FBDEV Console Patches Optimize Performance - Print Directory Listing ~25% Faster

[stormcrow]

I definitely have waited longer because I forgot to pipe away from the console or wanted to see what was going on. The directory listings are an example, but a real world case is a kernel build on a Raspberry Pi. This does have practical value.

Also, if someone's job is to maintain something, shouldn't they... optimize it within reason?

Absolutely.

It's not just the Pis. Many ARM systems use the fbdev for video output. Personally I'd direct the kernel build output to a file instead not just to speed things up, but to have a record of build info. Video output is never completely instantaneous no matter what hardware you use without a direct hardwired display. Even on PCs with much beefier video hardware it's better to direct video output to a file rather than the screen, if possible. There's definitely much that could be done to improve the situation for Linux TUI/CLI output.

[Nuc!eoN]

Would anybody be able to ELI5 this to me? What is FBDEV?

Is it what a TTY uses to display stuff?

[ElectricPrism]

This reminds me of how much I miss fbsplash ( https://wiki.archlinux.org/title/Fbs...le_backgrounds )

[mangeek]

In fact, just this morning, I was sitting there drinking coffee and bitching that I had to wait 6.1 seconds to see everything in /usr/share/doc and telling myself I didn't go to Vietnam just so I can have it in 4.7 seconds.

I definitely have waited longer because I forgot to pipe away from the console or wanted to see what was going on. The directory listings are an example, but a real world case is a kernel build on a Raspberry Pi. This does have practical value.

Also, if someone's job is to maintain something, shouldn't they... optimize it within reason?

[sophisticles]

Thank God, this was really needed since all the other Linux issues have been solved, now we finally get the much promised, highly anticipated reduction of 1.4 seconds in the time it takes to print the directory listing of /usr/share/doc.

In fact, just this morning, I was sitting there drinking coffee and bitching that I had to wait 6.1 seconds to see everything in /usr/share/doc and telling myself I didn't go to Vietnam just so I can have it in 4.7 seconds.

Glad that fancy comp sci degree paid off for Thomas Zimmermann, his parents must be so proud.

[willmore]

Oh, back in the days when the text console was actually a text console and scrolling was nearly instantanious.

[stormcrow]

Ha. To this day, devs give me weird looks when I tell them it's faster if they don't actually print the output...

Only junior devs I hope. This is not only well known, it's easily proven (in most cases).

[middy]

so this will bring back the ability to scroll using shift + PageUp / Down?

[anarki2]

Ha. To this day, devs give me weird looks when I tell them it's faster if they don't actually print the output...

[CochainComplex]

This is really nice - I was allways wondering why CPU, RAM, IOs get so much faster but often you can't "feel" this in console. Still more or less like the first time on a fast SSD. Or am I missing something?