Running CNN inference in parallel on a phone's CPU, GPU, and NPU can deliver up to 2x the image throughput of the fastest single component, on the two SoCs tested.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Neural Network Inference on Mobile SoCs
Running CNN inference in parallel on a phone's CPU, GPU, and NPU can deliver up to 2x the image throughput of the fastest single component, on the two SoCs tested.