enable SIMD on power9 (requires gcc>=8)

[kif]

This speeds up the reading of Eiger2 dataset from 36ms to 19ms (on one core) on our power9 computers.
x86 computers are much faster for this task (~12ms)

[vasole]

@t20100

Should we check the compiler version? We are (supposedly) shipping manylinux2014 wheels for power9...

[kif]

Not easy to get the version of the compiler from Python if I remember well ... I did it for Blosc but the build is driven by Cmake. The same directive exists for clang, but it appeared in version 9 (IIRC).

[vasole]

There are distutils hacks, but the simplest seems to me to recover the output of` gcc --version

[t20100]

On Power9, we are now building manylinux2014 wheels and Ubuntu20.04 packages.
Running gcc --version on manylinux2014_ppc64le gives: gcc (GCC) 9.3.1 20200408 (Red Hat 9.3.1-2)
I launched a complete build on your branch to see the outcome: https://gitlab.esrf.fr/silx/bob/hdf5plugin/-/pipelines/34424

It would be best not to modify files under src/ since it is subtrees of other projects.
There is no need to change src/bitshuffle/setup.py since it is not used, and change in src/bitshuffle/src/bitshuffle_core.c can be replaced by "-DUSESSE2" in the project's setup.py as it is already done for "-DNO_WARN_X86_INTRINSICS"

[kif]

On Wed, 30 Sep 2020 01:03:04 -0700 Thomas VINCENT ***@***.***> wrote: It would be best not to modify files under `src/` since it is subtrees of other projects. There is no need to change ` src/bitshuffle/setup.py` since it is not used, and change in ` src/bitshuffle/src/bitshuffle_core.c` can be replaced by `"-DUSESSE2"` in the project's `setup.py` as it is already done for `"-DNO_WARN_X86_INTRINSICS"`

I am not sure about that ... I see no problem in submitting a PR to bitshuffle

[t20100]

Yes, a PR to bitshuffle would be best.

[vasole]

Changing line 361 in your setup.py to extra_compile_args += ["-DNO_WARN_X86_INTRINSICS", "-DUSESSE2"] and not touching any file under src should be enough to allow the desired behavior without waiting for upstream changes.

[t20100]

You can have a look at this branch: https://github.com/t20100/hdf5plugin/tree/pr89
No change to src/bithsuffle and a slightly more generic approach to enabline SSE2 support on Power9.

[kif]

I tried and gcc emits a warning that USESSE2 is already defined ...

[kif]

My PR is available upstream to:
kiyo-masui/bitshuffle#87

[t20100]

I get no warning with: https://github.com/t20100/hdf5plugin/tree/pr89

[t20100]

Closing as supersedes by #90