Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!dgh!dgh
From: dgh%dgh@Sun.COM (David Hough)
Newsgroups: comp.lang.c
Subject: Re: Bad optimizations (was Re: comma operator)
Message-ID: <39616@sun.uucp>
Date: 21 Jan 88 21:25:55 GMT
References: <3819@sigi.Colorado.EDU> <5080013@hpfcdc.HP.COM> <7120@brl-smoke.ARPA> <269@vsi1.UUCP>
Sender: news@sun.uucp
Lines: 97
Keywords: optimization swap sun compiler
Summary: modern optimizers fix trivial test programs

In article <269@vsi1.UUCP>, steve@vsi1.UUCP (Steve Maurer) writes:
> to try this out, I decided to compile the
> following section of code on our Sun3 (without optimization).
> 
>     #define swap(a,b) ((a) = ((b) = ((a) = (a) ^ (b)) ^ (b)) ^ (a))
>     #define swap2(A, B, TYPE)  { TYPE temp; temp = A; A = B; B = temp; }
> 
>     main()
>     {
>     int a = 3, b = 5;
> 
>     swap(a, b);
>     foo_marker();	/* marker to deleniate code sections */
>     swap2(a, b, int);
>     }
> 
> The resulting code (without headers or trailers) was as follows:
> 
>     movl	a6@(-0x4),d0		<-- xor method
>     movl	a6@(-0x8),d1
>     eorl	d1,d0
>     movl	d0,a6@(-0x4)
>     movl	a6@(-0x8),d1
>     eorl	d1,d0
>     movl	d0,a6@(-0x8)
>     movl	a6@(-0x4),d1
>     eorl	d1,d0
>     movl	d0,a6@(-0x4)
> 
>     jbsr	_foo_marker
> 
>     movl	a6@(-0x4),a6@(-0xc)	<-- move method
>     movl	a6@(-0x8),a6@(-0x4)
>     movl	a6@(-0xc),a6@(-0x8)
> 
> Here is the same exclusive or swap() macro, under Sun's optimizer:
> 
>     eorl	d1,d0
>     movl	d0,a6@(-4)
>     eorl	d1,d0
>     movl	d0,a6@(-8)
>     movl	a6@(-4),d1
>     eorl	d1,d0
>     movl	d0,a6@(-4)

The above reflects what's called -O1 optimization in SunOS 4.0,
currently under development.  I tried the example program under
maximal optimization (-O4) and it optimized away to just about nothing,
so I revised the program:

#define swap(a,b) ((a) = ((b) = ((a) = (a) ^ (b)) ^ (b)) ^ (a))
#define swap2(A, B, TYPE)  { TYPE temp; temp = A; A = B; B = temp; }

main()
{
int a,b;

getab(&a,&b);		/* pretend that a and b get new values */
swap(a, b);
foo_marker(a,b);        /* marker to delineate code sections */
getab(&a,&b);		/* pretend that a and b get new values */
swap2(a, b, int);
foo_marker(a,b);        /* marker to delineate code sections */
}

and found:

	jbsr	_getab
	addqw	#8,sp
	movl	a6@(-8),d0			ex or macro
	eorl	d0,a6@(-4)
	movl	a6@(-4),d0
	eorl	d0,a6@(-8)
	movl	a6@(-8),d0
	eorl	d0,a6@(-4)
	movl	d0,sp@-
	movl	a6@(-4),sp@-
	jbsr	_foo_marker
...
	jbsr	_getab
	addqw	#8,sp
	movl	a6@(-4),d7			move via temp
	movl	a6@(-8),a6@(-4)
	movl	d7,a6@(-8)
	movl	d7,sp@-
	movl	a6@(-4),sp@-
	jbsr	_foo_marker

8 instructions instead of 5, the difference being the 3 eorl's.
In general, you wouldn't use the EOR method unless you were out of
space for temporaries, which might be the case if you were microcoding
within a small register file, but is hardly typical of C applications.

David Hough

ARPA: dhough@sun.com
UUCP: {ucbvax,decvax,allegra,decwrl,cbosgd,ihnp4,seismo}!sun!dhough