New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Ubuntu 32-bit Linux Builds Fading Away by 2018
I believe that some groups will step up and keep them available, but they won't get the testing and support of the mainstream, popular distros. See:
https://linux.slashdot.org/story/16/07/05/1838245/linux-letting-go-32-bit-builds-on-the-way-out
http://www.theregister.co.uk/2016/07/05/linux_letting_go_32bit_builds_on_the_way_out/
Comments
I corrected the title because it's only Ubuntu pissing on its users, not most distros.
I'd wager you'll have no problem downloading a current 32-bit Debian build in 2018.
You won't be able to get an Ubuntu build but then...you shouldn't be using Ubuntu anyway.
What is Ubuntu?
thx, debian.
Why would you want a 32-bit OS?
embedded
VMs. I find the difference between 32-bit and 64-bit on small LEBs to be quite significant.
old gear that doesn't support 64-bit. Sure, current stuff does, but lots of us still run old gear.
32-bit platforms aren't THAT old. You could still buy then much more recently than 10 years ago...
I think it's a little different than 16-bit vs 32-bit. In the 16-bit days, it was really hard to do anything practical with 16-bit memory addresses (other than embedded, where 16-bit is still quite popular) but 32-bit is a much bigger universe.
Of course, you're right that in 90%+ of the cases, 64-bit is the right choice. For now. Come back to this question in 2026 when people are talking about their 128-bit OSes and those old 64-bit dinosaurs...
Why waste memory on tiny virtual machines used for tiny jobs, as small vpn or icecast restreamer?
How much are you wasting exactly? Not enough to justify the continued support with the continual drop in memory cost, for sure.
The 32bit placebo effect isn't enough. Out with it. People who have special needs for their OS should look out for themselves, there's no justification that I can think of for distros to make, test, and support 32bit anymore. It's one thing for it to have a few uses, another for it to be justified in development alongside the rest.
COBOL has plenty of uses still, doesn't mean everyone should learn/support it.
OK, in terms of wasting efforts, I suppose that the more needed ram in 64 bit systems is minimum, in an era that 64MB vps is just for enthusiasts that are trying to push a whole cms in it, and does not worth the extra time of a distro community to develop and maintain any new edition.
If a community decides that it will stop developing 32bit systems, it's OK with me. But, as long as they continue for any reason to develop it, there always will be uses for that.
And if in Linux world, especially for proffesional usage, there is that strong argument, 64bit it is a bigger issue if you use Windows OS: a 2 core 2GB mem PC/netbook is still common for a lot of people, especially in poor or average income housholds.
A 64 bit Win 10 there, is clearly lacking performance and MS is really trying to push replacing some not so old hardware that still does the job for an average user.
Internet browsing, an Office use, skype and Kodi will have better performance (as of my experience) in Win7 x86 than Win10 x64. So, to use a better technology for the exactly same tasks, is demanding for buying a better hardware, 4+ cores with 4+ GB mem.
This is a waste of money for an average user.
So first, let's keep in mind you're writing on LET, where one of our most famous articles is how to run MySQL and Wordpress on a 64MB VM, so saying "pfft, RAM is abundant" may not resonate with the audience :-)
But more to the point...I just happen to have those numbers from some testing I did a few weeks ago:
deb8 x64 base image on DO 512: 428MB free at boot
cen7 x64 base image on DO 512: 424MB free at boot
deb8 x32 base image on DO 512: 473MB free at boot
So, 10%.
To me, that's significant...especially when you get over to 128, 96, or 64MB VMs.
I'm aware that for a lot of people, they aren't significant...e.g., my employer who routinely spins up 1TB VMs. But for a lot of people, they are significant.
Lots of justification - the same Linux kernel that powers your Cen7 box is the same kernel that powers a Raspberry Pi and the same Linux kernel that runs in an elevator's control system. Are you going to make everyone who wants to run 32-bit have their own kernel tree?
Also the amount of work to maintain this is...nil? If there's one thing C people are good at, it's dealing with the fact that their language doesn't tell them the size of an int. So every significant piece of C code starts out by determining what kind of platform it's on, and then adjusting itself, and this is all done through copy-pasted macros from the 70s. Most well-written C code compiles and runs just fine regardless of whether the underlying architecture is 32-bit or 64-bit. The current Linux kernel, for example.
What you're suggesting, essentially, is hard-coding the Linux tree to 64 bits. What happens when we get to 128-bit?
