Thanks, Dima. 
Installing openblas-devel did the trick. Confusingly, the system default 
installation is libopenblas_pthreads0 and its companion header file,, while 
the headers and libraries Sage wants are found in openblas-devel. Also, the 
system default is older (0.3.7) even than the version included in 
SageMath-9.2 (0.3.9). But now I get no errors from ptestlong.

I'll probably have to revisit this in a few months. openSUSE 15.3 will drop 
soon, and doubtless will include a much newer openblas.

On Monday, May 17, 2021 at 12:34:46 PM UTC-7 dim...@gmail.com wrote:

> please check the last part of the output of ./configure 
>
> among other things it lists system libraries to install (rather than build 
> from source).
>
> it particular it would print the contents of
>
>
> https://github.com/sagemath/sage/blob/develop/build/pkgs/openblas/distros/opensuse.txt
> which is
>
> openblas-devel
>
> Most probably the pc file for openblas will be there, and we would not 
> have to try solving this puzzle for you 
>
> first do
>
> make openblas-clean
>
> to remove (broken?) Sage installation of 
> openblas.
>
>
>
> On Mon, 17 May 2021, 19:04 Richard_L, <rich...@lozestech.com> wrote:
>
>> To your first point, the development headers are not installed. That can 
>> be remedied easily.
>> To your second point, yes, pkg-config is installed. There is no 
>> openblas.pc, apart from that in the sage-9.3 directory tree.
>>
>> BTW, I had no problem with Sage-9.2, which also built openblas (0.3.9) 
>> from source.
>>
>> On Monday, May 17, 2021 at 9:08:07 AM UTC-7 dim...@gmail.com wrote:
>>
>>>
>>>
>>> On Mon, 17 May 2021, 16:42 Richard_L, <rich...@lozestech.com> wrote:
>>>
>>>> What I find in common for the 7 failures is:
>>>> "/home/rllozes/sage-9.3/local/lib/libopenblas.so.0(zgemm_kernel_n+0x9d)[0x7f2b0f8c549d]",
>>>>  
>>>> and further
>>>> "0x00007f2b0f8c549d in zgemm_kernel_n () at 
>>>> ../kernel/x86_64/zgemm_kernel_1x4_nehalem.S:214"
>>>> indicating an openblas error. 
>>>> The install.log shows that we built openblas with correct detection of 
>>>> AVX (yes), AVX2 (no), and AVX-512 (no), which was once a common root cause 
>>>> for openblas failures (ca. 2012). OTOH, the system already supplies an 
>>>> openblas in /usr/lib64, which configure did not find.
>>>>
>>>
>>> you might need the development headers etc for system openblas installed.
>>> Or, perhaps, they are installed, but pkg-config does not know about it.
>>>
>>> Do you have system wide pkg-config installed?
>>>
>>>
>>>
>>>
>>>
>>>>
>>>> On Monday, May 17, 2021 at 2:32:59 AM UTC-7 dim...@gmail.com wrote:
>>>>
>>>>> On Sun, May 16, 2021 at 4:26 PM Richard_L <rich...@lozestech.com> 
>>>>> wrote: 
>>>>> > 
>>>>> > 'make ptestlong' gave the following error summary on a new build 
>>>>> from source: 
>>>>> > 
>>>>> > 
>>>>> ---------------------------------------------------------------------- 
>>>>> > sage -t --long --warn-long 103.3 --random-seed=0 
>>>>> src/sage/calculus/riemann.pyx # Killed due to illegal instruction 
>>>>>
>>>>>
>>>>> this kind of error indicates that a component is using CPU 
>>>>> instructions not understood by your CPU. 
>>>>> It could be either something you installed from the system, or 
>>>>> something that you built from source as a part of Sage. 
>>>>> You can look in logs/ptestlong.log to locate more precise errors/stack 
>>>>> dumps. 
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> > sage -t --long --warn-long 103.3 --random-seed=0 
>>>>> src/sage/schemes/riemann_surfaces/riemann_surface.py # Killed due to 
>>>>> illegal instruction 
>>>>> > sage -t --long --warn-long 103.3 --random-seed=0 
>>>>> src/sage/matrix/matrix2.pyx # Killed due to illegal instruction 
>>>>> > sage -t --long --warn-long 103.3 --random-seed=0 
>>>>> src/sage/misc/functional.py # Killed due to illegal instruction 
>>>>> > sage -t --long --warn-long 103.3 --random-seed=0 
>>>>> src/sage/calculus/interpolators.pyx # Killed due to illegal instruction 
>>>>> > sage -t --long --warn-long 103.3 --random-seed=0 
>>>>> src/sage/matrix/matrix_double_dense.pyx # Killed due to illegal 
>>>>> instruction 
>>>>> > sage -t --long --warn-long 103.3 --random-seed=0 
>>>>> src/sage/schemes/affine/affine_homset.py # Killed due to illegal 
>>>>> instruction 
>>>>> > sage -t --long --warn-long 103.3 --random-seed=0 
>>>>> src/sage/modular/modform/numerical.py # Killed due to illegal instruction 
>>>>> > 
>>>>> ---------------------------------------------------------------------- 
>>>>> > Total time for all tests: 3514.1 seconds 
>>>>> > cpu time: 22425.1 seconds 
>>>>> > cumulative wall time: 27133.4 seconds 
>>>>> > make: *** [Makefile:201: ptestlong] Error 16 
>>>>> > 
>>>>> > I allowed the package to build Python 3.9.2. The OS is openSUSE Leap 
>>>>> 15.2. There were 8 threads active during testing. 
>>>>> > 
>>>>> > -- 
>>>>> > You received this message because you are subscribed to the Google 
>>>>> Groups "sage-devel" group. 
>>>>> > To unsubscribe from this group and stop receiving emails from it, 
>>>>> send an email to sage-devel+...@googlegroups.com. 
>>>>> > To view this discussion on the web visit 
>>>>> https://groups.google.com/d/msgid/sage-devel/f499990e-b662-4ace-b2ee-74a4cec2bc7cn%40googlegroups.com.
>>>>>  
>>>>>
>>>>>
>>>> -- 
>>>> You received this message because you are subscribed to the Google 
>>>> Groups "sage-devel" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send 
>>>> an email to sage-devel+...@googlegroups.com.
>>>>
>>> To view this discussion on the web visit 
>>>> https://groups.google.com/d/msgid/sage-devel/0d3ad5a5-891b-4005-b75a-07c378406d7dn%40googlegroups.com
>>>>  
>>>> <https://groups.google.com/d/msgid/sage-devel/0d3ad5a5-891b-4005-b75a-07c378406d7dn%40googlegroups.com?utm_medium=email&utm_source=footer>
>>>> .
>>>>
>>> -- 
>> You received this message because you are subscribed to the Google Groups 
>> "sage-devel" group.
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to sage-devel+...@googlegroups.com.
>>
> To view this discussion on the web visit 
>> https://groups.google.com/d/msgid/sage-devel/f4a85ad8-770f-4a25-9e9f-110c0479b08fn%40googlegroups.com
>>  
>> <https://groups.google.com/d/msgid/sage-devel/f4a85ad8-770f-4a25-9e9f-110c0479b08fn%40googlegroups.com?utm_medium=email&utm_source=footer>
>> .
>>
>

-- 
You received this message because you are subscribed to the Google Groups 
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to sage-devel+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/sage-devel/ed6d10b4-c05a-41a2-8cd2-32a1520d6da8n%40googlegroups.com.

Reply via email to