I am using STM32F427ZG and have a query about the CMSIS FIR Filter functions. We require highest possible speed in a 16 bit environment. I have been comparing speeds for the q15,q31 and float32 FIR filters. The STM32F427 has floating point hardware, but I had hoped the q15 implementation of the filters would be faster than the q31 or float32 due to the SIMD instructions. However I am finding that they are all very much the same speed, if anything the float 32 is slightly faster, This applies at all levels of optimisation. Can anyone advise me please?
View all questions in Keil forum