We Use IQOO 11s (8gen2) and iqoo neo race 7 (8gen1+) , when we load the model ,it showss do now support DSP an GPU .However ,we need it to accelerate the model
try { Loger.i("rgbBitmapAsFloat == " + mModel.file.getAbsolutePath()) val builder = NeuralNetworkBuilder(App.getInstance()) .setDebugEnabled(false) .setModel(mModel.file) .setCpuFallbackEnabled(true) .setUnsignedPD(true) .setRuntimeOrder(NeuralNetwork.Runtime.DSP,NeuralNetwork.Runtime.GPU,NeuralNetwork.Runtime.CPU)
com.qualcomm.qti.snpe.SnpeError$IllegalArgsOrStateException: error_code : 73 error_message: None of the selected runtime targets are supported.
Dear developer,
You might need to add below module to your Mainifests.xml in APK project.
BR.
Wei
Thank you
DSP can be used ,but predict time stll very slow ,and the error like that:
vendor/qcom/proprietary/adsprpc/src/fastrpc_apps_user.c:2989: Error 0xd: open_shell failed for domain 3 search paths used are /dsp/, /vendor/dsp/, /vendor/dsp/xdsp/ (errno Permission denied)
how to add it in the app? like the doc says?
the model can be loaded in CPU or GPU.But in DSP,It shows that
Unable to create network! Cause: error_code=1002; error_message=Layer parameter value is invalid. error_code=1002; error_message=Layer parameter value is invalid. No backend could validate Op=Conv_0 Type=Conv2d error code=3110; error_component=Model Validation; line_no=769; thread_id=502275169456; error_component=Model Validation; line_no=276; thread_id=502620093616
Dear developer,
Looks like your model failed to run on device after loaded to SNPE.
You can try to execute your mode on snpe-net-run instead of APP to check model runable quickly.
BR.
Wei
Thank you!
Actually,before we make the Andriod App,we tested the models.For one ,we use yolov5s,and tested it.it cost about 6000us one frame with snpe-net-run . So ,we want to know that any dependency excpet the snpe-release.aar and platform-validator.aar?
We had export new Network,However,it didnot work.we also tried use sdk 29,but it alse doesn't work
This seems to be a complicated matter, I will consult with my instructor
Minesweeper