pyarrow.cuda.read_record_batch

pyarrow.cuda.read_record_batch(buffer, schema, *, DictionaryMemo dictionary_memo=None, pool=None)

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

schemaSchema

The schema for the record batch

dictionary_memoDictionaryMemo, optional

If message contains dictionaries, must pass a populated DictionaryMemo

poolMemoryPool (optional)

Pool to allocate metadata from

Returns
batchRecordBatch

Reconstructed record batch, with device pointers