Re: Promoting RISC-V to a Tier 3 supported platform

Charles R Harris via NumPy-Discussion <[email protected]> Tue, 12 May 2026 08:33:47 -0600
Newsgroups gmane.comp.python.numeric.general
Message-ID <CAB6mnxLCz92542BO=XB1xZC+et5FiVTXPpEbLQ-Mnf0iCNzPeg@mail.gmail.com>
--===============0802862702785256425==
Content-Type: multipart/alternative; boundary="000000000000f610bf06519fc085"

--000000000000f610bf06519fc085
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Tue, May 12, 2026 at 7:23=E2=80=AFAM Ralf Gommers via NumPy-Discussion <
[email protected]> wrote:

> Hi all,
>
> There has been sustained interest in, and responsive contributors for,
> RISC-V support in NumPy. There are now self-hosted runners from the RISE
> project (https://riseproject.dev/), which we're about to get green (I
> hope) on https://github.com/numpy/numpy/pull/30995.
>
> Their interest extends to adding RISC-V as a supported platform for which
> we ship wheels on PyPI - see https://github.com/numpy/numpy/issues/30216.
> It seems a little too early to consider crossing that bridge today, but
> it's an up-and-coming platform so it's fairly likely that we'll consider =
it
> in the future I'd think.
>
> For now, I propose we add RISC-V as a Tier 3 platform at
> https://numpy.org/neps/nep-0057-numpy-platform-support.html#tier-3, next
> to ppc64le and Pyodide. That means we have CI support and contributors wh=
o
> have stuck up their hand and can be pinged in case of platform-specific
> issues that need addressing.
>
> We also have SIMD and CPU detection code specific to RISC-V, so having
> native runners in CI will be quite helpful either way. And hopefully we c=
an
> disable the slow QEMU-based job once we're happy with the self-hosted
> runners.
>
> Thoughts or concerns?
>
>
+1.

Chuck

--000000000000f610bf06519fc085
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div dir=3D"ltr"><br></div><br><div class=3D"gmail_quote g=
mail_quote_container"><div dir=3D"ltr" class=3D"gmail_attr">On Tue, May 12,=
 2026 at 7:23=E2=80=AFAM Ralf Gommers via NumPy-Discussion &lt;<a href=3D"m=
ailto:[email protected]">[email protected]</a>&gt; wrot=
e:<br></div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0=
.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir=3D"l=
tr"><div>Hi all,</div><div><br></div><div>There has been sustained interest=
 in, and responsive contributors for, RISC-V support in NumPy. There are no=
w self-hosted runners from the RISE project (<a href=3D"https://riseproject=
.dev/" target=3D"_blank">https://riseproject.dev/</a>), which we&#39;re abo=
ut to get green (I hope) on=C2=A0<a href=3D"https://github.com/numpy/numpy/=
pull/30995" target=3D"_blank">https://github.com/numpy/numpy/pull/30995</a>=
.</div><div><br></div><div>Their interest extends to adding RISC-V as a sup=
ported platform for which we ship wheels on PyPI - see <a href=3D"https://g=
ithub.com/numpy/numpy/issues/30216" target=3D"_blank">https://github.com/nu=
mpy/numpy/issues/30216</a>. It seems a little too early to consider crossin=
g that bridge today, but it&#39;s an up-and-coming platform so it&#39;s fai=
rly likely that we&#39;ll consider it in the future I&#39;d think.</div><di=
v><br></div><div>For now, I propose we add RISC-V as a Tier 3 platform at=
=C2=A0<a href=3D"https://numpy.org/neps/nep-0057-numpy-platform-support.htm=
l#tier-3" target=3D"_blank">https://numpy.org/neps/nep-0057-numpy-platform-=
support.html#tier-3</a>, next to ppc64le and Pyodide. That means we have CI=
 support and contributors who have stuck up their hand and can be pinged in=
 case of platform-specific issues that need addressing.=C2=A0</div><div><br=
></div><div>We also have SIMD and CPU detection code specific to RISC-V, so=
 having native runners in CI will be quite helpful either way. And hopefull=
y we can disable the slow QEMU-based job once we&#39;re happy with the self=
-hosted runners.</div><div><br></div><div>Thoughts or concerns?</div><div><=
br></div></div></blockquote><div><br></div><div>+1.</div><div><br></div><di=
v>Chuck=C2=A0</div></div></div>

--000000000000f610bf06519fc085--

--===============0802862702785256425==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
NumPy-Discussion mailing list -- [email protected]
To unsubscribe send an email to [email protected]
https://mail.python.org/mailman3//lists/numpy-discussion.python.org
Member address: [email protected]

--===============0802862702785256425==--