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
The Chinese AI GPU Ecosystem at a Glance
- Comparative analysis of Huawei Ascend, Biren, and Cambricon MLU
- Contrasting CUDA with CANN, Biren SDK, and BANGPy models
- Market trends and vendor ecosystem developments
Preparing for the Migration Process
- Evaluating your existing CUDA codebase
- Selecting target platforms and appropriate SDK versions
- Installing toolchains and configuring the development environment
Techniques for Code Translation
- Adapting CUDA memory access patterns and kernel logic
- Mapping compute grid and thread models
- Evaluating automated versus manual translation approaches
Platform-Specific Implementation Strategies
- Leveraging Huawei CANN operators and custom kernels
- Utilizing the Biren SDK conversion pipeline
- Rebuilding models using BANGPy (Cambricon)
Cross-Platform Testing and Performance Tuning
- Profiling execution across each target platform
- Optimizing memory usage and comparing parallel execution methods
- Tracking performance metrics and iterating for improvement
Managing Mixed GPU Environments
- Implementing hybrid deployments across multiple architectures
- Developing fallback strategies and robust device detection
- Implementing abstraction layers to enhance code maintainability
Case Studies and Industry Best Practices
- Migrating vision/NLP models to Ascend or Cambricon
- Retrofitting inference pipelines on Biren clusters
- Managing version mismatches and API discrepancies
Key Takeaways and Recommended Next Steps
Requirements
- Practical experience in programming with CUDA or GPU-based applications
- Solid comprehension of GPU memory models and compute kernels
- Familiarity with AI model deployment or acceleration workflows
Target Audience
- GPU programmers
- System architects
- Migration specialists
