Mill Computing, Inc. › Forums › The Mill › Architecture › Extending 32-bit float precision, and minimizing rounding error
Tagged: 32-bit float, error, fma, polynomial, precision, rounding
- AuthorPosts
- #4076 |
Does the Mill have any specialized facilities to help reduce rounding errors when using polynomial expressions, or exponential functions; or, are we simply expected to manually implement something like ‘error recovery’ using FMA instructions?
- AuthorPosts
You must be logged in to reply to this topic.