Hi, sDNA can handle that sort of volume, albeit working from link centers rather than a point data set. Given the number of points involved I'm guessing your spatial model is large so the difference won't matter to you? (Though if it does you can split links where you want your start/endpoints).
The public version won't output the data you require but the beta of our next version will - if you would like to try it feel free to get in touch with me via the website. www.cf.ac.uk/sdna