I think I did not understand you. What’s the point of using a high performance device to handle files?
The answer is no, you can not open nor read/write files from the GPU, the GPU has no access to the hard drive. But if what you want to do is data partitioning based on complex operations on the data contained in that file, you can always read it in a buffer, upload it to video memory and perform those operations on the GPU (if parallelizable).
If you just want to split the file in smaller chunks what you need is a fast hard drive, not computational power.
For out of core algorithm, it seems they’re planning DMA to device in future release. However, I’m not sure about the hard drive part.