Contents

• 1. NVIDIA Ampere GPU Architecture Tuning Guide
  • 1.1. NVIDIA Ampere GPU Architecture
  • 1.2. CUDA Best Practices
  • 1.3. Application Compatibility
  • 1.4. NVIDIA Ampere GPU Architecture Tuning
    • 1.4.1. Streaming Multiprocessor
      • 1.4.1.1. Occupancy
      • 1.4.1.2. Asynchronous Data Copy from Global Memory to Shared Memory
      • 1.4.1.3. Hardware Acceleration for Split Arrive/Wait Barrier
      • 1.4.1.4. Warp level support for Reduction Operations
      • 1.4.1.5. Improved Tensor Core Operations
      • 1.4.1.6. Improved FP32 throughput
    • 1.4.2. Memory System
      • 1.4.2.1. Increased Memory Capacity and High Bandwidth Memory
      • 1.4.2.2. Increased L2 capacity and L2 Residency Controls
      • 1.4.2.3. Unified Shared Memory/L1/Texture Cache
    • 1.4.3. Third Generation NVLink
• 2. Revision History
• 3. Notices
  • 3.1. Notice
  • 3.2. OpenCL
  • 3.3. Trademarks