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

[phoronix]

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

With there being some renewed development interest around FBDEV with the kernel subsystem now having a maintainer, restoring hardware-accelerated scrolling, and other patches pending to clean-up and optimize the frame-buffer device code, there is also a new performance optimization series...

https://www.phoronix.com/scan.php?pa...le-Faster-2022

[onlyLinuxLuvUBack]

All i got to say...

push( @GNU_LINUX_Heroes, "Thomas Zimmermann" );

[s_j_newbury]

I don't know how I feel about this. I really hoped the lack of maintainer and loss of functionality would finally give impetus to a userspace solution. Personally, I use kmscon with hwaccel, it's okay, fast enough, looks okay, but really isn't a complete solution and feels unfinished.

[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?

[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...

[middy]

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

[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).

[willmore]

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

[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.

[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?