Ubuntu 25.04 "Plucky Puffin" Development Opens - Defaulting To -O3 Optimizations

[_blk]

Name those cases please.

Doesn't matter if he can name them. They're around indeed. But they won't matter.
But you definitelt don't need to "waste" a whole release with -O3 to "try it out."

Anyway, if it's as stable I guess it'd be ok (which I have my doubts, but then again I'm from too long ago to honestly have retried anything with O3). Size just doesn't matter that much anymore nowadays for most cases. Removing -doc pkgs will give back more when you really need it.

I'm not leading that pack but I'd rather like to see some other metrics like boot time (anyone remember the first few upstart releases after sysv, around 6.10 when that brought noticable improvements?) rather than bloating the system with snaps and trying to shave 10ms off starting firefox with -O3?

Enough ranting, I still like Ubuntu best, but I'm definitely having eye roll moments. I'm staying on LTS anyways these days, so hopefully the wind will blow over this one before 26.04.

[sophisticles]

I am not sure I have ever seen a plucky Puffin, I did try to pluck one years ago but it got away.

Maybe i will try again someday.

[_blk]

How is Ubuntu getting "slower and slower" with every consecutive release?

I hate to be this kind of person, but, do you have any numbers to back up your claim? I know it's typical here to hate Canonical/Ubuntu but spreading misinformation?

Here's the most (Not) scientifically accurate boot time results obtained from an Ubuntu 24.10 installation running on a (Laptop) Core i5 13500H, 32GB of 4800 MT/s RAM kit and a Samsung 970 EVO 512GB.

Startup finished in 15.248s (firmware) + 1.047s (loader) + 1.963s (kernel) + 6.597s (userspace) = 24.857s
graphical.target reached after 6.568s in userspace.

A sub 25 seconds boot time, is that slow by today standards (On a Laptop)? That's the same amount of time it takes Windows to boot on an x870E Motherboad.



You do realize a modern CPU will not boot on such an ancient Kernel? Ubuntu 10.04 was released on April 29th, 2010, Michael would need to find (Assuming he doesn't have) a RAM Kit, Motherboard and CPU from that era, totally defeating the purpose of the test because a Core i9 14900K, Ryzen 9 7950X completely smokes a Core i3, i5 from 2010.

Isn't your hardware getting older to today's software demands (amdgpu, Gnome, Kernel, systemd, etc)?

• The Linux 4.9 Kernel Is Around 22.3 Million Lines
• The Linux Kernel Enters 2020 At 27.8 Million Lines In Git But With Less Developers For 2019

If you really wanted to, you could probably boot a vm with quite old distros and compare. But I don't expect much insight from such a comparison.

[F.Ultra]

and just all this shit like openssh-server is installed on Ubuntu by default. You should know that.

It is not, which is why you have to manually install it after if you want to connect to it via ssh. Look I've installed thousands of Ubuntu desktops: openssh-server is not included by default. It's not even included by default on the server edition of Ubuntu, you have to manually select it during install.

I have no idea why you are lying, smells more and more like a troll here.

edit: just for kicks I made a fresh 24.04 install in VirtualBox, and look, not a single port open for external connections and no presence of the ssh daemon:

Skärmbild från 2024-10-31 04-13-15.png

[stan]

How is Ubuntu getting "slower and slower" with every consecutive release?

I hate to be this kind of person, but, do you have any numbers to back up your claim? I know it's typical here to hate Canonical/Ubuntu but spreading misinformation?

Here's the most (Not) scientifically accurate boot time results obtained from an Ubuntu 24.10 installation running on a (Laptop) Core i5 13500H, 32GB of 4800 MT/s RAM kit and a Samsung 970 EVO 512GB.

Startup finished in 15.248s (firmware) + 1.047s (loader) + 1.963s (kernel) + 6.597s (userspace) = 24.857s
graphical.target reached after 6.568s in userspace.

A sub 25 seconds boot time, is that slow by today standards (On a Laptop)? That's the same amount of time it takes Windows to boot on an x870E Motherboad.



You do realize a modern CPU will not boot on such an ancient Kernel? Ubuntu 10.04 was released on April 29th, 2010, Michael would need to find (Assuming he doesn't have) a RAM Kit, Motherboard and CPU from that era, totally defeating the purpose of the test because a Core i9 14900K, Ryzen 9 7950X completely smokes a Core i3, i5 from 2010.

Isn't your hardware getting older to today's software demands (amdgpu, Gnome, Kernel, systemd, etc)?

