anirudhacharya commented on issue #13447: Rewrite dataloader, improves responsiveness and reliability URL: https://github.com/apache/incubator-mxnet/pull/13447#issuecomment-445430380 > Dataloader with threadpool is faster than Dataloader with multiprocess(about 2~5x), especially when the dataset is small or most of codes are written with C++, for example, rotating the image is written by Opencv library what is the basis for this comment. If you have multiple CPU cores isn't multiprocessing more efficient, has easier memory management and bypasses GIL?
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
