Talk:cpp/numeric/random/mersenne twister engine/discard
From cppreference.com
< Talk:cpp | numeric/random
Wouldn't it be nice, to actually document which engines support "fast jump"? In parallel calculations, where one needs multiple instances of an engine spaced as far as possible (2^(internal bits)/N) it would be good to know up front which engines support jumping 2^67 in their sequence without actual generation.
- Yes it would be, but it's not requires by the standard, so someone has to look at each discard in each popular library (GNU, LLVM, Microsoft). e.g. GNU libstdc++ mersenne_twister_engine::discard, LLVM libc++ mesenne_twister_engine::discard.. If someone were to do that.. it would make a nice addition to the Notes. --Cubbi (talk) 07:35, 10 January 2017 (PST)
