hi. i am writing an application on the imx8mm that needs to copy video frames. As for now i use memcpy. the cpu usage is not high but the latency creates drop packets. is there a way to compile my application with using the NEON to vectorize the copy. i tried to compile with neon flags but all fail. does the imx8mm support neon compilations.
thanks
Hello
I also need to use NEON to accelerate my c++ program on iMX8M mini. In my case, it's only need to include the noen header by
#include <arm_neon.h>
and compile it without -mfpu=neon, then the program works!
I use vmulq_f32 and vmlaq_f32 in the program, and the speed indeed runs faster, the answer is also correct.
Maybe this will help you.
Hello greeran,
The i.MX8M don't have NEON on its architecture, for use neon you will have to use the MX6Q.
Regards
hi
thanks for the reply but i am a little confused. from the imx8mm datasheet (https://www.nxp.com/docs/en/data-sheet/IMX8MMCEC.pdf) from the nxp site i see that the a53 comes with neon by default
"Media Processing Engine (MPE) with NEON technology supporting the Advanced
Single Instruction Multiple Data architecture:"
am i missing something
thanks
hi
thanks for the reply but i am a little confused. from the imx8mm datasheet (https://www.nxp.com/docs/en/data-sheet/IMX8MMCEC.pdf) from the nxp site i see that the a53 comes with neon by default
"Media Processing Engine (MPE) with NEON technology supporting the Advanced
Single Instruction Multiple Data architecture:"
am i missing something
thanks