Alacritty worse than Konsole?
Posted by TheRealRubiksMaster@reddit | linux | View on Reddit | 79 comments
When I do some stress testing on both, alacritty differs from konsole by extremely marginally, like 5%. The difference is Alacritty eats like 2x the amount of cpu usage as konsole and a minor amount of gpu (like 2%, but konsole is technically 0%).
I tried a whole bunch of stress tests like yes'ing chinese liguatures, catting massive binary files, and other stress tests like that.
How exactly is Alacritty better than konsole? I hear so many people rave about how performant it is, but from my tests it seems to be very much not so. Is there some form of settings i have to switch or something?
audioen@reddit
If there is one thing I've not been worried about for like decades, it is the performance of my terminal emulator. The argument, in my opinion, is absolutely meaningless and bears no resemblance to any reality I'm familiar with. Even basic gnome-terminal which was famed for being "slow" is plenty fast as far as I'm concerned. Sure, I noticed step down from xterm in like 2005 or whatever, but it wasn't big enough for me to care about it even then.
Boring_Locksmith6551@reddit
I like Ghostty
sublime_369@reddit
I haven't used alacritty but IIRC one of the biggest selling points is that the text rendering is GPU accelerated which is not a feature of many console emulators.
Konsole doesn't make a big play of it, but it's actually also GPU accelerated. I ran the ghostty ASCII bouncing ghost on it and it was smooth.
Unless there's some specific feature you need, I would just stick with Konsole personally.
razorree@reddit
so it can display git log using my RTX card in a split second! but Konsole (or other term emulators using only CPU) can do that way slower, like in 10ms ...
comradeacc@reddit
I mean its better sure, not for every task.
I really like snappy terminal at work (I use mostly terminal editors) and it really makes a difference, but for most cases I just use the default on the machine that I have available.
sublime_369@reddit
You didn't read my comment.
razorree@reddit
yep. i understand what your saying :) my comment was more rhetoric question to all that GPU accelerated terminal emulators hype... :)
scottchiefbaker@reddit
I had no idea "ghostty ASCII bouncing ghost" was a thing? Is this something people use to "benchmark" terminals?
sublime_369@reddit
No idea mate but I thought it looked cool AF so I just wanted to try it.
https://ghostty.org/
scottchiefbaker@reddit
Neat... takes about 8Mb/s to stream ghosttime over SSH on my network :)
sublime_369@reddit
Cool.. and she runs smooth?
scottchiefbaker@reddit
Ya I'm "streaming" it to my Windows Terminal and it runs perfectly.
TheRealRubiksMaster@reddit (OP)
yah thats what im planning because its way more customizable. i was just trying to figure out if it was worth the tradeoff. this post was a question, but most people just downvoted me because they dont want to admit their glazing is misplaced.
sublime_369@reddit
Yeah I just ignore downvotes. I've noticed I make a very basic but relevant factual statement and it get's downvoted. Some people can't handle the truth.
Unaidedbutton86@reddit
Just choose whatever you prefer, a small performance difference doesn't matter in a terminal emulator
sebby2@reddit
Casey Muratori has entered the chat
TheRealRubiksMaster@reddit (OP)
Thats basically why im seeing. Ive heard so many people glaze alacritty and push down konsole. Just for it to not be true. Just look at the downvotes, people are pissed that im askign a question that insennuates their baby isnt what they glaze it as.
King_fisher1452@reddit
You’re basically comparing apple to oranges. I like konsole, it’s nice, convenient, and offer customization. They both have their functionalities and purposes. It’s derivative to just call something bad when all you have done is some arbitrary tests when you could just pipe them instead of standard output.
JockstrapCummies@reddit
All these minimal terminal emulators are really good at beating the boring DE-default emulators in things like microbenchmarks and perceived benefit in lending an air of elitism to the user.
Venar24@reddit
Alacritty good for customizing but sligthly worst for performance (irrelevent if you have a half decent rig)
Konsole better for performance (irrelevent if you like customizing your os to make it truly yours)
rdqsr@reddit
Even then the only time you'll see performance issues is with printing large amounts of text to the console very fast, and at that point in 99% of cases I'd argue you'd be better off piping it to a log file or using the pv command as needed.
Revolutionary_Click2@reddit
You say in a comment that you’re running a system with an RTX 5070 GPU. It’s a safe assumption that you have a powerful and modern system to go along with it, which is likely a gaming desktop. Why in the world would you be sweating 2x CPU usage for a terminal emulator, something which your system can run without breaking a sweat all day, every day? Who honestly gives a shit if one is slightly heavier on resources than the other? How does that impact you (or anyone) in the real world?
This reminds me of folks arguing over inconsequential differences in the RAM usage of KDE and GNOME. The memory is there to be used! The system or its applications will often claim memory they do not need for additional caching and whatnot, then give it back if other processes request it. If you have aren’t experiencing meaningfully constrained memory pressure, it is completely fine and expected that processes will use a bit of free RAM sometimes. It demonstrates nothing about the quality of an application if it is slightly more memory-hungry than some other application.
People use Alacritty because they like it. You like Konsole jnstead? Cool, use that! The extra CPU usage has no real impact on anything and you should not waste one more second thinking about such things.
anjumkaiser@reddit
I’ve been using yakuake for last 20 years. Even Konsole feels odd to me, and the hype about these new terminals, I just don’t get it. One you get used to quake styled terminal, nothing else matters.
LuckySage7@reddit
Stupid reason but I used it because I use zellij which has a keybind of alt+f to toggle floating multiplex. On konsole, alt+f conflicts with a kwin keybind for opening th windows context menu. Minor annoyance but I like how minimal alacritty is and that all config is done through a toml config instead of builtin settings menus.
icehuck@reddit
If you like konsole use it. There is no right or wrong answer here.
