S
SanchayanMaity
Hello,
I am using ADV7180 as a video decoder and using it to view the four analog channels.
I have to display four channels at the same time in four different windows. It's ok if the frame speed is low and not 25fps for the four windows. Can be 5 - 10 fps. Right now, even with two windows, it is not acceptable. I am using OpenCV for capturing and rendering.
I am using ioctl call to the adv driver to switch the channel. Though ultimately this is one byte, i think may be i can improve the timing and frame rate by circumventing the icotl call.
I read about mmap implementation in chapter 15 of LDD and was thinking of using it, but, i am not sure if this will really give me a performance improvement. Any thoughts on this?
Also, if i implement mmap, how should i synchronise the user and kernel space? Kernel space can inform user space by sending a signal and user space can write to the shared memory, but, how do i inform the kernel space to now read the memory and do this cycle vice versa?
Continue reading...
I am using ADV7180 as a video decoder and using it to view the four analog channels.
I have to display four channels at the same time in four different windows. It's ok if the frame speed is low and not 25fps for the four windows. Can be 5 - 10 fps. Right now, even with two windows, it is not acceptable. I am using OpenCV for capturing and rendering.
I am using ioctl call to the adv driver to switch the channel. Though ultimately this is one byte, i think may be i can improve the timing and frame rate by circumventing the icotl call.
I read about mmap implementation in chapter 15 of LDD and was thinking of using it, but, i am not sure if this will really give me a performance improvement. Any thoughts on this?
Also, if i implement mmap, how should i synchronise the user and kernel space? Kernel space can inform user space by sending a signal and user space can write to the shared memory, but, how do i inform the kernel space to now read the memory and do this cycle vice versa?
Continue reading...
