Hi,
Thanks for the SNPE - it is far superior in perfomance to straight caffe or tensorflow on Android.
Is there a plan to release arm64-v8a versions of the shared libraries?
When using ABI filters, we have to custom package the APK (with SNPE armeabi-v7a) in order to get this to work on target devices.
Thanks again.
Rex
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.
Thanks for using the Snapdragon Neural Processing Engine; the NPE was indeed made to use the Snapdragon acceleration to its fullest potential.
In our initial NPE release we offered the complete support for armeabi-v7a, which includes the runtime/scheduler and the optimized CPU backend (32-bit armeabi-v7a) for DLC model execution. We recognize the importance of simplifying our user's development cycle, and we will look into this matter. It is our policy to not discuss our roadmap in the forums; however, we will take your input into account.
Please subscribe to news from the Neural Processing Engine and check the Release Notes of future SDK versions.
I too would like to see a 64-bit release made available for Android. I struggle to understand why there is only a 32-bit variant to run on on shiny new Qualcomm 64-bit processors such as the 820 and 835. I would also like to see this work with clang vs gcc, as clang has been the default on Android for some time. Both of these factors causes issues trying to integrate this into an application that gets built inside the Android tree (CodeAurora based) vs outside using the NDK.
I too would like to see an arm64-v8a version of the SNPE libraries. I am trying to integrate SNPE with a 64-bit app running on an 820.
-rick
I was very disappointed when I saw no change in the release notes of SNPE 1.6.
I fail to understand why no one from Qualcomm is commenting on this. This is very frustrating ...
I am happy to see Android ARM 64-bit libraries in version 1.8 released today. :)
WOOT!!!