[test] New OpenACC/CUDA/C++ test from MCH #342

ajocksch · 2018-06-22T09:39:02Z

This PR also provides a small correction of automatic_arrays.py.

Closes #248
Closes #249

vkarak · 2018-06-22T09:40:39Z

@ajocksch Please associate this PR with the issue it solves.

vkarak · 2018-06-27T11:24:05Z

@ajocksch Are the failures on Kesch expected?

vkarak

Apart from my specific comments, I have the following more generic ones:

The test provided by MCH supports also two test cases (MPI and no MPI) without OpenACC. Should we generate tests for these as well? The corresponding issues do not require that, though.
Are the failures on Kesch expected?

vkarak · 2018-06-28T08:51:46Z

cscs-checks/mch/openacc_cuda_mpi_cppstd.py

+
+
+@rfm.parameterized_test(['mpi'], ['nompi'])
+class OpenaccCudaMpiCppstd(rfm.RegressionTest):


The class name should not have the Mpi inside it because you are also generating a test that is not using MPI. This is confusing.

vkarak · 2018-06-28T08:53:23Z

cscs-checks/mch/openacc_cuda_mpi_cppstd.py

+
+@rfm.parameterized_test(['mpi'], ['nompi'])
+class OpenaccCudaMpiCppstd(rfm.RegressionTest):
+    def __init__(self, withmpi):


The withmpi should be a boolean, not a string.

vkarak · 2018-06-28T08:54:41Z

cscs-checks/mch/openacc_cuda_mpi_cppstd.py

+        self.valid_prog_environs = ['PrgEnv-cray', 'PrgEnv-pgi']
+        if self.current_system.name in ['daint', 'dom']:
+            self.modules = ['craype-accel-nvidia60']
+            self._pgi_flags = '-O2 -acc -ta=tesla:cc60 -Mnorpath -lstdc++'


Why the -lstdc++ is needed here?

It throws link error std_cpp_call.o:(.eh_frame+0x12): undefined reference to __gxx_personality_v0 if the options is not there.

this for some reason required

std_cpp_call.o:(.eh_frame+0x12): undefined reference to `__gxx_personality_v0'

I could also try to move it into the makefile

It's fine here. No problem.

vkarak · 2018-06-28T08:56:25Z

cscs-checks/mch/openacc_cuda_mpi_cppstd.py

+            self._nvidia_sm = '37'
+
+        if withmpi == 'mpi':
+            self.mpiflag = ' -DUSEMPI'


You could also make this "private" as with the _nvidia_sm. Also the flag should be USE_MPI, not USEMPI.

teojgo

lgtm apart from the changes proposed by @vkarak

ajocksch · 2018-07-05T14:28:52Z

currently the check with pgi+mpi fails, the reeason should be the wrong PrgEnv modules; this should be fixed when merging into the master

ajocksch · 2018-07-05T15:17:20Z

about 1.: Do you mean to compile with -hnoacc? I think this is not possible since a cuda kernel is called, we would need to provide a cpu version of that part of the code as well.

teojgo · 2018-07-06T06:31:02Z

@jenkins-cscs retry all

teojgo · 2018-07-06T06:42:49Z

lgtm. @ajocksch are the kesch failures expected?

…cks/mch_openacc_cuda_mpi_cppstd

ajocksch · 2018-07-06T07:09:57Z

#348 needs to be merged in the master first

vkarak

Please change the name of the class, and I will approve.

vkarak · 2018-07-06T08:07:54Z

cscs-checks/mch/openacc_cuda_mpi_cppstd.py

+
+
+@rfm.parameterized_test([True], [False])
+class OpenaccCudaMpiNoMPICppstd(rfm.RegressionTest):


The name is weird! Why not just OpenaccCudaCpp, which is in accordance with the issue as well?

@teojgo Can you change the name of this class? Other than this, this is ready to be merged.

I don't think the automatically generated name will be very meaningful for this one: OpenaccCudaCpp_False and OpenaccCudaCpp_True. So it should be set manually.

…pstd

…scs/reframe into checks/mch_openacc_cuda_mpi_cppstd

codecov-io · 2018-07-10T14:01:39Z

Codecov Report

Merging #342 into master will decrease coverage by 0.02%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master     #342      +/-   ##
==========================================
- Coverage   91.12%   91.09%   -0.03%     
==========================================
  Files          68       68              
  Lines        8244     8244              
==========================================
- Hits         7512     7510       -2     
- Misses        732      734       +2

Impacted Files	Coverage Δ
reframe/core/config.py	`82.72% <0%> (-1.82%)`	⬇️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2018a71...fdd50af. Read the comment docs.

vkarak · 2018-07-17T09:12:00Z

@ajocksch Can you check if the failures on Kesch are expected? IMO, perhaps they are not. If they are, though, I can merge this immediately.

