CNN Accelerator Internal Buffering for Pooling Data Reuse

The patent merges multiple pooling operations across different layers into a single unified pooling operation. Instead of performing pooling separately for each layer (which would require multiple external memory accesses), the accelerator performs convolution and pooling operations jointly across multiple layers, consolidating data transfers and reducing the frequency of external memory accesses while maintaining complete data processing.

The patent implements preliminary action by loading and storing intermediate results of pooling operations in internal buffers before final external memory access is needed. The accelerator performs pooling operations on intermediate data and stores results in internal buffers, allowing subsequent operations to access this pre-processed data without requiring additional external memory accesses, thus reducing overall energy consumption.

The patent merges the storage operations for intermediate results across multiple layers into a single consolidated external memory access. Instead of writing intermediate results to external memory after each layer (which would require frequent accesses), the accelerator consolidates these writes into fewer batched operations, reducing memory access frequency while maintaining data accuracy through proper buffer management.

The patent introduces internal buffers as intermediary storage between processing elements and external memory. These buffers act as mediators that temporarily hold intermediate results, allowing processing elements to access data quickly without requiring immediate external memory access. The buffers manage data transfer to external memory in optimized batches, reducing access frequency while preserving data integrity.