- Experience with AI model deployment or acceleration workflows
Target Audience
- GPU developers
- System architects
- Software porting specialists
