[dataflow] [RFC] Remove many almost useless DCE passes

[Vladimir Makarov]

Steven Bosscher wrote:

> On 2/27/07, Vladimir Makarov <vmakarov@redhat.com> wrote:
>
>> I am just trying to compare apples with apples.
>
>
> You can't. Assuming mainline is your apple, then I assure you the
> dataflow branch is a completely different kind of fruit.
>
> You just can't compare the two that way.
>
> The only way to compare the two is by looking at the end result (code
> size, compilation speed, benchmarks). It's not really possible to
> compare how these results are obtained, because the compiler internals
> are completely different. Only the end result is comparable
>
Then you are probably moving in wrong direction worsening the generated 
code.  It is not only performance.  The generated code size for x86_64 
(SPECInt2000) was 0.5% bigger on the branch before the patch and I  am 
sure it will be even bigger after the patch.