Yeah, there's more testing...but it's automated and my sympathy is...non-existent? Because the Linux kernel will be worse off if you take out 32-bit code.
WHAT!?!? The reason is that the more different types of hardware you test your code on, the more bug-free it is. Taking a kernel and running it on all kinds of different hardware flushes out bugs, erroneous assumptions, etc.
If Linux people are complaining about 32-bit x86 vs. 64-bit x86_64 then they should consider themselves whiny little girls, because other kernels run on a far broader mix. OpenBSD doesn't run on Alphas and SparcStation 10s because people are actually doing productive work on those...they compile daily and test because it exposes bugs. Given the high cost of maintaining that hardware and power, it's an approach that is frequently challenged but every time, they've continued maintaining that ancient gear because they find lots of bugs they would not otherwise have found.
I realize Linux is not as quality-obsessed as others, but I think being forced to support both 32- and 64-bit (on ONE platform, boo hoo) is good for Linux.
You can write platform-independent COBOL code that compiles cleanly on 32-bit or 64-bit, x86 or POWER :-)
@raindog308 I agree with all that, but @Jarland has a point: the maintain costs (time, effort) to continue to provide x32 support. With a thing in mind: Ubuntu!
I think Canonical has some targeting groups. And the major may be the daily pc users. They are trying to provide (at least, they advertise so) a Linux distro friendly for the end user, so, maybe Canonical is approaching Microsoft's attitude. Newer technology needs newer machines and newer machines need newer technology. With this in mind, end user is being pushed to replace his gear every 3-4 years, instead of keeping a 10 years old computer for similar job. This brings profits for MS, for intel, for the hardware industry.
To sum up: Debian or Centos community will probably continue to maintain 32 bit kernels because they are targeting in different groups than Canonical. (we are all aware of Ubuntu's owners and their try to grab money as trademark for using their logo)
Right - it makes sense for the greedy blocksucking vultures at Canonical a distro like Ubuntu to only focus on desktop, but I was talking about the Linux kernel.
I wish MSFT and Canonical would just get a hotel room.
Another pivot is that as a generality across the population, people aren't running new operating systems on very old hardware. I'm currently typing to you on a machine running an 8-year-old 64-bit CPU that's four core/eight threads and can support up to 32gb of RAM, and have no realistic intention of upgrading any time soon, however, I'm running Windows 10 on it. This is certainly not the norm.
The reality is that people buy a pre-fab computer and use whatever version of Windows is installed on it until that computer's lifetime has ended, be it two years or 10 years, especially in business. One of, if not the largest manufacturer of private-label automotive lubricants in the United States was still found to be running Access 2.0 databases, from 1994. In their view, this wasn't a problem, and indeed it only manifested itself as an actual problem when Access 2.0 support was dropped in Access 2013, released literally 20 years later.
Canonical hasn't turned a profit in their existence, and both dropping support for 32-bit distributions AND the recent stab at our favorite French datacenter for royalties makes me think they're hurting for cash.
It looks like Ubuntu 16 is supported until April 2021. If you're still using a 32-bit-only processor at that juncture, it would most likely be akin to still attempting to use MS-DOS in the modern era.
Using MS-DOS would certainly save you on RAM usage.
LOL. What?
April 2021 is less than 5 years from now. So if that's true, then surely you're not using any piece of computing hardware made in 2011 or earlier...?
Like, oh I don't know, maybe an iPad 2 (which as a side note is 32-bit)? Release in March 2011. I can't quite agree that my kids using an iPad 2 is like my use of MS-DOS, which ended sometime around 1994.
We've had 64-bit desktop processors since ~2004. I'd say they became more "mainstream" by 2006 with the release of the Intel Core architecture. So by the time 2021 rolls around, we've had 15 years of 64-bit processors released to the masses.
If in 2021 someone really wants to run a cutting edge OS on their over-15-year-old computer, they would be better served by buying a new computer. More likely it's someone like us who will either build a new system, or upgrade it over time.
And indeed, Ubuntu is going in the correct direction by choosing to drop support for over-15-year-old computers. I would imagine that 15 years after the iPad 2 release, your children will no longer be using it in 2026.
Evolution progresses regardless of FUD. We're deploying servers with 384gb of RAM like we deployed single-core dual-processor Xeons with 32gb of RAM 15 years ago. I imagine in the next 5 years, servers will be into the terabytes of RAM range, and 1 GB VPS, if not more, will be the new 128mb VPS.
If only we could spend all this resource on figuring out how to make Wordpress not be shitty.
