Message-ID: <3367B958.88@silesia.top.pl>
Date: Wed, 30 Apr 1997 23:27:52 +0200
From: Michal <wapex AT silesia DOT top DOT pl>
MIME-Version: 1.0
To: djgpp AT delorie DOT com
Subject: Re: Alignment
References: <33636F49 DOT 4CE AT silesia DOT top DOT pl> <3366cd70 DOT 6952476 AT news-win DOT inp DOT nsk DOT su>
Content-Type: text/plain; charset=iso-8859-2
Content-Transfer-Encoding: 7bit
Precedence: bulk

Michael Bukin wrote:
> 
> struct test_t
> {
>   int a;
> } __attribute__ ((__aligned__ (16)));
> 
> About type attributes:
> info gcc "c extensions" "type attributes"
> 
> About variable attributes:
> info gcc "c extensions" "variable attributes"

I tryed that, it doesn't seem to work. Maybe my compiler isn't
configured in a right way?
-----------------------------------
To Eli Zaretski:
I'm shure my doubles are NOT aligned on a 64 byte boundary, if I put a
int variable beafore the inner's data I get about 10% increase or
decrease in performance, depending on the code and data before.
With data aligned on a 32 byte boudary I woudn't get any dramaticat
speed up, but it would be always something, and if it can be done why
don't do it? 
-----------------------------------
To Leath Muller:
As far as I know double and single operations are the same speed on
pentium. The only instruction, which is faster in single precision is
fdiv, but it takes more time to put 8 pixels(I interpolate u & v in
texture and light value lineary every 8 pixels) then to exectute double
precision fdiv. With double I can use some tricks, and have better
precision.