Arch Linux Developers Discuss Idea Of Providing An x86-64-v3 Port

[NateHubbard]

but why are they pushing AVX as minimum?

Because it's 10 years old now?

[Mat2]

> but why are they pushing AVX as minimum?
Because it's 10 years old now?

Yes, roughly 10 years ago the first processors with AVX (not AVX2) shipped, but Intel has continued to manufacture CPUs without support for AVX for many years.

After all, a 10 year old computer is still perfectly capable for desktop use, especially when equipped with an SSD.
Not everyone can afford an i7 with 64GB of RAM and RTX 2080.

[NateHubbard]

Not everyone can afford an i7 with 64GB of RAM and RTX 2080.

Instead of a 10 year old computer that is still "perfectly capable", you could get something incredibly cheap that would be faster and use less power.

[Mat2]

Well, "incredibly cheap" is a relative term - fortunately not so much for me, but for some still.

Edit: I've got a newish laptop, but around me is an older PC and a laptop from around 2008. They happily cope with anything thrown at them (browsing, remote learning, etc.).

[skeevy420]

Ehh ? I use UEFI with GRUB, are you saying they will deprecate GRUB in v3 ? Could you point me to where you read this ?

I didn't read that anywhere. It's all hypothetical.

AFAIK, all x86_64_v3 systems are UEFI. Because of that GRUB is unnecessary and can be replaced so no more /boot limitations that it brings. FWIW, GRUB already runs off the EFI partition on UEFI systems so the limitation isn't GRUB so much as it is needing an MBR loader that looks for a Fat32 EFI partition and goes from there.

As a long, long time GRUB user, I feel like I'm shitting on it when I'm just upset at the /boot limitations that have to be considered when legacy systems are considered. Hypothetically, when you/we don't have to consider "those people" it opens up a lot; especially the aspect of being able to shift the early disk initializing off of GRUB and onto the kernel so we don't have to reinvent the wheel twice to boot the system...only with Fat32.

[skeevy420]

Hope they go with v2, if i understand it correctly, v3 would remove support for my trusty T530 thinkpad

IMHO there should have been 1 through 5. 1 and 2 the same, 3 being basically Ivy, 4 adding AVX2/Skylake, and 5 adding in AVXBBQ. But you'd be covered by that v3

[AmericanLocomotive]

Yes, roughly 10 years ago the first processors with AVX (not AVX2) shipped, but Intel has continued to manufacture CPUs without support for AVX for many years.

After all, a 10 year old computer is still perfectly capable for desktop use, especially when equipped with an SSD.
Not everyone can afford an i7 with 64GB of RAM and RTX 2080.

10 year old PCs are Nehalem based, and are honestly getting pretty pokey these days, especially the mobile parts. My i7-2630qm laptop (which was a solid mid-pack quad-core mobile Sandy Bridge part in 2011) isn't the fastest on a lot of modern websites or even HD video playback, and it has an SSD. Even 2009 era pre-Nehalem Core CPUs are barely tolerable for general web browsing, with the lowest end Celeron chromebook chips offering similar or better performance.

It's like complaining about software no longer supporting your 800 MHz Pentium 3 in 2010 when people are running 4 and 6 Core Core "i" series chips, or whining that you can't install Windows XP on your 486 in 2001 when everyone else is running 1GHz Athlons....

Technology marches on. It's about time our software starts being targeted and optimized for CPUs that are 7 years old, instead of 17.

[Paradigm Shifter]

There are various options, none of which are perfect, but the two that would stand out for me as better solutions (at least to try to keep the maximum number of people happy?) would be the fat binary route, or the "stay with x86_64_v1 and make compile-from-source an easy option for those demanding ultimate speed".

On fairly recent hardware, Michael keep demonstrating that even kernel compiles are getting to the point where it's not unrealistic, on systems with extremely frequent updates.

[TemplarGR]

The workload that got the most boost in the benchmark was John the Ripper with MD5, perhaps because noone really cared to optimize it for AVX/AVX2. Other then this, the performance gains were modest - around 10% - 20%. I have tried to exclude binaries that use much asm, because there was really no point in benchmarking them.

10%-20% are NOT "modest" gains. They are the equivalent of a generation upgrade or two.

[Sonadow]

Yes, roughly 10 years ago the first processors with AVX (not AVX2) shipped, but Intel has continued to manufacture CPUs without support for AVX for many years.

After all, a 10 year old computer is still perfectly capable for desktop use, especially when equipped with an SSD.
Not everyone can afford an i7 with 64GB of RAM and RTX 2080.

Not just Intel. AMD also makes entry-level processors without AVX.

This is getting ridiculous. The goal is to make distributions more modern by catering to newer processors, and yet new entry-level processors released less than 18 months ago get shafted from these 'modern' distributions just because they lack AVX.

Imagine how idiotic and lopsided it is that a 2013-era Haswell can boot and run a x64-v3 optimized distribution but a Comet Lake processor released this year will fail to do so.