Re: Archive for SPEC CPU 2K/2006 results?

["H.J. Lu" <hjl dot tools at gmail dot com>]

On Sun, Jan 25, 2009 at 11:17 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Sun, Jan 25, 2009 at 11:04 AM, Richard Guenther
> <richard.guenther@gmail.com> wrote:
>> On Sun, Jan 25, 2009 at 7:56 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
>>> On Sun, Jan 25, 2009 at 10:45 AM, Richard Guenther
>>> <richard.guenther@gmail.com> wrote:
>>>> On Sun, Jan 25, 2009 at 7:34 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
>>>>> Hi,
>>>>>
>>>>> We have been running functional tests on SPEC CPU 2K/2006
>>>>> on Linux/ia32 and Linux/Intel64 at -O2 and -O3. We'd like to
>>>>> report pass and regressions.  We may send SPEC CPU
>>>>> regressions to
>>>>>
>>>>> http://gcc.gnu.org/ml/gcc-regression/
>>>>>
>>>>> But for passes, there is no suitable place to report.
>>>>>
>>>>> http://gcc.gnu.org/ml/gcc-testresults/
>>>>>
>>>>> is for gcc testsuite. If I send SPEC CPU pass to it, it
>>>>> may be buried by normal test results.
>>>>>
>>>>> Any suggestions.
>>>>
>>>> Put it on a website and link to it from
>>>> http://gcc.gnu.org/benchmarks/
>>>>
>>>
>>> My website isn't visible to public. Also it isn't really benchmark since
>>> I only run functional tests. My reports only show pass or which tests
>>> failed.
>>
>> I see.  I agree that gcc-regression is appropriate for FAILs then.  Is it
>> important to have PASSes available somewhere? (you can assume PASSes
>> if there are no FAIL reports).
>>
>
> But you won't know what exactly the last passing revision is and
> you can't tell if my SPEC CPU machines are running normally.
>

I changed my SPEC CPU machines to send SPEC CPU
regressions to gcc-regression. The report may look like

---
Subject: gcc (GCC) 4.3.3 20090117 (prerelease) [gcc-4_3-branch
revision 143474] failed SPEC CPU 2006 on i686
Date: Sun, 18 Jan 2009 09:22:44 -0800 (PST)
From: hjl@gnu-27.sc.intel.com (H.J. Lu (Intel64))

With runspec -c lnx-i686-gcc.cfg -n 1 -l -o asc -I all -T base -e o2
Error: 1x464.h264ref
---

I'd like to send my SPEC CPU success to a gcc mailing list. I am not
sure if SPEC SPU success is appropriate for gcc-regression. A new
mailing list, like, gcc-results or gcc-success, is useful for my purpose.
People can also use it to report gcc success for other important
packages, like Linux kernel or glibc.

Thanks.

-- 
H.J.