SR_Mobile_Quantization icon indicating copy to clipboard operation
SR_Mobile_Quantization copied to clipboard

Published 20 hours ago • verified publisher icon NJU-Jet

running time on AI Benchmark App

Open caijie0620 opened this issue 3 years ago • 3 comments

Hello: I tested 'base7_D4C28_bs16ps64_lr1e-3_qat_time.tflite' running time via AI Benchmark App. My device is Snapdragon 888 and the device's AI score is 54.4. It takes about 200ms NNAPI. In the paper, your device is Snapdragon 820 and it takes ~30ms. Do you have any idea about the running time?

Thanks so much.

caijie0620 avatar Jun 03 '21 22:06 caijie0620

It seems that you device doesn't use NNAPI to acclerate the model.

NJU-Jet avatar Jun 04 '21 05:06 NJU-Jet

My device is Snapdragon 835. It takes about 700ms NNAPI.

hacktmz avatar Dec 01 '21 09:12 hacktmz

It seems that you device doesn't use NNAPI to acclerate the model.

I enter the PRO Model and select the Custom Model. When I click the run button, the interface turns to the initial page and the operation information of the selected model is not displayed. What's the matter?

douzaikongcheng avatar Apr 08 '22 13:04 douzaikongcheng