Fsub
From: '\\\\o//'annabee ('\\o//'annabee)
Date: 03/08/05
- Next message: '\\\\o//'annabee: "Re: Fsub"
- Previous message: '\\\\o//'annabee: "Re: War"
- Next in thread: '\\\\o//'annabee: "Re: Fsub"
- Reply: '\\\\o//'annabee: "Re: Fsub"
- Reply: Stepan Polovnikov: "Re: Fsub"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Tue, 08 Mar 2005 12:16:27 +0100
Hello.
Do anyone know what is the diffrence between
fld D§ebx + D3DVector.z
fld D§eax + D3DVector.z
fsubrp ST0 ST1
and
fld D§eax + D3DVector.z
fld D§ebx + D3DVector.z
fsubp ST0 ST1
The result is the same, but is one better than the other ?
What would be even better way ? Maybe this ?
fld D§eax + D3DVector.x
fsub D§ebx + D3DVector.y
My test seem to indicate that this is slower:
fld D§eax + D3DVector.x
fsub D§ebx + D3DVector.y
And that this is slighly faster, and the same as the subp one
fld D§ebx + D3DVector.y
fld D§eax + D3DVector.y
fsubrp ST0 ST1
I am wondering how do do the fastest subtract of two floats (4 bytes /
singles)
Is it in anyway possible to setup the hardware to subtract all components
of the
vector at once ?
-- http://TheWannabee.org
- Next message: '\\\\o//'annabee: "Re: Fsub"
- Previous message: '\\\\o//'annabee: "Re: War"
- Next in thread: '\\\\o//'annabee: "Re: Fsub"
- Reply: '\\\\o//'annabee: "Re: Fsub"
- Reply: Stepan Polovnikov: "Re: Fsub"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]