Forums - CPU Fallback option in snpe-net-run

2 posts / 0 new
Last post
CPU Fallback option in snpe-net-run
sithara.tresa
Join Date: 5 Jul 20
Posts: 1
Posted: Thu, 2021-07-01 00:17
Hi,
 
I had tried to run a model using snpe-net-run which has an unsupported layer on DSP ,but supported on CPU.
SNPE doc quotes  "setCPUFallbackMode(): When the CPU fallback functionality is enabled, layers in model that violates runtime constraints will run on CPU while the rest of non-violating layers will run on the chosen runtime processor"
Assuming the same for enable_cpu_fallback option in snpe-net-run ,
1. Is it is expected to run unsupported layer on CPU and rest layers on DSP?The following command is used :
./snpe-net-run_1.51 --container dlc/xyz.dlc --input_list data/raw_list.txt --set_unconsumed_as_output --use_dsp --enable_cpu_fallback
 
2. If yes,I have also used snpe-diagview tool to get the information of timing and runtime of model inference on board,but the runtime for every layers was CPU.
 
Kindly guide us to resolve this.
 
Thanks,
Sithara Tresa Chacko
  • Up0
  • Down0
ap.arunraj
Join Date: 20 Apr 20
Posts: 21
Posted: Tue, 2021-07-13 04:21

Hello Sithara Tresa Chacko,

Did you use quantized model? If yes, is it throwing any error without CPU Fallback option enabled.

  • Up0
  • Down0
or Register

Opinions expressed in the content posted here are the personal opinions of the original authors, and do not necessarily reflect those of Qualcomm Incorporated or its subsidiaries (“Qualcomm”). The content is provided for informational purposes only and is not meant to be an endorsement or representation by Qualcomm or any other party. This site may also provide links or references to non-Qualcomm sites and resources. Qualcomm makes no representations, warranties, or other commitments whatsoever about any non-Qualcomm sites or third-party resources that may be referenced, accessible from, or linked to this site.