Sending large images to the GPU takes abnormally large time

Hi guys,

I am using CNN model(AlexNet) for Image processing. and I am using Torch(lua). I am modifying the Torch starter code which is this -> https://github.com/cvondrick/torch-starter. My problem is that I am making images with 18 channels instead of 3 channels for training the model , and for sending those images to GPU, it takes around 20 (2.13 s for every batch) times more than when it sends images with three channels (0.14s for every batch ) . I also tried to see how long does it take to send images with 4 channels to GPU, and I saw that as soon as the number of channels increased to more than 3 channels, consequently the time increased a lot, like 20 times. For example, even for images with 4 channels , it took around 2s for every batch, which is around 19 times more than running 3 channel images. I was wondering if there is a bug which makes it take this much time and if there aren’t any bugs, is there any way I can decrease this running time?

Thank you so much in advance!

I’m curious… why do you need so many channels and what do these channels represent?

I’m curious… why do you need so many channels and what do these channels represent?