You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In GpuJsonToStruct, after reading the input JSON using cudf::read_json, we need to convert the output table (consisting of many strings columns) into the final structs column with the desired schema. Currently, that step is done in Scala code, which can have a lot of overhead of unnecessarily copying the intermediate columns into the final output column. We can do that better in native C++ code, performing the conversion if needed while just moving the intermediate columns into the output column without any copying.
The text was updated successfully, but these errors were encountered:
In
GpuJsonToStruct
, after reading the input JSON usingcudf::read_json
, we need to convert the output table (consisting of many strings columns) into the final structs column with the desired schema. Currently, that step is done in Scala code, which can have a lot of overhead of unnecessarily copying the intermediate columns into the final output column. We can do that better in native C++ code, performing the conversion if needed while just moving the intermediate columns into the output column without any copying.The text was updated successfully, but these errors were encountered: