[Openmp-dev] Fwd: Re: OpenMP problems with clang 3.7.0

Mon Sep 7 06:18:04 PDT 2015

I'm sorry, your mail came in a second too late...

Btw, why isn't Outlook responding to the list? I remember this working at least before the transition @Tanya

Thanks,
Jonas

From: Openmp-dev [mailto:openmp-dev-bounces at lists.llvm.org] On Behalf Of Bataev, Alexey via Openmp-dev
Sent: Monday, September 07, 2015 2:54 PM
To: openmp-dev at lists.llvm.org
Subject: [Openmp-dev] Fwd: Re: OpenMP problems with clang 3.7.0

Sending this thread to OpenMP dev list to be sure it is not lost.

Best regards,
Alexey Bataev
=============
Software Engineer
Intel Compiler Team

-------- Перенаправленное сообщение -------- 
Тема: 
Re: [Openmp-dev] OpenMP problems with clang 3.7.0
Дата: 
Mon, 7 Sep 2015 09:07:54 +0200
От: 
Christof Söger <csoeger at uos.de>
Кому: 
Bataev, Alexey <a.bataev at hotmail.com>

Hi Alexey, 

thanks for your quick answer. For you it indeed looks okay. I will test it next week.

Best,
Christof
Am 7. September 2015 07:52:35 MESZ, schrieb "Bataev, Alexey" <a.bataev at hotmail.com>: 
Another one addition - I used trunk 3.8, not 3.7 release. Could you 
check everything using clang/LLVM 3.8svn?

Best regards,
Alexey Bataev
=============
Software Engineer
Intel Compiler Team

07.09.2015 8:51, Bataev, Alexey пишет:
 Hi Christof,
 Thank you for your report!
 I tried to reproduce your problem and here is what I get when I run 
 your example:
                                                     \.....|
                     Normaliz 2.99.4                  \....|
                                                       \...|
      (C) The Normaliz Team, University of Osnabrueck   \..|
                     September 2015                      \.|
                                                          \|

************************************************************
 Compute: HilbertSeries
 ************************************************************
 starting primal algorithm with full triangulation ...
 Roughness 1
 Generators sorted by degree and lexicographically
 Generators per degree:
 1: 40
 Start simplex 1 2 3 4 5 9 12 14 16 18 20 25 28 30 32 34 36
 gen=6, 18 hyp, 2 simpl
 gen=7, 21 hyp, 4 simpl
 gen=8, 28 hyp, 8 simpl
 gen=10, 28 hyp, 12 simpl
 gen=11, 28 hyp, 16 simpl
 gen=13, 32 hyp, 26 simpl
 gen=15, 46 hyp, 51 simpl
 gen=17, 62 hyp, 86 simpl
 gen=19, 77 hyp, 148 simpl
 gen=21, 85 hyp, 218 simpl
 gen=22, 95 hyp, 304 simpl
 gen=23, 108 hyp, 420 simpl
 gen=24, 198 hyp, 745 simpl
 gen=26, 205 hyp, 1039 simpl
 gen=27, 217 hyp, 1402 simpl
 gen=29, 290 hyp, 2280 simpl
 gen=31, 561 hyp, 4298 simpl
 gen=33, 846 hyp, 351 pyr, 4318 simpl
 gen!
 =35,
1147 hyp, 718 pyr, 4324 simpl
 gen=37, 1297 hyp, 1197 pyr, 4326 simpl
 gen=38, 1525 hyp, 1493 pyr, 4326 simpl
 gen=39, 1784 hyp, 1898 pyr, 4327 simpl
 gen=40, 3753 hyp, 2284 pyr, 4331 simpl
 Pointed since graded
 Select extreme rays via comparison ... done.
 **************************************************
 level 0 pyramids remaining: 2284
 **************************************************
 **************************************************
 all pyramids on level 0 done!
 **************************************************
 evaluating 61640 simplices
 ||||||||||||||||||||||||||||||||||||||||||||||||||
 61640 simplices accumulated.
 Adding 1 denominator classes... done.
 Total number of pyramids = 2317, among them simplicial 33
________________________________________

 transforming data... done.

 How can I understand that there are some troubles with the code? I 
 tried it on MacOS 10.10.5, gmp is !
 6.0.0,
boost is 1.55.0.
 Also tried debug build with assert(Pyramid_key.size() == dim); at the 
 beginning and at the end of the critical section and everything was fine.

 Best regards,
 Alexey Bataev
 =============
 Software Engineer
 Intel Compiler Team

 07.09.2015 0:25, Christof Söger via Openmp-dev пишет:
 Hi everybody,

 I just downloaded the new 3.7 release and also the OpenMP runtime for 
 Darwin (compiled versions) and tried to compile our software package 
 "Normaliz" with it.
 Without -fopenmp=libomp it compiles and runs fine.
 With the openmp flag it compiles but I ran in problems (even if I set 
 the number of threads to 1).

 I debugged it a bit and the problem is that at the entering of a 
 critical section one std::vector, which is firstpri!
 vate in
the outer 
 parallel loop gets, invalidated.
 As a workaround it works when I copy the vector right before the 
 critical, and use that copy inside the critical section.

 I don't know under which exact circumstances this happens and was not 
 able to create a small example in which it happens. If you are 
 interested to check it, you can find the source code on github 
 https://github.com/Normaliz/Normaliz. To reproduce the problem 
 compile normaliz via cmake and run it with "normaliz -cq -x=1 
 example/medium". The critical critical is the critical(TRIANG) on 
 line 1113 in process_pyramid and the effected vector is Pyramid_key. 
 Right before the critical section an
 assert(Pyramid_key.size() == dim);
 succeeds, but in the critical section it fails.

 I hope this helps to figure out what happens. If I can any other 
 information please l!
 et me
know.

 Best regards,
 Christof

________________________________________

 Openmp-dev mailing list
 Openmp-dev at lists.llvm.org
 http://lists.llvm.org/cgi-bin/mailman/listinfo/openmp-dev

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 5868 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/openmp-dev/attachments/20150907/698ce35c/attachment.bin>