mirror of
https://github.com/reactos/reactos.git
synced 2024-10-22 05:46:19 +00:00
- switch license to LGPL
- fix a typo svn path=/trunk/; revision=39615
This commit is contained in:
parent
2e0c09ded9
commit
dced580965
|
@ -1,5 +1,5 @@
|
||||||
/*
|
/*
|
||||||
* COPYRIGHT: See COPYING in the top level directory
|
* COPYRIGHT: LGPL, see LGPL.txt in the top level directory
|
||||||
* PROJECT: ReactOS Win32 subsystem
|
* PROJECT: ReactOS Win32 subsystem
|
||||||
* PURPOSE: FLOATOBJ floating point emulation functions for x86
|
* PURPOSE: FLOATOBJ floating point emulation functions for x86
|
||||||
* FILE: subsystems/win32/win32k/objects/i386/floatobj.S
|
* FILE: subsystems/win32/win32k/objects/i386/floatobj.S
|
||||||
|
@ -33,7 +33,7 @@
|
||||||
* Optimization notes:
|
* Optimization notes:
|
||||||
*
|
*
|
||||||
* - shld is slow (4 cycles) and not pairable, mov + shl is faster
|
* - shld is slow (4 cycles) and not pairable, mov + shl is faster
|
||||||
* - esp is used, beacuse it's available earlier
|
* - esp is used, because it's available earlier
|
||||||
* - bsr is very slow on old cpus (up to 72 cycles on a p1) while being much
|
* - bsr is very slow on old cpus (up to 72 cycles on a p1) while being much
|
||||||
* faster on modern cpus (2-11 cycles). Workarounds using branch trees or
|
* faster on modern cpus (2-11 cycles). Workarounds using branch trees or
|
||||||
* table lookups are of no use nowadays.
|
* table lookups are of no use nowadays.
|
||||||
|
|
Loading…
Reference in a new issue