EI0012 Execution_Error_SDMA
Symptom
SDMA memory copy task occurred exception during operator execution from remote rank: [%s]. base information: [%s]. task information: [%s]. group information: [%s]
Possible Cause
- The input or output address from SDMA memory copy task has not been allocated, or the allocated memory is smaller than size of task, or the allocated memory has already been released.
- The link status has abnormal during the execution of the SDMA task.
Solution
- Check whether the size of the input or ouput address matched the data.
- Check whether amy process have exited abormally in advance or the communication was destroyed.
父主题: HCCL Errors