Mesa Considers Raising CPU Support Baseline

[MadeUpName]

Who uses legacy hardware has to use legacy operating systems. The newest hardware cannot be penalized by the anachronism. SSE2 is 20 years old nowadays, and we are debating if a software must be based or not on SSE2. It's stupid enough. The hardware based on early sse2 costs 20$$$, in some cases it is also dismissed and free. Mesa could be integrate also up to sse4.x standard. Assumed that the best solution is the flexibility way got by a preliminary test of the hardware specification in order to apply the best solution for each specific hardware system.

You know they currently sell hardware that doesn't support SSE4 right? Or are you a believer that Linux is only for people with bleeding edge gaming rigs?

[skeevy420]

There goes another useless phrase, meanwhile there are people who are more than happy to define "old hardware" upwards into R600, RadeonSI, Sandy Bridge, Broadwell, even anything pre-Ryzen-class hardware levels.

Those of us who are using blisteringly old ancient Intel 8080(/sarc) hardware see these comments being made, I don't know why those of you using nebulously useless phrases never see it.

If SSE2 max isn't "old hardware" then I don't know what is. That's what the article is talking about.

Personally, I think anything before AVX is old hardware. Doesn't mean that I don't see uses for older hardware. I just replaced my old hardware because it died, not because I wanted to. It was perfectly capable for everything except keeping up with the Jones's (Michael's benchmarks and new games).

But on a performance per watt and efficiency standpoint 32mn and up stuff should be considered old.

[Azrael5]

Who uses legacy hardware has to use legacy operating systems. The newest hardware cannot be penalized by the anachronism. SSE2 is 20 years old nowadays, and we are debating if a software must be based or not on SSE2. It's stupid enough. The hardware based on early sse2 costs 20$$$, in some cases it is also dismissed and free. Mesa could be integrate also up to sse4.x standard. Assumed that the best solution is the flexibility way got by a preliminary test of the hardware specification in order to apply the best solution for each specific hardware system.

[TemplarGR]

Just curious: what do you mean by "you people"?

People who want bleeding edge software that has stopped introducing new features for their hardware many years ago to keep supporting their hardware.

[DanL]

<more inanity>

Fsck you.

[CochainComplex]

To be fair, Microsoft also enabled SSE2 on Wndows 7 through security update in one kernel component so it has been bluescreening for years if you have 32bit without SSE2. Very few people cared.

well...on the other hand a lot of Windowsusers dont even know about such issues....considering that the majority thinks that pcs kinda "wear out" when they age. If a bluescreen pops up usually that means to them the pc is over its zenith and is "just old".

[mangeek]

Big gains can be made by rewriting certain functions in inline assembly using newer instructions.

I think that's what they're talking about. Raising the baseline so they can write parts of the code itself for modern SIMD instead of depending on compiler intrinsics, auto-vectorization, or multiple blocks of code to do the same thing.

[mangeek]

Simply stated, there are lots of industrial systems... meant to last decades or longer and use some sort of display for interaction.

Those systems aren't ever going to run Mesa 21.2. They're going to run old versions of LTS distros, and likely never be upgraded. When they can't be patched, the vendor will recommend sequestering them off with network rules and host firewalls.

I deal with a LOT of these types of systems, and I don't think I can recall any of them making a change like upgrading from Ubuntu 16.04 to 18.04 for already-deployed hardware.

[Raka555]

If people played around with the compiler optimizations like -march=native , they will know that it does not make a spectacular difference.
In reality newer machines were not held back that much.

Big gains can be made by rewriting certain functions in inline assembly using newer instructions.
That can still be done in a way where a generic version is available for architectures which don't have the said instructions.
gcc had support for FMV since about version 4.8
Compilers and linkers are smart enough to do dead code elimination to remove the bloat.

In essence the best way to gain from newer instructions is to use a distro like Gentoo that builds everything from source.

[microcode]

What hardware from before 2003 is even currently supported by Mesa?