search

DeepRec-AI / DeepRec

DeepRec is a high-performance recommendation deep learning framework based on TensorFlow. It is hosted in incubation in LF AI & Data Foundation.
Apache License 2.0
1.03k stars 349 forks source link

the performance in a100 is slower than the nvidia-tensorflow #105

Closed thinking03 closed 2 years ago

thinking03 commented 2 years ago

env: A100 8gpu + horovod framework: DeepRec1.15.5 vs nvidia-tensorflow 1.15.4 model: transformer + mmoe result: DeepRec 1.8s/step, nv-tf:0.55s/step

liutongxuan commented 2 years ago

Duplicate issue as https://github.com/alibaba/DeepRec/issues/110