Update your plugins and use few. Things you have to do with every CMS. Only reason WordPress gets it as bad as it is is because it's the most used.
I know this may be considered a stupid question (sorry if you feel offended), but Is it a hard challenge for providers to remove Ubuntu from their OS choices towards customers?
If Canonical can ask others for trademark, money, or 32bit restrictions, providers can simply... ignore them. We can simply stop using Ubuntu.
Maybe. But you miss my point. You say 32-bit computers in 5 years will be like MS-DOS today, which is nearly the same as saying that a computer from 5 years ago (an iPad 2) is like MS-DOS. Really?
You have a valid point, I just think 5 years is a lot sooner than you realize. Windows Server 2008 is still very widely deployed in datacenters...that's 8 years ago. How many still run Windows XP? That's 2001. Etcetera...technology doesn't move that fast.
Not sure where the ad hom came from, but whatever.
When I created the thread, I thought about incorporating Ubuntu in the title, but also noted the SUSE discussion as well, which is why I wrote it the way I did.
My primary interest is from the perspective of a low-end VPS user. My VPSs are low priced, and thus have a limited amount of RAM. Running a 32-bit OS on my VPSs reduces RAM requirements for my VPSs.
I predicted that 32-bit support would wane over time in several posts that I wrote on LET over the years, but this is the first that I have seen it in the media. It did not take mystical clairvoyance to see it coming.
I believe that Ubuntu is a bellwether with respect to the abandonment of 32-bit support for Linux. I predict that other mainstream Linux distributions will follow, some faster than others. Some dedicated people will keep 32-bit forks going for their own reasons, the way we still see support for PowerPC, DEC Alpha, and embedded systems.
I'll comment on the tree instead of the forest
COBOL v6.1 was released by IBM in Feb 2016
On the subject, I don't think there's anything to argue about.
32-bit processors are alive and well (despite the "evolutionary" hate from @Damian ) and they too need an OS. Canonical's target group isn't people who work with 32-bit processors.
In other words - support for 32-bits is necessary but not by everyone. Case closed, you can send me your extra bitcoins at 1K7WevFtGwp8tJHZu4ubqjB5wEBErsoDLw for the time sav... oh wait
Of course there is the Y2038 problem that we should worry about, so sooner or later the 32 bit support must end. But i'd say it's too soon for now, there are still plenty of hardware and software that will run better in 32 bit mode.
Good point!
I was curious this morning and read that there is work being done on 32-bit Linux to support 64-bit time_t:
https://en.wikipedia.org/wiki/Year_2038_problem#Solutions
But the problems run deeper...you're into filesystem code, etc.
Of course, your 64-bit OSes will break at 15:30:08 on Sunday, 4 December 292,277,026,596 AD...
Just ran a test install on a couple of DO vpsses, on 32bit it uses 16mb less ram than on 64bit. I dont think this 16mb will matter in 2018 (or even 2021 when 16.04 is EOL).
Did save 180 meg of disk, which is nice
Is a simple OS installation comparing 32-bit vs. 64-bit RAM usage a valid test?
Don't the real RAM savings appear when you start running real-world applications on the servers?
Installed and ran the gameservers i usually run on these boxes, ram savings remained at 16mb.
Debian just dropped support for some 32-bit systems https://lists.debian.org/debian-devel-announce/2016/05/msg00001.html
Fedora no longer offers 32-bit Server editions
OpenSUSE no longer offers 32-bit with LEAP (SUSE 12 Enterprise supports only 64-Bit) with the OpenSUSE chairman saying "Over years we've seen a steady decline of 32bit downloads of openSUSE. Even when our overall download numbers have more than doubled, 32 bit downloads has declined, both as as absolute number and as a percentage of Downloads. They're now a small fraction of our users And yet 32bit support doubles our testing burden (actually, more so, do you know how hard it is to find 32bit hardware these days?)"
edit forgot to add that RHEL 7 doesn't natively support 32-bit hardware and only provides 64-bit ISOs - https://access.redhat.com/solutions/509373
Ubuntu is certainly not alone in this.
Regardless, if by 2021 (Ubuntu 16.04 LTS supports 32-bit until then and Jessie LTS will support until 2020 if they include the support) you still need a 32-bit linux OS then I'm sure you will still have options.
32-bit users will be just fine, there's still a lot of time and things change
.
Seems later 90's chips.
If Linux kernel will drop support to 32bit, people who needs 32bit can go to NetBSD?