Welcome Guest | Sign In

Linux, the 386, and Days of Auld Lang Syne

By Katherine Noyes
Dec 20, 2012 5:00 AM PT

The year 2012 may not yet have reached its final conclusion, but here in the Linux community another kind of curtain was recently dropped for the last time.

Linux, the 386, and Days of Auld Lang Syne

It's the end of the line for Linux's support of Intel's 386 chip, specifically, and tears are being shed across the land -- or not.

"This tree removes ancient-386-CPUs support and thus zaps quite a bit of complexity," wrote developer Ingo Molnar when submitting the change last week. "Unfortunately there's a nostalgic cost: your old original 386 DX33 system from early 1991 won't be able to boot modern Linux kernels anymore. Sniff."

Not everyone got choked up, however: "I'm not sentimental," wrote Linus Torvalds, for example. "Good riddance."

Is it the best of times? Is it the worst of times? Will anyone even care? That's what bloggers on Slashdot and across the land have been trying to decide.

'I Think It's a Good Thing'

"This development kind of reminds me of the New Year," said Google+ blogger Linux Rants over a Tequila Tux cocktail down at the Linux blogosphere's seedy Punchy Penguin Saloon.

"It gives us an opportunity to look back on where we were, reflect on where we are, and look forward to where we're going," Linux Rants explained. "It's kind of a sad moment to look at what's moving into the past, but when all is said and done, it's time for us to move forward and go on to bigger and better things."

The end of 386 support is "the passing of an era, but I think it's a good thing," he told Linux Girl. "386 support in many ways holds us back and makes things more difficult, so getting rid of it really makes sense. Seriously, how many people are going to be affected by this in a negative way?

"Still, kinda sad," he concluded.

'This Makes Me a Little Sad'

Indeed, "this makes me a little sad, because my first Linux machine was a 386DX25 with 8MB of DIP DRAM onboard, on which I ran Slackware 2.0," Hyperlogos blogger Martin Espinoza began. "On the other hand, I understand; I was running kernel 1.1.47 back then!

"If you're running such an antiquated processor, do you really want the latest kernel, anyway?" he pointed out. "You might not even want a 2.x kernel, let alone 3.x."

Most of the 386s in the world today are probably running DOS and performing industrial control tasks, Espinoza suggested.

"It can be difficult to maintain a proper sense of perspective if you started computing on something with less power than a 386 (for some of us, it was a lot less), but even NASA likely isn't using them any more, as Intel now makes radiation-hardened Pentiums," he pointed out.

'Thanks for the Memories'

Blogger and educator Robert Pogson had similar memories.

"My last '486 died around 2000 when it was dropped on a runway," Pogson recounted.

"In schools, the oldest machine I have seen recently was 15 years old, so it is time '386 died," he opined. "The few antiques left can run reasonably well on older versions of GNU/Linux but not on new ones just because of the bloat.

"Good-bye and thanks for the memories (Lose 3.1 and freezing)," he added.

In fact, "I was still using that '486 when I discovered GNU/Linux ran solidly on hardware which Lose '95 froze constantly," he told Linux Girl. "I can still build an older kernel and an older distro to run something in a few megabytes if necessary, but I doubt it will ever be necessary again."

'History Moves On'

Similarly, "I can't imagine anyone will miss the 386," consultant and Slashdot blogger Gerhard Mack agreed. "Most distros have dropped support for it ages ago, and some have even dropped the 486.

"Uselessly old CPUs that only serve to complicate the rest of the code are better off dropped," Mack concluded.

And again: "I am an enthusiast of saving old machines from the recycling pile with GNU/Linux, but since packages are continually evolving, it is becoming increasingly difficult to work with any (even light) distribution on a Pentium III era computer," Google+ blogger Gonzalo Velasco C. agreed. "Imagine that with a 386 DX33!!!

"History moves on; also technology does," he added.

'Good Riddance'

One more time: "What took them so long?" wondered Slashdot blogger hairyfeet. "Who still uses 386 anywhere for anything but nostalgia? You can find several orders of magnitude better than a 386 on an average street corner!"

In fact, "I think they should throw out everything before 686 as there really isn't a point -- nobody is gonna be trying to run a modern OS on a Pentium 1, much less a 386 or 486, not to mention that even if the kernel supports such an old CPU, how much of the modern software required to make the kernel actually do anything is gonna run on a CPU that is THAT old without it having each action literally counted in how many hours it takes to perform?"

Bottom line: "I agree with Torvalds -- good riddance," hairyfeet concluded.

'A Non-Issue'

Last but not least, "If you maintain support perpetually for every piece of kit ever invented, the software just grows and gets bloated," Google+ blogger Kevin O'Brien explained. "So you need to trim the list occasionally, and this seems perfectly justified."

Indeed, "the great benefit of Free Software is that if anyone really needs this, they can use an older kernel, or they can hire a programmer to add in support, or if there is enough demand someone will probably provide a kernel that has it built-in," he added. "This is really a non-issue."

Katherine Noyes has been writing from behind Linux Girl's cape since late 2007, but she knows how to be a reporter in real life, too. She's particularly interested in space, science, open source software and geeky things in general. You can also find her on Twitter and Google+.

Facebook Twitter LinkedIn Google+ RSS
How do you feel about accidents that occur when self-driving vehicles are being tested?
Self-driving vehicles should be banned -- one death is one too many.
Autonomous vehicles could save thousands of lives -- the tests should continue.
Companies with bad safety records should have to stop testing.
Accidents happen -- we should investigate and learn from them.
The tests are pointless -- most people will never trust software and sensors.
Most injuries and fatalities in self-driving auto tests are due to human error.