That’s what I thought too. But there deffinitely seem to be a lot of people claiming that something redicolous like 90% of their PCs will stop working with that change. I believe the only ‘PCs’ I have that ‘dont have avx2’ are VMs on my NAS witch itself does have avx2, but I selected some random 2nd or 3rd gen xeon when setting up ovirt.
And you can count the pie I guess.
Either way I dont install or plan to install fedora on any of them. Its on the PCs I use (so the ‘workstation’) and thats it.
But the real question I think should be why must moving to avx2 on an OS level must mean dropping functionality on older processors entierly?
4th gen intel was a while ago. How many subsequent processors must implement the same feature for it to be implementable?
It would be cool if this could be done with some dnf magic. But its likely not feasible to maintain 2 versions effectively. If it was Im sure the proposal would have been different.
If you wanna know possible performance gains you should look at Intels “Clear Linux” distribution benchmarks. That distribution being a thing seems to be the motivation for such changes.
compile for AVX2 support - the executable will not run on a processor without AVX2
provide detection at runtime within the executable - which means multiple code paths included (and likely using inline assembly for those parts because you can’t just tell the compiler to use AVX2 globally?), which means both bloat, and a detection at runtime to determine whether or not you can use AVX2 at best which slows things down and makes the code size larger. for what benefit?
It could be maybe done with DNF magic, etc. but you’re talking small number of non-target users vs. significant re-archtecture of the release process. Additional complexity, bloat, etc… for minimal gain. All of that would cost time and effort that is better spent working on more important issues.
Better to just drop support entirely, and if people care enough offer a re-spin (just build the same distro with different compiler flags)… or direct them to a more suitable platform.
After you make this one, you gotta make a “How to Survive Your First Upgrade : A friendly guide on recompiling your kernel headers just to get stuff to work”
How about you try it in a VM like vbox, and then install it if you like it. Freenode in IRC has a fedora channel as well as efnet if you have problems. That’s how I learned, as well as having some college buddies to help collab when we had to solve something. I come from the dark ages of linuxx when you had to use ndiswrapper for windows drivers because nobody wrote linux drivers for bleeding edge hardware. You had to buy a 2 year old computerbecause nobody developed for certain chipsets yet. But everything was cheaper, faster and more secure than the alternative…
I used debian kde and it wasn’t all that bad, pretty much everything is on the internet… just got out of it cause of lazyness of searching things and switched back to mint, the friendliest distro I used so far.
I think pop will probably pop out of ubuntu soon enough and become something like solus (pun intended)… that would be something of a logical upgrade.
Me too, I liked it. Came here because minty is a *buntu.
Would it be? It’s a pretty large step going from being Ubuntu plus a theme and an addon repository to your own, full-blown distribution with your own package base to maintain. It’s a hell of a lot of work and if maintaining “only” the 32bit packages for apt would suffice, i’m pretty sure the pop devs wouldn’t be too interested in maintaining every other package under the sun themselves.
Solus is in a completely different ballpark than a Ubuntu spin.
Either you didn’t understand what I typed or I didn’t write correctly.
I said becoming an independent distro instead of a *buntu would be the logical upgrade.
If they’re willing to maintain the 32bit libs on their own that’s a lot of work, but if they managed to pull out some independent distro like solus it would be amazing.
But I don’t mind much 32 bit, steam and games are not vital to me.
Oh i totally understood what you wrote.
I just think that you are underestimating the amount of work it would be to be an “independent” distro without having a repository to fall back on. Which is what Solus is.
Pop at this point in time is 98% Ubuntu. Maybe more. And they don’t have to care at all about those 98%. No work for them. Maintaining the 32bit libraries that are needed for steam maybe puts them at 95%, or heck, 90%.
This means that going completely independent would mean 10 times the work for them. Maybe that’s possible, i just don’t see a reason why they would do that.
Going away from Ubuntu and maybe changing their base to Debian is more likely than that.
Or wait. Maybe a misread you after all? Did you mean Pop rebasing themselves off of Solus?
I’ll stay here rooting for it. Maybe they’d ship some other DE, kde would be sweet. But if they let you choose many they should do something to “hide” the tools that aren’t on the session you started. It was confusing having 3 terminal emulators when I took debian for the first spin with gnome, kde and xfce all at the same time.
Pop has, afaik, no interest in maintaining another DE. They take what ever Ubuntu or Debian use. The sole purpose of Pop to exist is to have a distro that works on the hardware they sell. It just happens to be great for everyone else too. I’m pretty sure they don’t want to take this any further atm as they are a relatively small team.
With debian, there are a few ways to install a DE. Most “short” tutorials recommend that tasks. Those are the combination of packages that get installed when you select that DE at installation. So of course they’ll come with all that DE’s “default” applications.
Debian is a lot like Arch in that sense that it doesn’t really care to make it easy for you. It does what you tell it to do and if you install 3 terminal emulators, there is no reason for Debian to hide those from you. You are perfectly able to install KDE without Konsole or xfce without xfce4-terminal. You just have to install the “correct” packages instead of the one-stop shop tasks.
Fedora was, and in my opinion still is, the only desktop variant with the goal of keeping documentation up. When I say desktop - I am not hating on Debian as they do amazing job too.
That makes it a good choice for a Beginner’s guide here, because you can reference not just the man pages, but documentation.
I would love to se a category with individual topis though.
I.e. Moving and sharing files over network. Most people I meet jsut default to SMB and then deal with the rabbit-hole there.
I remember when Wendell first introduced Linux to the forum, he was using Fedora, and that’s the distro that a lot of people here moved to and recommended. Those were the days.
Lots of us use it because it’s just an all-around, solid distro. I checked out Fedora initially on Wendell’s insistence that it was good, and after a few days of getting used to DNF being shit, I found out that he’s right.
Because DNF is functionally perfect, but slow as hell.
If you’re willing to wait, it’s a great package manager.
I’m willing to put up with the sluggishness for a distro that doesn’t randomly decide to break all the packages and stop installing anything (coughcough debian/ubuntu)
I agree and i the speed is bearable if you know what you are doing.
If you are using package manager to “browse” - you are in for a lot of clicks and waiting like its 1999.
but then again that is not the functionality of yum/dnf. If anything is providing a nice front-end it should accommodate the difference between a store and a manager
