Dear all,
I recently learned about the Advanced SIMD/NEON extension on ARM processors. Although compilers can do a great job in generating NEON instructions, I started with hand-written assembly, so that I can get to know the possibilities of this technology. I managed to write a few simple audio processing routines, using NEON instructions.
Using Instruments' Time Profiler, I can get some insight in the efficiency of my code. However, I would also like to use the Counters instrument to get event-based information, such as pipeline stalls. The counters instruments seems unavailable for iOS ("The current OS does not have support for your CPU"), even though the Cortex-A9 has a Performance Measurement Unit (PMU).
Is there any way to access this Performance Measurement Unit, or are there any other tools available that can give me more insight in the timing aspect of my routines (pipeline stall, waiting for load/store, number of cycles spent in the routine, ...)?
I'm using Xcode 7.0 beta 4 on OS X Yosemite (10.10.4), and iPhone 4S running iOS 8.4
Thanks in advance,
Maarten