Linux 6.0 Merges The AMD Performance Fix For The Old "Dummy Wait" Workaround

[Sin2x]

Half a monthly income is not that much for a new PC, that's definitely still affordable.

And that's like you said at minimum wage and it would be a single person, but two person household and even more are much more common in third world countries, so it would be only 25% of monthly houselhold income for a 2 person household. Very affordable.
You are definitely ignoring that saving money exist and that you use the PC for at least 5 years. So saving 10% of your income for 5 month or 5% for 10 month is not as unreasonable as you make it out to be. And that's again for a single person household at minimum wage and not being able to sell their old PC to finance some of the price.

You are also ignoring that you could still run an older Linux kernel and still a new software stack, even Linux 2.x should already support your hardware flawlessly, just wanting to run the newest kernel "because I want to" is ridiculous. But even then as of right now you could use Linux 5.10 up to December 2026. As well as all the other potential hardware you could upgrade to.
Even on the raspberry pi side of things, like a PI zero or Raspberry Pi 3.

Quite honestly this complain is just not justified.

You clearly have no children.

[NateHubbard]

I know there is the usual discussion about ancient hardware and possible performance issues with this patch, but as someone who patched it yesterday with a Zen 2 processor, my initial report is that you're not going to notice anything.

[coder]

You are also ignoring that you could still run an older Linux kernel and still a new software stack, even Linux 2.x should already support your hardware flawlessly, just wanting to run the newest kernel "because I want to" is ridiculous. But even then as of right now you could use Linux 5.10 up to December 2026. As well as all the other potential hardware you could upgrade to.

This is tricky, because web browsers continually need security fixes and have some amount of dependencies on the underlying system. I don't know how old a kernel/glibc current web browsers support, but there's certainly some limit.

Even on the raspberry pi side of things, like a PI zero or Raspberry Pi 3.

I don't see any Pi 3's in stock. The cheapest option I see is a Pi 400, costing $139. A regular Pi 4 is a bit more than that.

As for Pi Zero, is that even usable for desktop computing stuff? Can you get a case for it with all the required connectors? I thought it was intended only for embedded computing applications.

[sinepgib]

This is tricky, because web browsers continually need security fixes and have some amount of dependencies on the underlying system. I don't know how old a kernel/glibc current web browsers support, but there's certainly some limit.

I wouldn't expect glibc to require too recent a kernel, and the same goes for browsers. Specially browsers are supposed to be portable between OSes. They're likely more limited by vector extensions and GPUs than the kernel version, for which the hardware itself is limiting you.

[DRanged]

My Duron 800 says you're wrong, while ripping 3 CDs with EAC over wine, checking them for correctnes, auto naming them and converting to flac.


Because old hardware is here and costs nothing while new stuff needs to be purchased.


Arch Linux 32 is currently at 5.18 but will ship 6.0 if its ready.

Just to be cheeky Debian testing (bookworm) is already on 5.19

[jorgepl]

It was 20 years ago lol.

You are basically mixing unmixable. K7 are 32 bit CPUs even without SSE2 support, K8 are 64 bit CPUs. That's an extreme difference in terms of performance and distro support.

That's not true. I had multiple Athlon XPs in past - sold them as soon as I got my hands on 64-bit first on socket 754 then on 939. My father still uses a notebook with Pentium M (comparable performance from the Athlon XP era) for OBD car diagnostics and it is a complete disaster under any distro that call itself lightweight. Antix, MXLinux, even Debian minimal install struggle just to decently run a desktop. Not saying about running that apps under Wine. And this notebook already has 3Gb of dual channel DDR2 and Intel SATA SSD. Eventually I just installed Windows 7 there because every time I tried to do minimal modern (not a 15 y/o distro) Linux install it turned out to be a complete waste of time. My experience shows that "outdated" for Linux nowadays means 2 cores 64 bit as a bare minimum to run desktop with browser at least bearable.

Using 20+ y/o hardware as a server that can die at any moment? Unless you don't have anything else and you have free electricity.

I strongly struggle to believe Windows 7 runs better than AntiX

[rob11311]

You might also stop for a moment and think about your priorities. Should we focus on current PCs when we develop our current kernel, or junk from 20 years ago? Hmmm... Tough one indeed.

That's a very narrow attitude, Linux supports a huge vareity of processors, many of them that were never running desktops and kernel developers have benefited hugely over time by scaling well and NOT killing things for trivial reasons.
By your own logic, Linux Kernel devs should drop "current PCs" entirely and focus on HPC, server & Android devices where they have the most market penetration, and simply leave the dying legacy Wintel desktop platform to Microsloth's WSL.

[rob11311]

Wait, what about the old AMD systems with VIA chipsets that were supposedly affected?

I built one of those horrible things, Via chipsets were unreliable especially with Windows. AMD hardware improved massively once they developed other chipsets for the Athlon.
As it stands on 64bit you don't want extra tests on AMD systems, do you? Why risk branch misprediction and pipeline stalls trying to idle?

Those old AMD/Via chipsets IIRC we had to use APM not ACPI which was new fangled and utterly broken for years as what Windows did rather than the ACPI spec was what BIOS developers went by, only checking if Windows booted. That meant the ACPI developers had to change approach and use reverse engineering to make ACPI actually work.
If there are users who have 32bit AMD/Via using ACPI and track supported kernels, they need the Intel workround, then they can flag up a regression, but I doubt it as the x86 32bit toolchain started breaking down at least a decade ago. Distros have been dropping support.

I suggest you get on your Via chipset box and install the last kernel, then test a new one and report a regression.

Now note, presently only on Intel is it calling an ACPI function, punishing them for the fault their hardware required a work round for.

[sinepgib]

That's a very narrow attitude, Linux supports a huge vareity of processors, many of them that were never running desktops and kernel developers have benefited hugely over time by scaling well and NOT killing things for trivial reasons.
By your own logic, Linux Kernel devs should drop "current PCs" entirely and focus on HPC, server & Android devices where they have the most market penetration, and simply leave the dying legacy Wintel desktop platform to Microsloth's WSL.

Prioritizing something doesn't necessarily mean dropping everything else. It means, in case of conflicting needs, you bias in favor of your priority.

[birdie]

Not included in just released 5.19.12