Android hardware VP9: only 1 spatial layer despite L3T3_KEY / multi-spatial SVC config?

[Pratim Mallick]

We're publishing VP9 SVC from Android (WebRTC m145) with multi-spatial modes like L3T3_KEY, but on hardware encoders (c2.exynos.vp9.encoder, c2.google.vp9.encoder, etc.) the outgoing stream appears to contain only 1 spatial layer, not 3.

Has anyone else seen Android hardware VP9 behave this way — accepting or being configured for multi-spatial SVC, but effectively sending a single spatial layer (temporal layers only)?

Also, how can we verify whether the underlying hardware encoder actually supports 3 spatial layers? MediaCodec doesn't seem to expose scalability modes directly.Is this expected for MediaCodec-based VP9 on Android, or device/implementation-specific?

Thanks
Pratim