Anurag Ranjan

Thanks @cgo . I can start working on it, but if you already have something, could you send a pull request?

We now have a bleeding edge version that uses PIL instead of skimage. It is available on this branch. https://github.com/anuragranj/cc/tree/pil Please let me know if this works, or if there...

Yes, I am keeping this open until we are sure that this works. Could you try the pil branch and see if there are any problems?

To my knowledge, it worked at the time.

@EderSantana I got a similar OOM error ``` Error allocating 74400000 bytes of device memory (out of memory). Driver report 34045952 bytes free and 12079136768 bytes total ``` So, it...

@ClementPinard What is the cuda, ubuntu, and gcc version on which the test passes? It seems this is related to age old compatibility issues between nvcc and gcc. https://devtalk.nvidia.com/default/topic/1028112/cuda-setup-and-installation/nvcc-bug-related-to-gcc-6-lt-tuple-gt-header-/

I tried to set up with Ubuntu 18.04, cuda 9.0 and gcc 6. The problem is nvcc 9.0 is not compatible with gcc>5. And if I compile your package with...

I din't try the https://github.com/pytorch/extension-cpp on cuda 9.0., and will avoid going back to it. However, to give you a sense of the bug, this is what happens when I...

May be, we should add it to readme. The following works/ doesnt work with pytorch 0.4: Works: `gcc-5` `cuda-9.0` Works: `gcc-7` `cuda-9.2` Doesnt work: `gcc-7` `cuda-9.0` If you have both...

Yes, this sounds like reasonable fix. Although, I don't have sudo access on my gpu machines. :D