Work by Ben Hilburn, Philip Balister, and Doug Geiger focused on using the ARM cycle counter for profiling VOLK kernels. Ben wrote a patch that lets you use the ARM cycle counter from within the VOLK profiler. This is working, and you can use it immediately using this patch here:
That said, it is not immediately useful because the VOLK profiler runs operations hundreds / thousands of times to amortize overheads. That is great for measuring performance in terms of time, but makes measuring cycles nearly useless.
Ben will continue developing the VOLK profiler to make the ARM cycle counter more useful. There are a few commits on the plane, but there hasn't been the chance to finish the work up yet. This development is taking place here: