Thanks. I was looking at bfin. MT's implementation looks similar but
simpler.
> -Original Message-
> From: Ramana Radhakrishnan [mailto:[EMAIL PROTECTED]
> Sent: 16 July 2008 19:17
> To: Bingfeng Mei
> Cc: gcc@gcc.gnu.org
> Subject: Re: Question about dol
Hi Bingfeng,
> Hello,
> I tried to use doloop_end pattern to reduce loop overhead for our target
> processor, which features a dedicated loop instruction. Somehow even a
> simple loop just cannot pass the test of doloop_condition_get, which
> requires following canonical pattern.
I checked this
L PROTECTED]
> Sent: 16 July 2008 16:05
> To: gcc@gcc.gnu.org
> Cc: Ian Lance Taylor; Bingfeng Mei
> Subject: Re: Question about doloop_end pattern
>
> I can confirm that the doloop optimization is applied for
> ARC600 / ARC700
> in a compiler based on gcc 4.4.0 20080606 (exp
I can confirm that the doloop optimization is applied for ARC600 / ARC700
in a compiler based on gcc 4.4.0 20080606 (experimental) .
OTOH, it doesn't use any of the PRE_INC, POST_INC, PRE_MODIFY or
POST_MODIFY addressing modes.
lp .L__GCC__LP2
.align 4
.L2:
add r0,r1,r4
"Bingfeng Mei" <[EMAIL PROTECTED]> writes:
> I tried to use doloop_end pattern to reduce loop overhead for our target
> processor, which features a dedicated loop instruction. Somehow even a
> simple loop just cannot pass the test of doloop_condition_get, which
> requires following canonical patt