search

intel / cloud-native-ai-pipeline

AI cloud native pipeline for confidential and sustainable computing
https://intel.github.io/cloud-native-ai-pipeline/
Apache License 2.0
37 stars 15 forks source link

Long data conversion time when using QAT compression/decompression #127

Closed Hulongyin closed 1 year ago

Hulongyin commented 1 year ago

Converting data to ctypes takes too long (about 10ms) when using QAT for compression and decompression.