This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: [RFC PR43721] Optimize a/b and a%b to single divmod call

From: Richard Henderson <rth at redhat dot com>
To: Richard Biener <rguenther at suse dot de>, Prathamesh Kulkarni <prathamesh dot kulkarni at linaro dot org>
Cc: Jim Wilson <jim dot wilson at linaro dot org>, Kugan Vivekanandarajah <kugan dot vivekanandarajah at linaro dot org>, Ramana Radhakrishnan <ramana dot radhakrishnan at arm dot com>, "gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>, richard dot sandiford at arm dot com
Date: Mon, 1 Feb 2016 09:15:40 +1100
Subject: Re: [RFC PR43721] Optimize a/b and a%b to single divmod call
Authentication-results: sourceware.org; auth=none
References: <CAAgBjMmSzFUwa8W2Tc6GzGcA3nTZ_7oOfyB-UPmRLtfKXioUMQ at mail dot gmail dot com> <CAFiYyc1OVk2L76pTjhRFzgco-D7YbA=uK+xG95Xn-whuFh292Q at mail dot gmail dot com> <CAAgBjM=XtqUa60TOdRXnpMkMcGO7fAupbuYWGmTPbt_qkkOrgQ at mail dot gmail dot com> <CAAgBjMmF4kUK2Uovw=rxYF1vpjAmfzFTsXQaRj9yj7v+ep1-Ag at mail dot gmail dot com> <alpine dot LSU dot 2 dot 11 dot 1511021342240 dot 10078 at zhemvz dot fhfr dot qr> <CAAgBjMmB=ow24szF+qW01+ordKGcOEHRLcx5fLyQnzvHND0fow at mail dot gmail dot com> <alpine dot LSU dot 2 dot 11 dot 1511041546420 dot 10078 at zhemvz dot fhfr dot qr> <CAAgBjMnL__ZhkLZG=kOL3xrf-s6gGvW6EGBvqeizrE0WDdQ8Zg at mail dot gmail dot com> <alpine dot LSU dot 2 dot 11 dot 1511101522180 dot 10078 at zhemvz dot fhfr dot qr> <CAAgBjM=9Uu_=SQp+v6nM3xK_7Eg6AFcDTfVO2EQP_Z_vUeg93g at mail dot gmail dot com> <alpine dot LSU dot 2 dot 11 dot 1511111127310 dot 4884 at t29 dot fhfr dot qr> <CAAgBjMn2MtfChCkk0ZQ-n10t_gzHviUBu+NDZNfLNFBmT35xAw at mail dot gmail dot com> <alpine dot LSU dot 2 dot 11 dot 1511111429220 dot 4884 at t29 dot fhfr dot qr> <CAAgBjM=FZEoU2c+OAnunCx4teZdBXOxAwyf5V3BFSD1PL08vUg at mail dot gmail dot com> <alpine dot LSU dot 2 dot 11 dot 1601281434470 dot 31122 at t29 dot fhfr dot qr>

On 01/29/2016 12:37 AM, Richard Biener wrote:

To workaround this, I defined a new hook expand_divmod_libfunc, which
targets must override for expanding call to target-specific dimovd.
The "default" hook default_expand_divmod_libfunc() expands call to
libgcc2.c:__udivmoddi4() since that's the only "generic" divmod
available.
Is this a reasonable approach ?


Hum.  How do they get to expand/generate it today?  That said, I'm
no expert in this area.

A simpler solution may be to not do the transform if there is no
instruction for divmod.


Are we certain that the libcall is a win for any target?

I would have expected a default of

	q = x / y
	r = x - (q * y)

to be most efficient on modern machines. Even more so on targets like ARM thathave multiply-and-subtract instructions.

I suppose there's the case of the really tiny embedded chips that don't havemultiply patterns either. So that this default expansion still results inmultiple libcalls.

I do like the transformation, because for machines that don't have a divmodinstruction, being able to strength-reduce a mod operation to a multiplyoperation is a nice win.

r~

References:
- Re: [RFC PR43721] Optimize a/b and a%b to single divmod call
  - From: Prathamesh Kulkarni
- Re: [RFC PR43721] Optimize a/b and a%b to single divmod call
  - From: Richard Biener

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]