On Sat, Jun 14, 2025 at 6:23 PM Jerome Kieffer
wrote:
> Hi Ralph,
>
> There is no problem here except "any hardware more than 10yo* is
> deprecated and we can discontinue its support" which is something Intel
> would love but not me (nor anybody who realized we are living in a
> finite world).
>
This is a pretty wildly incorrect representation, please re-read the first
message in this thread.
I'll also summarize again:
1. The default build settings increase the hardware requirements to
x86-64-v2, dropping support for pre-2009 hardware and a few specific Intel
Atom CPUs produced in the 2009-2013 timeframe.
2. This will affect <0.22% of machines based on the best data we have
3. A previous such bump to SSE3 caused no problems. We set the threshold
for allowing to drop support at <0.5% at that time
4. If anyone with ancient hardware is affected, they can still build NumPy
from source with the `-Ddisable-optimizations=true` flag, that will still
work and use the scalar implementations. Or pin to <2.4.0 if they want to
use wheels from PyPI.
Cheers,
Ralf
___
NumPy-Discussion mailing list -- numpy-discussion@python.org
To unsubscribe send an email to numpy-discussion-le...@python.org
https://mail.python.org/mailman3//lists/numpy-discussion.python.org
Member address: arch...@mail-archive.com
