[llvm-commits] [PATCH] Adding the Linpack test to the test suite

[Duncan Sands]

Hi Nadav,

On 25/11/12 22:05, Nadav Rotem wrote:
> Hi David,
>
> Thanks for letting me know about this.  It looks like Linpack, SciMark and Livermoore are still failing. Do we know when SciMark started failing ?

SciMark2 and LivermoreLoops also fail with dragonegg.

I think the problem with SciMark2 is that it is non-deterministic: it
keeps iterating until a fixed amount of time has passed, rather than doing
a fixed number of iterations.  (I didn't look at it myself yet but this is
what I understood from discussion on the mailing list).

LivermoreLoops is more interesting.  With dragonegg it works fine as long as you
don't do link-time optimization (which the testsuite does by default).  In fact
to get the failures it is enough to run "internalize" and then some standard set
of optimization passes.  I didn't get to the bottom of this yet.

Ciao, Duncan.


> I will look at it later today or tomorrow.
>
>
> On Nov 25, 2012, at 11:17 AM, David Blaikie <dblaikie at gmail.com> wrote:
>
>> I'm not sure if this has passed previously, but it's failing on
>> http://lab.llvm.org:8011/builders/clang-x86_64-debian-fnt/builds/12074/steps/make.test-suite/logs/fail.LLC
>>
>> Could you please investigate/fix. (it'd be nice to get the bots green)
>>
>> On Fri, Nov 16, 2012 at 6:07 AM, Nadav Rotem <nrotem at apple.com> wrote:
>>> At first I thought that only stdout is compared. But after I noticed that stderr is also compared I removed the time prints altogether. I also added a reference file for the test.
>>> Thanks.
>>>
>>> On Nov 16, 2012, at 1:02 AM, Duncan Sands <baldrick at free.fr> wrote:
>>>
>>>> Hi Nadav,
>>>>
>>>> On 16/11/12 02:02, Nadav Rotem wrote:
>>>>> I removed the parts of the code that rely on timers to calculate the number of iterations.  However, I left the time measurements that are printed to stderr.
>>>>
>>>> won't that result in spurious failures when comparing against the GCC output
>>>> or the reference output, because the printed times won't be the same?
>>>>
>>>> Ciao, Duncan.
>>>>
>>>>>
>>>>> On Nov 15, 2012, at 4:58 PM, Evan Cheng <evan.cheng at apple.com> wrote:
>>>>>
>>>>>> Hi Nadav,
>>>>>>
>>>>>> Did you remove all the timing stuff from the code?
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>> Evan
>>>>>>
>>>>>> On Nov 15, 2012, at 3:33 PM, Nadav Rotem <nrotem at apple.com> wrote:
>>>>>>
>>>>>>> Hi,
>>>>>>>
>>>>>>> Here is a patch for adding Roy Longbottom's Linpack benchmark.  Roy agreed that we include the benchmark in our test suite.
>>>>>>>
>>>>>>> <test_suite_linpack.diff>
>>>>>>>
>>>>>>> Thanks,
>>>>>>> Nadav
>>>>>>> _______________________________________________
>>>>>>> llvm-commits mailing list
>>>>>>> llvm-commits at cs.uiuc.edu
>>>>>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> llvm-commits mailing list
>>>>> llvm-commits at cs.uiuc.edu
>>>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>>>>
>>>>
>>>> _______________________________________________
>>>> llvm-commits mailing list
>>>> llvm-commits at cs.uiuc.edu
>>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>>
>>> _______________________________________________
>>> llvm-commits mailing list
>>> llvm-commits at cs.uiuc.edu
>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>