• The Linux 4.9 Kernel Is Around 22.3 Million Lines
• The Linux Kernel Enters 2020 At 27.8 Million Lines In Git But With Less Developers For 2019

Do you seriously think a 25-second boot on modern hardware that has screaming-fast CPU/RAM/Graphics is something to be proud of? It is in fact an awful 250% regression (at least) compared to Ubuntu from 2009, which was able to boot in 10 seconds on tortoise-slow hardware of that era (see Phoronix articles below):

https://www.phoronix.com/news/NzMxMw

https://www.phoronix.com/review/ubuntu_lucid_final

Back in 2012 they were even striving for a 2-second boot:

https://www.phoronix.com/news/MTEwMjc

Running a boot time comparison of Ubuntu versions from 10.4 till now will show just how wasteful of CPU cycles and IO accesses Ubuntu has become.

I don’t actually care that much about boot time itself, but I find it a more easily measured reflection of latency, which is notoriously difficult to benchmark. From personal experience, I can attest that piss-poor boot time in recent Ubuntu versions go hand-in-hand with piss-poor latency and sluggishness.

[ffs_]

Here's the most (Not) scientifically accurate boot time results obtained from an Ubuntu 24.10 installation running on a (Laptop) Core i5 13500H, 32GB of 4800 MT/s RAM kit and a Samsung 970 EVO 512GB.

Startup finished in 15.248s (firmware) + 1.047s (loader) + 1.963s (kernel) + 6.597s (userspace) = 24.857s
graphical.target reached after 6.568s in userspace.

22.04 -> 22.10 -> 23.04 -> 23.10 -> 24.04: for me boot time increased by ~1 second total.

Startup finished in 13.914s (firmware) + 1.024s (loader) + 2.137s (kernel) + 614ms (userspace) = 17.691s
graphical.target reached after 605ms in userspace.​

Haven't tested 24.10 yet.

[MillionToOne]

simplicity and convenience for intruders?

This is a desktop! why use ssh by default? this is not a server. if it's a desktop, then you always have physical access to the computer because it's on your desk. but the hole is huge. It's just a stab in the back for a beginner.

Ever heard of anything called remote control?

[rogerx]

Ran Gentoo for a decade or two decades, and very rarely were there any software packages not compiling or functioning incorrectly with O3 optimizations. More so, more recent software code would fail with O3, or mostly due to poorly written code or poorly maintained code.

It is very likely Apple/Mac and MS Windows compile with optimizations enabled from what I've seen, and for a very long time now.

Think O3 optimizations is a good thing, as it will further force software programmers to write good code and/or fix broken code.

I'm now running my computers on a popular de facto binary non-systemd Linux distribution, and at my age, I could really care less about things running quickly... albeit it is nice. Slow-downs nowadays are mostly due to Internet browser code bloat.

The difference between O2 and O3, similar to driving a vehicle with either a 4-cylinder engine or 8-cylinder engine. Code executes noticeably much quicker, however code will still have similar speeds when reading databases or functions using the bus, due to through-put bottle-necks. (eg. audio/video processing, databases, ...)

[Mark Rose]

Ran Gentoo for a decade or two decades, and very rarely were there any software packages not compiling or functioning incorrectly with O3 optimizations. More so, more recent software code would fail with O3, or mostly due to poorly written code or poorly maintained code.

It is very likely Apple/Mac and MS Windows compile with optimizations enabled from what I've seen, and for a very long time now.

Think O3 optimizations is a good thing, as it will further force software programmers to write good code and/or fix broken code.

I'm now running my computers on a popular de facto binary non-systemd Linux distribution, and at my age, I could really care less about things running quickly... albeit it is nice. Slow-downs nowadays are mostly due to Internet browser code bloat.

The difference between O2 and O3, similar to driving a vehicle with either a 4-cylinder engine or 8-cylinder engine. Code executes noticeably much quicker, however code will still have similar speeds when reading databases or functions using the bus, due to through-put bottle-necks. (eg. audio/video processing, databases, ...)

Long ago I believe -O3 enabled --no-strict-aliasing which caused bugs when used to compile code that had aliased variables. The code was not wrong.

[patrick1946]

Ran Gentoo for a decade or two decades, and very rarely were there any software packages not compiling or functioning incorrectly with O3 optimizations. More so, more recent software code would fail with O3, or mostly due to poorly written code or poorly maintained code.

It is very likely Apple/Mac and MS Windows compile with optimizations enabled from what I've seen, and for a very long time now.

Think O3 optimizations is a good thing, as it will further force software programmers to write good code and/or fix broken code.

I'm now running my computers on a popular de facto binary non-systemd Linux distribution, and at my age, I could really care less about things running quickly... albeit it is nice. Slow-downs nowadays are mostly due to Internet browser code bloat.

The difference between O2 and O3, similar to driving a vehicle with either a 4-cylinder engine or 8-cylinder engine. Code executes noticeably much quicker, however code will still have similar speeds when reading databases or functions using the bus, due to through-put bottle-necks. (eg. audio/video processing, databases, ...)

Wow, do you really think the compiler has no bugs? O3 is producing more machine code, which has the disadvantage of filling the caches. So, in the end, O3 can be slower than Os. Modern hardware does not profit so much from unrolling because the branch prediction has improved.