Ubuntu MATE / Studio / Budgie All End Their 32-bit ISOs For New Releases

[dungeon]

They were all 64-bit in hardware.

https://en.wikipedia.org/wiki/Intel_Compute_Stick

Look them up on the Intel Ark site, all 64bit.

There was 32bit Quark CPUs even in products of year 2016.

https://en.wikipedia.org/wiki/Intel_Quark

But yeah these are not intended for Desktop, usually runs embedded OSes, Yocto or such Altough can't say for sure, one never know.... there might be someone run these on usual distros

[leipero]

While there are certainly a few people still using this hardware, chances are you don't need the latest Ubuntu release on your 15+ year old hardware.

Exactly, for those Users, Ubuntu 14.04 with support till mid 2019 and Ubuntu 16.04 with support till mid 2021, so no problem here in reality. As for others mentioning retro Gaming, yes, old hardware is better in some (or even most) cases, but GNU/Linux for retro gaming is just stupid idea, especially on old hardware, for such tasks you run native OS for those games (meaning DOS, Windows 9x etc.) in offline mode (large majority of old games do not have online DRM, and those who have CD based DRM are likely cracked already so you can use 3rd party crack to spare yourself of all that CD-ROM nonsense, and yes I am endorsing "piracy" regardless if you own the game or not, and can give tons of reasons why everyone should, so take it whoever disagree with it lol).

For anyone who for whatever reason want to use old hardware with new software, that's just terrible idea, most applications for the web dropped support for 32-bit OS's, so what's the point? If offline, support is really irelevant, who cares if it is supported if you are using it offline? You can always find or build packages for the specific need and call it a day.

[dungeon]

From the other side distros or at least upstream should start dropping support for generic x86 64bit... yeah, really 64bit is kind of more than decade with us, which is too much

I expected Intel and AMD to promote new 64bit arch which include AVX2 CPUs as a minimum requirement, that way distros would follow a suite and optimize everthing for these...
Altough one never know, they might plan 128bit CPUs sooner Even RiscV could be 128bit already

Ryzen 5 128bit, no 32bit compatability and 64bit obsolete... why not, that would gain envelopes

[kneekoo]

64bit is kind of more than decade with us, which is too much

Hush, someone might just take you seriously.

I expected Intel and AMD to promote new 64bit arch which include AVX2 CPUs as a minimum requirement, that way distros would follow a suite and optimize everthing for these...
Altough one never know, they might plan 128bit CPUs sooner Even RiscV could be 128bit already

Honestly, if everyone would be aligned to AVX2+, that would be awesome. And if everyone had a minimum of 8GB of RAM, even more awesomeness. But it will be quite a while until we'll see brand new PCs with no less than 8GB RAM. Still many years from now, unfortunately.

[dungeon]

Hush, someone might just take you seriously.

Well, i am serious even when i laugh sometimes

I am not that young, so this is normal too me as i remember 8bit CPUs, then 16bit, these 32bit and this old current 64bit... and i can only say this 64bit taked a lot of time, so it is time for replacement and no that won't surprise me when it happens

Honestly, if everyone would be aligned to AVX2+, that would be awesome.

To be honest i expected Intel and AMD to promote these couple years ago... that is what ClearOS is doing of some sort. But it does not do a right thing, they should together just ask all distros to create new arch with a new requirement, so who ever fullfill requirement to just run this new arch...

And now because they didn't do this, we can only guess that probably 128bit is near

[leipero]

dungeon While you are sarcastic, there is a huge difference between 32-bit and 64-bit, the main reason (as you probably know) why they moved to 64-bit is theoretical/practical RAM allocation limit, and you go from 4GB to close to 17 million TB, while almost everyone back in 80's did understand that will be a problem much sooner, now, you can't really say the same. At this pace, it's very possible that 64-bit will not become problem in decades and very likely century, unless whole principle of computing changes, and if that happens, topic as "moving to 128-bit or whatever" would be irelevant. Especially if you take into account that supercomputers and servers address memory in modular way, because of both physical/software limitation and practicality/efficiency, so that problem is really so far away, if ever becomes a problem in the first place (likely not).

[dungeon]

I expect 128bit in next decade and on reasons... well, they could find whatever other reason as a main reason to switch to 128bit From 32to 64bit was a memory, now for 128bit would be something else

You know how that goes, find and marketize just one useful reason and they will do it Just by making first 128bit consumer CPU is already enough for the masses to jump

[leipero]

dungeon I understand marketing, but there is no real reason to move from 64-bit afaik, and making compatibility with new extensions for both 64 and 32 bit makes it even more complicated that I don't see it anyone doing it for that purpose.

[dungeon]

dungeon I understand marketing, but there is no real reason to move from 64-bit afaik, and making compatibility with new extensions for both 64 and 32 bit makes it even more complicated that I don't see it anyone doing it for that purpose.

RiscV already have 128bit, Linux is most used on supercomputers and if supercomputers feel that they are capped in progression... what you think will happen?. From RiscV papers:

"At historic rates of growth, it is possible that greater than 64 bits of address space might be required before 2030."

Is that enough reasoning? These pushovers does not happen just because me or you are not capped on our Desktops

[leipero]

dungeon Yes, but that is prediction "at historic rates", and in last decade plus it's not stagnant but it did slow down substantionaly. If I understand it corectly supercomputers use modular approach, so really it should not be required at all (there's always a posibility that I do not understand it in the right way). My point here is that there are real physical limitations approaching 5nm regardless of materials used, so what will come next, it's hard to say, if it's quantum computing, then it will require completely different approach in memory addressing, so whole talk about X-bits will become irelevant.

At current rate of growth, even supercomputers are far away from 18 million terabytes in 2050, let alone 2030, and our little desktops will probably never use that amount of memory in billion years .