cudaStreamCreateWithPriority on TK1

Does compute 3.2 (the TK1) support cudaStreamCreateWithPriority functionality?
The API guide says 3.5 only, Tesla and Quadro. I know Dynamic parallelism is not supported, but is this also missing from 3.2? If so, is this something which needs hardware support, or just a feature which could be enabled in software?
The programming guide, even in CUDA 6.5, is still missing a definition of the compute 3.2 capabilities.

You can test for support by calling ‘cudaDeviceGetStreamPriorityRange()’ and inspecting the ‘leastPriority’ and ‘greatestPriority’ values.

Note that concurrent kernel support is currently broken in the CUDA Toolkit for L4T 19.2.

If stream priorities aren’t supported now, I’m guessing that once concurrent kernel support is fixed (L4T 21.1?) we’ll find out if stream priorities are supported.