This is a victory.

– Every loser.

100%. There might be a slight uptick in Linux use, but the vast majority of people will either just keep using Windows 10, or buy a newer computer.

10-14-25

The 10th of Duember?

I mean the bar charts are not useful for deciding whether or not each parameter makes a difference. You need to plot e.g. the speed difference turning on exceptions makes in each case. Maybe use some colour coding and line styles to group other attributes.

Uhm… You went to all that effort and then didn’t actually answer the question! Can you put the data in a readable format so we can see whether using exceptions made a difference?

The liquor store presumably? Just submitting patches is a simple 12 step process. I can’t imagine the development and review process is any easier.

I have implemented a Git client from scratch which involved quite a bit of reading the Git source code. It’s not bad code but it’s definitely the sort of code that would break in all sorts of unexpected ways if you changed something. I wouldn’t volunteer my time their tbh.

Yep, whenever they fix a bug it’s added in a new flag that nobody knows about.

git --enable-sane-behaviour

Git has four build systems?? Meson seems overkill if you already have CMake too. The only thing it really adds is that it’s nicer to write (CMake is somewhere between Bash and PHP in sanity), but if you have to write CMake anyway…

Fair point - 4.1% for desktop, which is more than I would have guessed.

It’s 1.5% according to StatCounter which is the least biased source I know of.

Yeah it’s easier to compile software with support for the latest vector extensions etc. if you do it from source. It is also possible to do runtime detection and switch between implementations that way, but it requires more work.

Tbh I don’t think it would make much difference in practice. If you are ok with supporting only recentish CPUs you can use one of these options:

• -march=x86-64: CMOV, CMPXCHG8B, FPU, FXSR, MMX, FXSR, SCE, SSE, SSE2

• -march=x86-64-v2: (close to Nehalem) CMPXCHG16B, LAHF-SAHF, POPCNT, SSE3, SSE4.1, SSE4.2, SSSE3

• -march=x86-64-v3: (close to Haswell) AVX, AVX2, BMI1, BMI2, F16C, FMA, LZCNT, MOVBE, XSAVE

• -march=x86-64-v4: AVX512F, AVX512BW, AVX512CD, AVX512DQ, AVX512VL

v2 is definitely fine and v3 is probably acceptable by now.

In short I don’t think -march is a compelling argument for avoiding binary distribution. If it really makes a big difference either distribute multiple versions of your software using the flags above, or do runtime detection.

Good luck with this approach on a server.

Indeed, obviously I’m talking about desktops here.

If by ‘suspend’ you mean that the process will just halt, then: Which processes? All of them? Good luck displaying a message then.

You could use some kind of heuristic to suspend ones using the most memory/CPU. Or just suspend them all. Obviously you would exclude the processes needed to display the message.

If by ‘suspend’ you mean moving the memory to disk

No I meant just pausing their execution. I’m pretty sure ctrl-alt-del does something like this on Windows.

Ah maybe. I’m still on RHEL8. Even so, “it hangs a bit and kills a random process” is still shit! What it should do is suspend processes, and show you a GUI saying “you’re running low on memory, here are your running programs and how much they are using” and allow you to choose which processes to kill, or whatever.

That would be far too user friendly for Linux though. I don’t think the kernel/Wayland Devs could really comprehend that tbh. They’ll say something along the lines of “users shouldn’t be doing that”.

Don’t hard-reboot when memory runs out.

You basically shouldn’t until you are forced to move. Almost all of the improvements so far are in the internal architecture.

You might notice some tiny differences if you switch, like logging in doesn’t show a black screen at any point, and window choosers when screen sharing show a (totally broken) grid of previews instead of a plain list of window titles.

Hopefully when X is fully dead (give it another 10 years) we’ll see some actual improvements, e.g. RDP-style remote desktop, good support for multi-monitor, HDR, HiDPI, etc.

Huge speed-ups for the will-it-scale tlb_flush2_threads test - presumably not very representative!

I wonder how much effect it has on real world workloads.

Definitely true for a minority. Not the way most of these articles are presented though (including this one).

I don’t see why. You can be interested in Linux and like some aspects of it but still get annoyed at the blinkered zealots claiming that there’s no reason to use Windows.

Pornhub’s audience is 75% male, and Linux users are almost all male too (and … no offence guys but probably more likely to rely on porn), so I wouldn’t take that as a representative global figure.

Statcounter gives 1.4% which is much more likely to be unbiased (and sounds way more plausible based on my experience of real life).

You won’t get very far in life with such primitive black and white thinking.