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
Core Performance Principles and Key Metrics
- Analyzing latency, throughput, energy consumption, and resource efficiency
- Distinguishing between system-wide and model-specific constraints
- Profiling strategies for inference versus training workloads
Performance Analysis on Huawei Ascend
- Leveraging CANN Profiler and MindInsight tools
- Diagnosing kernel and operator performance issues
- Managing offload patterns and memory mapping strategies
Performance Analysis on Biren GPU
- Utilizing Biren SDK monitoring capabilities
- Optimizing kernel fusion, memory alignment, and execution queues
- Conducting power and temperature-conscious profiling
Performance Analysis on Cambricon MLU
- Employing BANGPy and Neuware performance utilities
- Gaining kernel-level insights and interpreting performance logs
- Integrating the MLU profiler with various deployment frameworks
Graph and Model Architecture Optimization
- Implementing graph pruning and quantization methodologies
- Executing operator fusion and restructuring computational graphs
- Standardizing input dimensions and adjusting batch sizes
Memory and Kernel Refinement
- Improving memory layout and data reuse patterns
- Managing buffers efficiently across different chipsets
- Applying platform-specific kernel tuning methods
Best Practices Across Multiple Platforms
- Achieving performance portability through abstraction techniques
- Developing unified tuning pipelines for multi-chip setups
- Case study: Optimizing an object detection model across Ascend, Biren, and MLU
Concluding Remarks and Future Directions
Requirements
- Practical experience managing AI model training or deployment workflows
- Solid grasp of GPU/MLU computational concepts and model refinement techniques
- Familiarity with standard performance profiling instruments and key metrics
Target Participants
- Performance Engineering Specialists
- Machine Learning Infrastructure Teams
- AI System Architects
