| 1 |
And Victor just announce ntl 9.0 on sage-devel: |
| 2 |
|
| 3 |
With much trepidation, I have introduced a (hopefully minor) |
| 4 |
backward incompatibility into NTL. |
| 5 |
|
| 6 |
The interface to the single-precision modular arithmetic |
| 7 |
routines has been modified slightly. |
| 8 |
This interface change allows for more flexible and more |
| 9 |
efficient implementation of these routines, |
| 10 |
which play a crucial role at many levels in NTL. |
| 11 |
|
| 12 |
Basically, these changes to the interface abstract away |
| 13 |
some implementation details that arguably should never been there |
| 14 |
in the first place. |
| 15 |
By coding to the new interface, NTL clients will be able to |
| 16 |
benefit from the current and future improvements. |
| 17 |
|
| 18 |
In particular, on 64-bit x86/GCC platforms, single precision |
| 19 |
moduli can now be up to 60 bits, rather than 50 bits. |
| 20 |
While some operations may in fact be a little slower, the most important |
| 21 |
ones (like MulModPrecon) should not be. |
| 22 |
Using larger moduli speeds up a number of things, like ZZ_pX |
| 23 |
arithmetic, as fewer primes need to be used in Chinese Remaindering steps. |
| 24 |
Other applications benefit from larger moduli as well. |
| 25 |
|
| 26 |
It is expected that most NTL clients will not be affected at all. |
| 27 |
Moreover, any code that needs to be updated will be detected |
| 28 |
by the compiler, and the updates should be simple and mechanical. |
| 29 |
There is also a configuration flag that will enable the legacy |
| 30 |
interface (although this is not recommended practice). |
| 31 |
|
| 32 |
For more, go to http://www.shoup.net/ntl |
Archives