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.
Course Outline
Foundational Performance Concepts and Metrics
- Latency, throughput, power consumption, and resource utilization
- Distinguishing between system-level and model-level bottlenecks
- Profiling strategies for inference versus training
Profiling on Huawei Ascend
- Utilizing CANN Profiler and MindInsight
- Diagnosing kernels and operators
- Understanding offload patterns and memory mapping
Profiling on Biren GPU
- Leveraging Biren SDK performance monitoring capabilities
- Managing kernel fusion, memory alignment, and execution queues
- Implementing power and temperature-aware profiling
Profiling on Cambricon MLU
- Using BANGPy and Neuware performance tools
- Gaining kernel-level visibility and interpreting logs
- Integrating the MLU profiler with deployment frameworks
Graph and Model-Level Optimization
- Strategies for graph pruning and quantization
- Restructuring computational graphs through operator fusion
- Standardizing input sizes and tuning batch sizes
Memory and Kernel Optimization
- Optimizing memory layout and reuse
- Managing buffers efficiently across different chipsets
- Applying kernel-level tuning techniques specific to each platform
Cross-Platform Best Practices
- Achieving performance portability through abstraction strategies
- Developing shared tuning pipelines for multi-chip environments
- Case study: Tuning an object detection model across Ascend, Biren, and MLU
Summary and Next Steps
Requirements
- Hands-on experience with AI model training or deployment pipelines
- A solid understanding of GPU/MLU compute principles and model optimization
- Basic proficiency with performance profiling tools and key metrics
Target Audience
- Performance engineers
- Machine learning infrastructure teams
- AI system architects
21 Hours
