{lang}[foss/2021.04] SciPy-bundle v2021.05

[boegel]

(created using eb --new-pr)

[boegel]

@boegelbot please test @ generoso

[boegelbot]

@boegel: Request for testing this PR well received on generoso

PR test command 'EB_PR=12935 EB_ARGS= /apps/slurm/default/bin/sbatch --job-name test_PR_12935 --ntasks=4 ~/boegelbot/eb_from_pr_upload_generoso.sh' executed!

• exit code: 0
• output:

Submitted batch job 17297

Test results coming soon (I hope)...

- notification for comment with ID 847013338 processed

Message to humans: this is just bookkeeping information for me,
it is of no use to you (unless you think I have a bug, which I don't).

[verdurin]

Test report by @verdurin
FAILED
Build succeeded for 31 out of 39 (3 easyconfigs in total)
easybuild-c7.novalocal - Linux centos linux 7.9.2009, x86_64, Intel Xeon Processor (Skylake, IBRS), Python 3.6.8
See https://gist.github.com/54c8b9e7268413bc805d184dfdc7f5c0 for a full test report.

[boegel]

@verdurin Looks like we need to include the OpenSSL wrapper dependency to Rust? Can you verify if that fixes this problem for you?

cargo: error while loading shared libraries: libssl.so.1.1: cannot open shared object file: No such file or directory

[boegelbot]

Test report by @boegelbot
SUCCESS
Build succeeded for 3 out of 3 (3 easyconfigs in total)
generoso-c1-s-1 - Linux centos linux 8.2.2004, x86_64, Intel(R) Xeon(R) CPU E5-2667 v3 @ 3.20GHz (haswell), Python 3.6.8
See https://gist.github.com/397a5ffe5bfb932023c7beaf21a01783 for a full test report.

[boegel]

Test report by @boegel
SUCCESS
Build succeeded for 3 out of 3 (3 easyconfigs in total)
node3161.skitty.os - Linux centos linux 7.9.2009, x86_64, Intel(R) Xeon(R) Gold 6140 CPU @ 2.30GHz, Python 3.6.8
See https://gist.github.com/87cf1d1752e0a62cc0e0d31905f22a2c for a full test report.

[boegel]

Test report by @boegel
SUCCESS
Build succeeded for 3 out of 3 (3 easyconfigs in total)
node2617.swalot.os - Linux centos linux 7.9.2009, x86_64, Intel(R) Xeon(R) CPU E5-2660 v3 @ 2.60GHz (haswell), Python 3.6.8
See https://gist.github.com/76ebfb350ff7dd0e042e31d296fafad3 for a full test report.

[boegel]

Test report by @boegel
SUCCESS
Build succeeded for 3 out of 3 (3 easyconfigs in total)
node3532.doduo.os - Linux RHEL 8.2, x86_64, AMD EPYC 7552 48-Core Processor (zen2), Python 3.6.8
See https://gist.github.com/81d6648614a3e51ef5b44bbf056df1a0 for a full test report.

[verdurin]

Test report by @verdurin
FAILED
Build succeeded for 27 out of 28 (3 easyconfigs in total)
nuc.lan - Linux Fedora 33, x86_64, Intel(R) Core(TM) i7-8650U CPU @ 1.90GHz, Python 3.9.4
See https://gist.github.com/dabae0a1f54ff156219a56ac71e29cb9 for a full test report.

[boegel]

@verdurin The following module(s) are unknown: "Rust/1.52.1-GCCcore-10.3.0"

Multiple sessions in parallel?

[verdurin]

Test report by @verdurin
SUCCESS
Build succeeded for 1 out of 1 (3 easyconfigs in total)
nuc.lan - Linux Fedora 33, x86_64, Intel(R) Core(TM) i7-8650U CPU @ 1.90GHz, Python 3.9.4
See https://gist.github.com/a46c0443a3884af5ca43ca6b6adabe2c for a full test report.

[verdurin]

@boegel the most recent build is with Rust modified with the OpenSSL wrapper as a dependency.

[boegel]

Looks ready to go on x86_64, but there are known problems on aarch64 and pp64le:

• On aarch64 (AWS Graviton2), the numpy test suite fails, more details in np.linalg.eigvals() unexpected results on ARM CPUs numpy/numpy#18422.

• On ppc64le, I'm seeing a segmentation fault, which only occurs when using FlexiBLAS with OpenBLAS as a backend, not when using OpenBLAS directory, or using BLIS as a backend for FlexiBLAS. Reported upstream in segmentation fault with numpy on POWER9 (only) when using FlexiBLAS mpimd-csc/flexiblas#17 .

Neither of these issues should block this PR though, especially since the problem is not in numpy itself, but in OpenBLAS and/or FlexiBLAS...

[verdurin]

Looks fine.

[verdurin]

Going in, thanks @boegel!

[Flamefire]

Test report by @Flamefire
FAILED
Build succeeded for 20 out of 21 (3 easyconfigs in total)
taurusml20 - Linux RHEL 7.6, POWER, 8335-GTX (power9le), Python 2.7.5
See https://gist.github.com/7766b6b58fd10ed162f936a67a0553a0 for a full test report.

[boegel]

@Flamefire Anything useful in the core dump?

[Flamefire]

The segfault I get is a double-free:
*** Error in /beegfs/global0/ws/s3248973-easybuild/easybuild-ml/software/Python/3.9.5-GCCcore-10.3.0/bin/python': double free or corruption (out): 0x0000000013434ed0 ***`

(gdb) bt
#0  0x0000200000b6fbf0 in raise () from /lib64/libc.so.6
#1  0x0000200000b71f6c in abort () from /lib64/libc.so.6
#2  0x0000200000bb8d10 in __libc_message () from /lib64/libc.so.6
#3  0x0000200000bc9be0 in free () from /lib64/libc.so.6
#4  0x0000200b0abf9920 in release_dgeev (params=0x7ffffffdcf00) at numpy/linalg/umath_linalg.c.src:2229
#5  DOUBLE_eig_wrapper (JOBVL=JOBVL@entry=78 'N', JOBVR=JOBVR@entry=86 'V', args=0x200b1f804120, dimensions=<optimized out>, steps=<optimized out>) at numpy/linalg/umath_linalg.c.src:2324
#6  0x0000200b0abf9a70 in DOUBLE_eig (args=<optimized out>, dimensions=<optimized out>, steps=<optimized out>, __NPY_UNUSED_TAGGEDfunc=<optimized out>) at numpy/linalg/umath_linalg.c.src:2336

[boegel]

@Flamefire Please report that in mpimd-csc/flexiblas#17 (keeping track of things in merged PRs is impossible)

That seems to suggest the problem may be in numpy rather than FlexiBLAS though...

[Flamefire]

Not sure. It doesn't happen with OpenBLAS alone and there are some failing lapack calls with FlexiBLAS so it might as well be memory corruption
And yes reported there

[boegel]

I opened a dedicated issue for the segfault on POWER, see #12968 .

For the failing numpy tests on aarch64, we can track that in #11959.

[Flamefire]

Test report by @Flamefire
SUCCESS
Build succeeded for 9 out of 9 (3 easyconfigs in total)
taurusi8006 - Linux centos linux 7.9.2009, x86_64, AMD EPYC 7352 24-Core Processor (zen2), Python 2.7.5
See https://gist.github.com/592685426ccac14e05e81dc8ad116794 for a full test report.