pyarrow.cuda.read_record_batch

pyarrow.cuda.read_record_batch()

Construct RecordBatch referencing IPC message located on CUDA device.

While the metadata is copied to host memory for deserialization, the record batch data remains on the device.

Parameters:

• buffer – Device buffer containing the complete IPC message
• schema (Schema) – The schema for the record batch
• pool (MemoryPool (optional)) – Pool to allocate metadata from

Returns:

batch (RecordBatch) – Reconstructed record batch, with device pointers