Contents

• 1. NVIDIA Hopper Tuning Guide
  • 1.1. NVIDIA Hopper GPU Architecture
  • 1.2. CUDA Best Practices
  • 1.3. Application Compatibility
  • 1.4. NVIDIA Hopper Tuning
    • 1.4.1. Streaming Multiprocessor
      • 1.4.1.1. Occupancy
      • 1.4.1.2. Tensor Memory Accelerator
      • 1.4.1.3. Thread Block Clusters
      • 1.4.1.4. Improved FP32 Throughput
      • 1.4.1.5. Dynamic Programming Instructions
    • 1.4.2. Memory System
      • 1.4.2.1. High-Bandwidth Memory HBM3 Subsystem
      • 1.4.2.2. Increased L2 Capacity
      • 1.4.2.3. Inline Compression
      • 1.4.2.4. Unified Shared Memory/L1/Texture Cache
    • 1.4.3. Fourth-Generation NVLink
• 2. Revision History
• 3. Notices
  • 3.1. Notice
  • 3.2. OpenCL
  • 3.3. Trademarks