SoylentNews is people
SoylentNews
posted by
Fnord666
on Wednesday January 10 2018, @01:04AM
from the found-at-the-bottom-of-the-closet dept.
from the found-at-the-bottom-of-the-closet dept.
Submitted via IRC for TheMightyBuzzard
[Yeo Kheng Meng] had a question: what is the oldest x86 processor that is still supported by a modern Linux kernel? Furthermore, is it actually possible to use modern software with this processor? It's a question that surely involves experimentation, staring into the bluescreen abyss of BIOS configurations, and compiling your own kernel. Considering Linux dropped support for the 386 in 2012, the obvious answer is a 486. This supposition was tested, and the results are fantastic. You can, indeed, install a modern Linux on an ancient desktop.
Source: https://hackaday.com/2018/01/07/go-retro-to-build-a-spectre-and-meltdown-proof-x86-desktop/
This discussion has been archived.
No new comments can be posted.
Go Retro in Order to Build a Spectre and Meltdown-Proof x86 Desktop
|
Log In/Create an Account
| Top
| 43 comments
| Search Discussion
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
Most of our lives are about proving something, either to ourselves or to
someone else.
(Score: 5, Informative) by takyon on Wednesday January 10 2018, @01:19AM (6 children)
Isn't the relevant question: "What is the most modern/powerful x86 processor that is not affected by Meltdown and Spectre" rather than "What is the oldest x86 processor I can run modern Linux on"? Or do you want to have a slower experience?
There are much newer Intel chips than the 486 that are not affected by Meltdown. Probably no IME too if you throw that criteria in. Although many Intel chips since 1995 are affected by Meltdown, pre-2013 Atom are not. Wimpy as it may be, a 2012 Atom will demolish a 1989 80486.
This just looks like a cool project that was repackaged as Meltdown/Spectre clickbait on Hackaday.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 3, Insightful) by Hyperturtle on Wednesday January 10 2018, @03:04PM (2 children)
yeah
Why build a machine that is immune to this if one that is patched is faster and runs modern stuff anyway?
Go retro to go retro, or to fulfil a specific requirement (personal or business.. hobbies are cool too you know). But this article just seems to be clickbait shaped around recent tech news.
My understanding is that Pentium II's and newer are suspectible, give or take a few in the family around 1995. Pentium Pros and similar competing cpus are not impacted. I have an HP L4 Netserver with four Pentium III 500mhz cpus and it's affected.
But that doesn't make my HP Proliant with four Pentium Pro 200mhz cpus a good choice compared to that 4x500 P3 computer. Nor is it compared to an Athlon XP 2200+ which is a faster CPU and cheaper to run. It can even be overclocked to some extent (the servers are pretty much as-is). But the best option is that patched i5 or i7 you already have from 5 years ago or last week.
Whatever you have is good to keep using. Just decide to patch it or not and take precautions. There is no building a new system that performs like a modern computer when using old cpus to "not be impacted. There won't be a performance solution around this problem until they release CPUs that don't have the flaws.
AMD is the best bet, but their modern CPUs are suspectible to Spectre-- but not Meltdown. 1 out of 2 isn't bad i suppose.
More to the point, my c64 with wireless internet access with the SD flash card that replaced the floppy drive and a flash rom cartridge that loads programs so fast I am still in denial (minutes to seconds--faster than I ever thought was even possible on the platform) STILL is vulnerable to the WPA2 'krack' or whatever vulnerability that came out not long ago.
I am seriously not going to dmz my c64 because its wireless capabilities are not likely to be patched any time soon, despite my ability to connect to it via a USB connector I have to copy files faster than over the wifi (it's still limited to... like 38400bps or so, but realistically I telnet out from it to boards at 9600 since I can't read much faster than that and the scroll/pause isn't always reliable depending on the BBS answering...)
The takeway is that there is no escape. It will all be vulnerable. It is better to react and defend, unless it's just for fun and the risks are small and costs are negligble.
There's value in building an old system--there's a logic required that doesn't often apply today for many young IT people. They just get a sealed box that gets replaced after a while. Knowing how it all works, and with old gear, can teach a lot... how it works, how to troubleshoot when it doesn't, how light overclocking can make big differences in performance... and most of all patience. because it's just as hard now as it was then, since it is hard to find drivers online and back then it was hard to find drivers online, too.. (Of course for me I do it because I am a masochist, but that's a different problem...)
That said I wish we didn't have to dream up workarounds for modern problems, and I have systems I am reluctant to patch until I see more evidence of how it impacts peformance. I haven't seen too much about the impact overall, just specific worst case scenarios. But in the end I will end up patching nearly everything that needs it because it's better off slower than compromised.
(Score: 2) by takyon on Thursday January 11 2018, @06:59AM (1 child)
We'll end up with some neuromorphic + quantum chips in PCs and phones mounting an active defense against intrusion. Or just becoming a more dynamic and interesting way to pwn the system.
The good news is that the desire for convenience and riskiness of technology means humans will continue to be employed for at least a couple of decades more.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by Hyperturtle on Thursday January 11 2018, @07:22PM
I've been wondering if the transition to mostly robot manufacturing and mining and even cars takes place, who has been planning on how to deal with the resultant loss of revenue because of people that can no longer afford to buy the products being manufactured? We're not at that tipping point, but some industries are closer to it than others.
They can't keep raising the prices of stuff that isn't selling in order to meet their profit projections -- eventually, people with money have to buy stuff to make those robots worth the price of not ever being a consumer and never contributing to a tax base.
But I bet if we pooled our resources and made a bitcoin mining robot that was outsourced--we'd make millions!
(Score: 2) by Wootery on Thursday January 11 2018, @10:57AM (2 children)
The question is broader than this, no? Why commit to x86? If I understand correctly, your average 'unpatched' ARM/PowerPC chip is no better so it doesn't help to just dig out that ancient Mac, [blogspot.co.uk] but still.
Also, as this is SN: roll on RISC-V, I say.
(Score: 2) by takyon on Thursday January 11 2018, @11:12AM (1 child)
The same reason why so many are committed to x86, or even Windows: Software.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by Wootery on Thursday January 11 2018, @11:22AM
It's pretty clear we aren't talking about mainstream desktop machines or servers, or no-one would have mentioned the 486.