Steve (sberens.Throwaway.delete@this.gmail.com) on July 14, 2013 8:10 pm wrote:
> > > No. It shows that Intel's optimization is still superior to others. (Closest one is VC not GCC/CLANG)
> >
>
> Actually Intel's optimization is still superior in one way. Removal of code
> that is actually needed in benchmarks thus inflating benchmark results to beat
> the competition then post said tainted results or hire a lacky to do it.
>
> Intel cheating at benchmarks has happened many many times in the past. This is just one more example.
>
> ARM Beats Intel With Revised AnTuTu Benchmark
> http://www.eetimes.com/author.asp?section_id=36&doc_id=1318894&
>
> Some quotes:
>
> > Under the revised benchmark, overall scores for the Atom Z2580 dropped by about 20
> > percent.
>
> > Technical consulting firm BDTI pointed out that the compiled code for the Intel
> > processor was not executing all instructions that were intended for the RAM test.
>
> > The revision still uses the ICC compiler, but the resulting scores are drastically
> > different for the Intel processor. The AnTuTu CPU and overall scores dropped by
> > approximately 20 percent, while the AnTuTu RAM score plummeted by approximately 50
> > percent
>

Removal of code if results are unused is valid optimization. It is up to programmer to ensure that in benchmark, such optimization cannot be done. (Only code whose results are needed is required.)
Bugs and errors in coed are not compilers responsibility...