Hello,
Do IMX8 SOMs provide hardware acceleration of Opencv applications via multithreading and GPU acceleration? With opencv 3.0 and after, this is available via Transparent API, however there seems to be not documentation clarifying whether Transparent API is supported on IMX8M/IMX8.
DO any of the available IMX8 offerings support Opencl 1.2 FP?
Please assist in acceleration options for opencv on IMX8 platform. Thank you
Hi
Do IMX8 SOMs provide hardware acceleration of Opencv applications via multithreading and GPU acceleration?
With opencv 3.0 and after, this is available via Transparent API, however there seems to be not documentation clarifying whether Transparent API is supported on IMX8M/IMX8.
OpenCV built with Openembedded by default builds with TBB so internal functions which are programmed to use multithreading do use multithreading.
That is true for any i.MX 8 series based module.
However, OpenCV in Openembedded is built with OpenCL support disabled by default. You would have to rebuild it with the relevant packageconfig set.
PACKAGECONFIG_append_mx8qm = " opencl"
PACKAGECONFIG_append_mx8qxp = " opencl"
DO any of the available IMX8 offerings support Opencl 1.2 FP?
According to NXP’s datasheet the i.MX 8 and i.MX 8X series support OpenCL 1.2 Full Profile.
So the Apalis and Colbri modules based on those SoMs do support it.
Note that the i.MX8 M Mini series of SoMs does not support OpenCL.
Max