Get in Touch
 Duration 21 hours

Course Outline

Introduction to Biren GPU Architecture

  • Overview of Biren technology and its primary use cases
  • Hardware configuration: cores, memory structure, and compute clusters
  • Comparative analysis with NVIDIA and AMD GPUs

Configuring the Biren Programming Environment

  • Installation of the Biren SDK and runtime components
  • Exploring the toolchain and compiler architecture
  • Standard project structures and build workflows

GPU Programming with the Biren Stack

  • Thread and block modeling concepts
  • Managing memory and handling data transfers
  • Developing kernels and optimizing launch patterns

Migrating from CUDA to Biren

  • Strategies for translating CUDA codebase
  • Mapping common APIs and necessary adaptations
  • Hands-on labs and practice for code conversion

Debugging and Profiling

  • Leveraging Biren's debugger and profiler tools
  • Pinpointing performance bottlenecks
  • Analyzing memory access patterns and applying optimizations

Optimization Techniques

  • Managing thread scheduling and instruction pipelining
  • Utilizing loop unrolling and shared memory effectively
  • Advanced kernel tuning to maximize throughput

Case Studies and Real-World Applications

  • Training models using Biren accelerators
  • Porting and profiling vision or NLP models
  • Performance benchmarking against CUDA/NVIDIA standards

Summary and Future Steps

Requirements

  • A solid grasp of GPU architecture and parallel processing concepts
  • Prior experience with CUDA, OpenCL, or comparable GPU programming environments
  • Proficiency with deep learning frameworks such as PyTorch or TensorFlow

Target Audience

  • HPC developers
  • AI infrastructure engineers
  • Specialists in performance optimization

Number of participants


Price per participant

Testimonials (2)

Provisional Upcoming Courses (Require 5+ participants)

Related Categories