GPU Programming on Biren AI Accelerators

• An understanding of GPU architecture and parallel processing
• Experience with CUDA, OpenCL, or similar GPU programming environments
• Familiarity with deep learning frameworks such as PyTorch or TensorFlow

Audience

• HPC developers
• AI infrastructure engineers
• Performance optimization specialists

Biren AI Accelerators are high-performance GPUs designed for AI and HPC workloads with support for large-scale training and inference.

This instructor-led, live training (online or onsite) is aimed at intermediate-level to advanced-level developers who wish to program and optimize applications using Biren’s proprietary GPU stack, with practical comparisons to CUDA-based environments.

By the end of this training, participants will be able to:

• Understand Biren GPU architecture and memory hierarchy.
• Set up the development environment and use Biren’s programming model.
• Translate and optimize CUDA-style code for Biren platforms.
• Apply performance tuning and debugging techniques.

Format of the Course

• Interactive lecture and discussion.
• Hands-on use of Biren SDK in sample GPU workloads.
• Guided exercises focused on porting and performance tuning.

Course Customization Options

• To request a customized training for this course based on your application stack or integration needs, please contact us to arrange.

Introduction to Biren GPU Architecture

• Biren overview and use cases
• Hardware layout: cores, memory, compute clusters
• Comparison with NVIDIA and AMD GPUs

Setting Up the Biren Programming Environment

• Installing Biren SDK and runtime
• Understanding the toolchain and compiler model
• Basic project structure and build process

GPU Programming with the Biren Stack

• Thread and block models
• Memory management and data transfers
• Kernel development and launch patterns

Porting from CUDA to Biren

• Translation techniques for CUDA code
• Common API mappings and adaptations
• Code conversion labs and practice

Debugging and Profiling

• Using Biren’s debugger and profiler
• Identifying bottlenecks
• Memory access patterns and optimization

Optimization Techniques

• Thread scheduling and instruction pipelining
• Loop unrolling and shared memory use
• Advanced kernel tuning for throughput

Case Study and Application Examples

• Training a model with Biren accelerators
• Porting and profiling a vision or NLP model
• Comparing performance vs CUDA/NVIDIA

Summary and Next Steps