[CentOS-devel] Feedback on Red Hat Enterprise Linux 10 plans

Mon Jan 8 23:13:44 UTC 2024
Gerd v. Egidy <lists at egidy.de>

Hi,

I would also like to voice my concerns about requiring x86_64-v3 for RHEL 10.

In the blog post https://developers.redhat.com/articles/2024/01/02/exploring-x86-64-v3-red-hat-enterprise-linux-10 
Florian writes:

"... and adopting x86-64-v3 will exclude some systems from being able to run 
RHEL 10, just as the choice of x86-64-v2 for RHEL 9 excluded some systems."

I would like to point out a major difference between what was done on RHEL 9 
and what is now planned for 10: the last CPU core designs that did not include 
support for x86_64-v2 were Intel Cedar Trail and AMD Bobcat. While it is hard 
to find reliable discontinuation dates, it looks to me like they were both 
discontinued in 2014. So when RHEL 9 was released in 2022, the affected systems 
were already 8 years or older. So they were already aging systems.

With the plans for RHEL 10 this is a completely different matter: as Florian 
writes in the blogpost, Intel still released new CPU variants without support 
for x86-64-v3 in 2023 so it is very likely that new CPUs without x86-64-v3 
will be still sold by the time RHEL 10 will be released. This is of course in 
addition to the systems sold in the last years without x86-64-v3 and that 
won't be able to upgrade.

I don't think it is a good idea to raise the minimum requirements so much that 
even systems sold as new don't meet them.

Also the impacts of such a decision on the environment should be considered. 
When you believe for example the figures in 
https://www.tomshardware.com/software/windows/microsofts-draconian-windows-11-restrictions-will-send-an-estimated-240-million-pcs-to-the-landfill-when-windows-10-hits-end-of-life-in-2025
then the hardware requirements decided on by Microsoft for Windows 11 alone 
will lead to hundred thousands of tons of additional electronic waste. I think 
Linux vendors should set a better example in this regard.

It is these two points that lead me to believe that the raising the baseline 
to x86-64-v3 should not be implemented in RHEL 10. Instead further 
research&development should go into hwcaps and should then be reviewed by the 
time RHEL 11 will be branched.

Kind regards,

Gerd