…om/eth-cscs/reframe into checks/mch_openacc_cuda_mpi_cppstd

Conflicts: config/cscs.py cscs-checks/mch/automatic_arrays.py

vkarak · 2018-10-08T15:34:31Z

@jenkins-cscs retry dom

added OpenACC+CUDA+C++ check; small correction of automatic_arrays.py

b2fbe86

ajocksch self-assigned this Jun 22, 2018

ajocksch requested a review from victorusu June 22, 2018 09:39

vkarak added this to the Upcoming sprint milestone Jun 22, 2018

vkarak added prio: important regression test labels Jun 22, 2018

vkarak self-requested a review June 22, 2018 09:40

vkarak changed the title ~~added OpenACC+CUDA+C++ check; small correction of automatic_arrays.py~~ [test] Add OpenACC/CUDA/C++ test from MCH Jun 27, 2018

vkarak changed the title ~~[test] Add OpenACC/CUDA/C++ test from MCH~~ [test] New OpenACC/CUDA/C++ test from MCH Jun 27, 2018

vkarak requested a review from teojgo June 27, 2018 13:45

vkarak requested changes Jun 28, 2018

View reviewed changes

teojgo requested changes Jul 5, 2018

View reviewed changes

made changes required

c01e517

Merge branch 'master' into checks/mch_openacc_cuda_mpi_cppstd

bcba544

teojgo approved these changes Jul 6, 2018

View reviewed changes

Merge branch 'master' of https://github.com/eth-cscs/reframe into che…

534c5e2

…cks/mch_openacc_cuda_mpi_cppstd

vkarak requested changes Jul 6, 2018

View reviewed changes

Merge branch 'master' into checks/mch_openacc_cuda_mpi_cppstd

0333499

vkarak assigned teojgo Jul 10, 2018

Theofilos Manitaras added 3 commits July 10, 2018 15:46

Merge github.com:eth-cscs/reframe into checks/mch_openacc_cuda_mpi_cp…

f470147

…pstd

Address PR comments on check name

ed8911d

Merge branch 'checks/mch_openacc_cuda_mpi_cppstd' of github.com:eth-c…

fdd50af

…scs/reframe into checks/mch_openacc_cuda_mpi_cppstd

Merge branch 'master' into checks/mch_openacc_cuda_mpi_cppstd

ac2bd2d

vkarak modified the milestones: ReFrame sprint 2018w26, Upcoming sprint Jul 12, 2018

Merge branch 'master' into checks/mch_openacc_cuda_mpi_cppstd

0bcbba1

vkarak approved these changes Jul 17, 2018

View reviewed changes

vkarak added prio: normal and removed prio: important labels Jul 18, 2018

Merge branch 'master' into checks/mch_openacc_cuda_mpi_cppstd

a265f9a

vkarak modified the milestones: Summer sprint, ReFrame sprint 2018w35 Aug 24, 2018

ajocksch added 5 commits August 28, 2018 11:05

intermediate commit

887e594

merge master into branch

3b6dbae

intermediate stage | not working without MPI

be589f7

WIP: only MPI case is running

c2b6e64

Merge branch 'checks/mch_openacc_cuda_mpi_cppstd' of https://github.c…

c3ac9b3

…om/eth-cscs/reframe into checks/mch_openacc_cuda_mpi_cppstd

vkarak removed this from the ReFrame sprint 2018w35 milestone Sep 4, 2018

Vasileios Karakasis added 2 commits October 8, 2018 11:51

Merge branch 'master' into checks/mch_openacc_cuda_mpi_cppstd

adfc0ad

Conflicts: config/cscs.py cscs-checks/mch/automatic_arrays.py

Fix OpenACC+CUDA+MPI test on Kesch+Dom

e1f44fb

vkarak added this to the Upcoming sprint milestone Oct 8, 2018

vkarak merged commit 9069bab into master Oct 8, 2018

vkarak deleted the checks/mch_openacc_cuda_mpi_cppstd branch October 8, 2018 15:44



		@rfm.parameterized_test(['mpi'], ['nompi'])
		class OpenaccCudaMpiCppstd(rfm.RegressionTest):



		@rfm.parameterized_test([True], [False])
		class OpenaccCudaMpiNoMPICppstd(rfm.RegressionTest):

[test] New OpenACC/CUDA/C++ test from MCH #342

[test] New OpenACC/CUDA/C++ test from MCH #342

Uh oh!

Conversation

ajocksch commented Jun 22, 2018 • edited by vkarak Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

vkarak commented Jun 22, 2018

Uh oh!

vkarak commented Jun 27, 2018

Uh oh!

vkarak left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

teojgo Jul 5, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

teojgo left a comment

Choose a reason for hiding this comment

Uh oh!

ajocksch commented Jul 5, 2018

Uh oh!

ajocksch commented Jul 5, 2018

Uh oh!

teojgo commented Jul 6, 2018

Uh oh!

teojgo commented Jul 6, 2018

Uh oh!

ajocksch commented Jul 6, 2018

Uh oh!

vkarak left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

codecov-io commented Jul 10, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

vkarak commented Jul 17, 2018

Uh oh!

vkarak commented Oct 8, 2018

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

ajocksch commented Jun 22, 2018 •

edited by vkarak

Loading

teojgo Jul 5, 2018 •

edited

Loading

codecov-io commented Jul 10, 2018 •

edited

Loading