< :前の番号
^ :番号順リスト
> :次の番号
P :前の記事
N :次の記事(自分と同じ返事先を持つ)
|<:スレッドの先頭
>|:次のスレッド
^ :返事先
_:自分への返事
>:同じ返事先を持つ記事(前)
<:同じ返事先を持つ記事(後)
---:分割してスレッド表示、再表示
| :分割して(縦)スレッド表示、再表示
~ :スレッドのフレーム消去
.:インデックス
..:インデックスのインデックス
Hello,
In message "Re: bignums"
on Tue, 5 Sep 2006 01:20:06 +0900, Ondrej Bilka <neleai / seznam.cz> writes:
|
|I want ask how look integration of faster bignums.
|It was mentioned at this ml 2 years ago where it seemed that MBignum
|will be integrated but isn't.
|
|GMP: There is problem how replace bignum with it otherwise it will be
|slower with small numbers.
|And problem with license(Unofficial patch | GMP taints ruby :-)
|
|With GMP benchmark I discovered that multipling 2 fixnums into bignum is
|slow because it multiplies 2 temporary bignums. Without heavy magic its unsolvable.
|
|Addition and substraction use same technique. But because Fixnum has
|one byte less than long result will fit into long. So LONG2NUM would work too.
|
|I tried add Karatsuba multiplication to bignums and when debugging I
|discovered that rb_big_pow computing doesn't normalize temps. Its not problem
|with big numbers but computing power of 2 temporary variable has x these
|lengths:
|2**2 :1
|2**4 :3
|2**8 :7
|2**32 31
|)
Right. I will merge your patch. Thank you!
matz.