> On Apr 6, 2021, at 12:06 PM, Satish Balay <[email protected]> wrote:
>
>> See the attachements. alltest.log is on a machine with 96 cores, ARM, with
>> FC,gcc9.3.5,mpich3.4.1,fblaslapack; 6 failures
>
> Perhaps this is an issue with ARM - and such diffs are expected - as we
> already have multiple alt files for some of these tests
The function values are hugely different even at the beginning of the
optimization which is why I don't think it is just round-off differences.
>
> $ ls -lt src/tao/bound/tutorials/output/plate2f_*
> -rw-r--r--. 1 balay balay 1029 Mar 23 19:48
> src/tao/bound/tutorials/output/plate2f_1_alt.out
> -rw-r--r--. 1 balay balay 1071 Mar 23 19:48
> src/tao/bound/tutorials/output/plate2f_1.out
> -rw-r--r--. 1 balay balay 1029 Mar 23 19:48
> src/tao/bound/tutorials/output/plate2f_2_alt.out
> -rw-r--r--. 1 balay balay 1071 Mar 23 19:48
> src/tao/bound/tutorials/output/plate2f_2.out
>
>>>>>>>
> not ok diff-vec_is_is_tutorials-ex2f_1 # Error code: 1
> # 16,24d15
> # < 5
> # < 7
> # < 9
> # < 11
> # < 13
> # < 15
> # < 17
> # < 19
> # < 21
> <<<<<
>
> This one is puzzling - missing fortran stdout? Perhaps compile issue on ARM?
> [its a sequential example - so can't blame MPI]
>
> Or they are all related to the optimization flags used? What configure
> options were used for the build?
>
> Satish
>
> On Tue, 6 Apr 2021, Barry Smith wrote:
>
>>
>> Alp,
>>
>> Except for the first test, these are all optimization problems (mostly in
>> Fortran). The function values are very different so I am sending it to our
>> optimization expert to take a look at it. The differences could possibly be
>> related to the use of real() and maybe the direct use of floating point
>> numbers that the compiler first treats as single and then converts to double
>> thus losing precision.
>>
>> Chen Gang, I assume you compiled with the default standard precision PETSc
>> configure options?
>>
>>
>>
>> On Apr 6, 2021, at 3:56 AM, Chen Gang
>> <[email protected]<mailto:[email protected]>> wrote:
>>
>>
>> See the attachements. alltest.log is on a machine with 96 cores, ARM, with
>> FC,gcc9.3.5,mpich3.4.1,fblaslapack; 6 failures
>> alltest2.log is on an intel machine with 40
>> cores,x86, without FC; icc&mkl& intel mpi; only 1 failure
>>
>> ------------------ 原始邮件 ------------------
>> 发件人: "petsc-dev" <[email protected]<mailto:[email protected]>>;
>> 发送时间: 2021年4月6日(星期二) 中午12:38
>> 收件人: "petsc-dev"<[email protected]<mailto:[email protected]>>;
>> 抄送: "Chen
>> Gang"<[email protected]<mailto:[email protected]>>;"cglwdm"<[email protected]<mailto:[email protected]>>;
>> 主题: Re: [petsc-dev] Petsc: Error code 1
>>
>> Note: do not use '-j' with alltests.
>>
>> And run alltests on both machines [but *not* at the same time on machines]
>> and send us logs from both the runs.
>>
>> Satish
>>
>>
>> On Mon, 5 Apr 2021, Satish Balay wrote:
>>
>>> Try:
>>>
>>> make alltests TIMEOUT=600
>>>
>>> And send us the complete log (alltests.log)
>>>
>>> Satish
>>>
>>> On Tue, 6 Apr 2021, Chen Gang wrote:
>>>
>>>> Dear sir,
>>>>
>>>>
>>>> The result of make check is OK. And I do set the timeout to a larger
>>>> value, which keeps me from getting timeout error. The thing is I have two
>>>> machines. And I get the error code 1 in different tests on different
>>>> machines.I don’t know what is error code1. What case this? How can I fix
>>>> the failure tests.
>>>>
>>>>
>>>> ------------------ Original ------------------
>>>> From: Satish Balay <[email protected]<mailto:[email protected]>>
>>>> Date: Tue,Apr 6,2021 0:18 PM
>>>> To: Chen Gang <[email protected]<mailto:[email protected]>>
>>>> Cc: petsc-dev <[email protected]<mailto:[email protected]>>,
>>>> cglwdm <[email protected]<mailto:[email protected]>>
>>>> Subject: Re: [petsc-dev] Petsc: Error code 1
>>>
>>
>> <alltests2.log><alltests.log>
>>
