[prev in list] [next in list] [prev in thread] [next in thread] 

List:       gcc-patches
Subject:    Re: Patch: fix SPEC regressions
From:       Dale Johannesen <dalej () apple ! com>
Date:       2002-10-31 23:22:19
[Download RAW message or body]


On Thursday, October 31, 2002, at 03:18  PM, Richard Henderson wrote:

> On Tue, Oct 29, 2002 at 03:10:20PM -0800, Dale Johannesen wrote:
>> As you point out this simplifies to
>>    (abs(final-initial) + abs_inc*unroll_number - 1) / t1
>
> Thanks for the detailed explanation, but that still begs the
> question of why we don't simplify the expression before
> computing it.

So the code would correspond better to the algorithmic explanation
in the comments; not a great reason, I agree.  I'll rewrite it
simplified if you prefer.  Are you comfortable with the algorithm?

[prev in list] [next in list] [prev in thread] [next in thread]