Glibc Lands HWCAPs Support For LoongArch

[phoronix]

Phoronix: Glibc Lands HWCAPs Support For LoongArch

When it comes to Glibc HWCAPs for allowing the C library to load optimized libraries based upon the CPU features at run-time, it's mostly been focused on the x86_64 world for targeting higher x86-64 levels or being able to load optimized libraries for systems with AVX support. Loongson though has now contributed initial LoongArch HWCAPs support...

https://www.phoronix.com/news/Glibc-HWCAPs-LoongArch

[stormcrow]

The branding on the English version of the Loongson logo has an unfortunate connotation that I doubt occurred to those that came up with it... "Cored..." can also mean they removed/ate the best to leave the undesirable bits behind. The pitfalls of international branding initiatives.

[Vistaus]

The branding on the English version of the Loongson logo has an unfortunate connotation that I doubt occurred to those that came up with it... "Cored..." can also mean they removed/ate the best to leave the undesirable bits behind. The pitfalls of international branding initiatives.

Still better than that other project called “Mold”.

[stormcrow]

Still better than that other project called “Mold”.

True, it's a bad name for a project if nothing else because it's impossible to search for without additional context. But, at least it's not for product branding. It's just an unfortunate portmanteau of "modern" and "ld", which is in the realm of terse, often obscure or even ironically obtuse, Unix naming conventions. If we stick to just bad names for product branding, nothing at all beats the stupidity behind "X" (and I don't mean X, the successor to W).

[zexelon]

True, it's a bad name for a project if nothing else because it's impossible to search for without additional context. But, at least it's not for product branding. It's just an unfortunate portmanteau of "modern" and "ld", which is in the realm of terse, often obscure or even ironically obtuse, Unix naming conventions. If we stick to just bad names for product branding, nothing at all beats the stupidity behind "X" (and I don't mean X, the successor to W).

I think we need to push to rename wayland to "Y" to keep with convention... that way when wayland/Y is finally abandoned we can have the true penultimate display system "Z"!!!!

And since "Z" should be functionally usable by about the end of the suns fuel supply it will be the last display system released before the sun moves into its red giant phase and consumes the earth.

[stormcrow]

I think we need to push to rename wayland to "Y" to keep with convention... that way when wayland/Y is finally abandoned we can have the true penultimate display system "Z"!!!!

And since "Z" should be functionally usable by about the end of the suns fuel supply it will be the last display system released before the sun moves into its red giant phase and consumes the earth.

I know some higher stack GUI programmers hope the sun swallows the Earth before they have to deal with wlroot because programming for multiple different compositors isn't in the cards.

[lectrode]

HWCAPs seems like it would provide the best solution to support both newer optimizations and backward compatibility in compiled binaries. Why is this not a viable solution for repos to adapt? Why are repos like Arch Linux seeming to lean toward instead duplicating the entire repo? Are the binaries it makes too big or something? Or am I completely misunderstanding the point of HWCAPs?

EDIT: Just saw OpenSUSE's breakdown of pros and cons of HWCAPs. Useful info. Would still like to know why this isn't a preferred solution...