Re: [patch] Performance patch for MIPS conditional move in expr.c

[Andrew Pinski <andrew.pinski@caviumnetworks.com>]

[-- Attachment #1: Type: text/plain, Size: 1157 bytes --]

On Wed, Nov 14, 2012 at 1:45 PM, Steve Ellcey <sellcey@mips.com> wrote:
> On Wed, 2012-11-14 at 12:00 -0800, Andrew Pinski wrote:
>
>> I know exactly where this code comes from; I have looked at the
>> benchmark as one of the reason why I add expand_cond_expr_using_cmove
>> in the first place.  Anyways you should look into removing
>> TARGET_PROMOTE_PROTOTYPES because I found that also fixes the problem
>> mentioned here.
>>
>> Thanks,
>> Andrew Pinski
>
> Removing TARGET_PROMOTE_PROTOTYPES looks interesting but I don't know if
> it is possible for compatibility reasons.  I am still looking at my
> example though, I see GCC doing:
>
> andi    $5,$5,0x1
> xori    $5,$5,0x1
> movz    $2,$4,$5
>
> When it should just do:
>
> andi    $5,$5,0x1
> movn    $2,$4,$5

Yes I have a few patches for improving this case.  I have not
submitted them yet though.
Attached is the assembly I get from a 4.7 toolchain with all of the
changes I have internally applied; this is for n32 (though o32
produces the exact same code in this case).  I will try to post some
more in the next coming weeks.

Thanks,
Andrew Pinski


>
>
> Steve Ellcey
> sellcey@mips.com
>
>

[-- Attachment #2: n32.s --]
[-- Type: application/octet-stream, Size: 1655 bytes --]

	.file	1 "t5.c"
	.section .mdebug.abiN32
	.previous
	.gnu_attribute 4, 3
	.abicalls
	.option	pic0
	.text
	.align	2
	.align	3
	.globl	foo
.LFB0 = .
	.cfi_startproc
	.set	nomips16
	.ent	foo
	.type	foo, @function
foo:
	.frame	$sp,0,$31		# vars= 0, regs= 0/0, args= 0, gp= 0
	.mask	0x00000000,0
	.fmask	0x00000000,0
	.set	noreorder
	.set	nomacro
	xori	$3,$5,0x2006
	xor	$6,$5,$4
	srl	$7,$3,1
	srl	$2,$5,1
	ori	$8,$7,0x8800
	andi	$5,$6,0x1
	andi	$9,$2,0x77ff
	srl	$4,$4,1
	movn	$9,$8,$5
	srl	$10,$4,1
	srl	$11,$10,1
	xori	$12,$9,0x2006
	xor	$13,$9,$4
	srl	$14,$12,1
	srl	$15,$9,1
	andi	$24,$13,0x1
	ori	$25,$14,0x8800
	andi	$6,$15,0x77ff
	srl	$5,$11,1
	movn	$6,$25,$24
	srl	$4,$5,1
	srl	$3,$4,1
	xori	$2,$6,0x2006
	xor	$7,$6,$10
	srl	$8,$2,1
	srl	$10,$6,1
	andi	$12,$7,0x1
	ori	$9,$8,0x8800
	andi	$13,$10,0x77ff
	srl	$14,$3,1
	movn	$13,$9,$12
	xori	$15,$13,0x2006
	xor	$11,$13,$11
	srl	$24,$15,1
	srl	$25,$13,1
	andi	$6,$11,0x1
	ori	$7,$24,0x8800
	andi	$2,$25,0x77ff
	movn	$2,$7,$6
	xori	$8,$2,0x2006
	xor	$5,$2,$5
	srl	$10,$8,1
	srl	$12,$2,1
	andi	$9,$5,0x1
	ori	$13,$10,0x8800
	andi	$15,$12,0x77ff
	movn	$15,$13,$9
	xori	$11,$15,0x2006
	xor	$4,$15,$4
	srl	$24,$11,1
	srl	$25,$15,1
	andi	$6,$4,0x1
	ori	$7,$24,0x8800
	andi	$2,$25,0x77ff
	movn	$2,$7,$6
	xori	$8,$2,0x2006
	xor	$3,$2,$3
	srl	$10,$2,1
	srl	$5,$8,1
	andi	$12,$3,0x1
	ori	$9,$5,0x8800
	andi	$13,$10,0x77ff
	movn	$13,$9,$12
	xori	$15,$13,0x2006
	xor	$14,$13,$14
	srl	$11,$15,1
	srl	$24,$13,1
	andi	$25,$14,0x1
	ori	$4,$11,0x8800
	andi	$2,$24,0x77ff
	j	$31
	movn	$2,$4,$25

	.set	macro
	.set	reorder
	.end	foo
	.cfi_endproc
.LFE0:
	.size	foo, .-foo
	.ident	"GCC: (Cavium Development Version) 4.7.0"