EtiamTinciduntNullam@reddit
Alacritty always feels the most snappy for neovim. I just checked now and Konsole will stutter badly on something so simple as holding
j(move one line down continuously), while Alacritty will do it smoothly. Seems like Konsole is the worst in performance I've seen yet.I don't need a toolbar in terminal like I get in Konsole, it's still thick even if you remove everything from it.
I guess if you don't use any interactive terminal application then any terminal program will work and other things than performance matters for you.
Schreq@reddit
Relying on your keyboard repeat rate is such an anti-pattern. There are better ways to scroll in vim, which also make the terminal rendering speed irrelevant.
EtiamTinciduntNullam@reddit
Holding
jis not how I usually scroll, but it's not relevant as this is not a discussion about keystroke efficiency in (n)vim, this is a simple "benchmark" that Konsole fails miserably at. Try something more difficult and difference is even bigger.Tight action-feedback loop makes it more fun to use PC (including terminal). What if I choose to hold
j? Why "modern PC" has to choke when scrolling text window one line at a time in Konsole when an "e-waste" level machine handles it well in Alacritty?Schreq@reddit
That is just a realistic "benchmark" when you use vim and other software in a manner it was not designed for, tho.
Byron_th@reddit
I think this section from alacritty's readme is relevant:
Benchmarking terminal emulators is complicated. Alacritty uses vtebench to quantify terminal emulator throughput and manages to consistently score better than the competition using it. If you have found an example where this is not the case, please report a bug.
Alacritty does not optimize for low CPU usage.
scottchiefbaker@reddit
I've been using Konsole for years. It's probably the best terminal emulator I've ever used. Does "performance" really matter in a terminal emulator? Short of "cat 100mb.txt" when does performance really come into play. If it's reasonably fast/responsive and has the features you want that seems good enough for me.
5% performance seems very minor. Also, 5% performance in what? It outputs and scrolls 5% faster? I struggle to find a real-world use case where that would be relevant other than just comparing numbers on a spreadsheet.
hxka@reddit
Tried that. Depending on the size of the window, I got:
That's not within 5%
TheRealRubiksMaster@reddit (OP)
"it works on my machine"
on mine i did gigabyte files and got liek 95 MB/s on alacritty and 88 MB/s on konsole, which is basically a rounding error.
peterhoeg@reddit
I think the question you should consider asking yourself is "what would I stand up gain from using A over B" (substitute whatever values you want for A and B). Unless there is something tangible you need, pick one and get on with life.
TheRealRubiksMaster@reddit (OP)
The title was a bait, i was hoping the glazer would come in and defend it, but they are just mad but dont explain anything.
kalzEOS@reddit
I've never used any terminal emulator other than the one that comes with the distro. I genuinely have no clue why others choose different ones. Like is there an advantage or something? Or is it preference?
TheRealRubiksMaster@reddit (OP)
thats what im trying to figure out too
Hug_The_NSA@reddit
The difference is Alacritty eats like 2x the amount of cpu usage as konsole and a minor amount of gpu (like 2%, but konsole is technically 0%).
I'd be curious to see the FPS of the terminal while this is happening. Just because alacritty is using more resources doesn't necessarily mean it's doing anything wrong. It may be displaying it at a higher or more consistant FPS. I'd argue that's entirely worth the extra resources.
I also really love konsole, but for me alacritty just seems much smoother overall.
ronchaine@reddit
Terminal emulator having a consistent FPS is doing things wrong. A terminal emulator should not refresh what it's drawing unless it needs to.
TheRealRubiksMaster@reddit (OP)
thats what i was thinking too
TheRealRubiksMaster@reddit (OP)
Is there a way to check that?
ipsirc@reddit
Run bbdemo, it's a good performance test for terminals.
https://aa-project.sourceforge.net/bb/
Hug_The_NSA@reddit
Not that I know of right off hand. If any terminal benchmarkers are ITT now is your time to shine.
admica@reddit
The alacritty hype train is straight out of the twitch station.
razorree@reddit
have you seen Ghostty hype ? lol ....
LvS@reddit
Have you seen this link?
robprobasco@reddit
It’s a terminal. The only reason to switch from default is like…style preferences. I like mine to be able to highlight file and command flags and stuff. I pick any of the ones that do that and call it a day.
Sosowski@reddit
This is a terminal emulator, not Unreal Engine 5. Any “optimisation” and performance benefits are gonna be in nanoseconds. Not noticeable.
A console is not being redrawn if it doesn’t update. There’s no reason to seek out the most performant one unless you’re doing something really weird.
jorgejhms@reddit
Nothing weird, but anyone that have coded on nvim knows there is a difference between a low performance terminal and a optimized one.
Sosowski@reddit
Is there really? Up to the point where GPU acceleration makes a difference too?
jorgejhms@reddit
Yes. Non GPU terminal feel laggy when working on a big codebase. It's not a terminal thing. Zed code editor is also GPU enabled for text rendering.
TheRealRubiksMaster@reddit (OP)
exactly, ikr
tripanossoma_cruzes@reddit
Being an ex urxvt user, I found my home with foot.
Never understood the need for GPU on terminals.
ipsirc@reddit
https://aa-project.sourceforge.net/bb/
OfflerCrocGod@reddit
I don't care about performance in a terminal, they are all fast enough. Very few are as configurable as kitty or wezterm so I stick with those two. I work all day writing code in my terminal so it needs to programmable or highly customizable.
stiggg@reddit
Konsole is great software, always was. If you’re on KDE it totally make sense to use it.
__konrad@reddit
It's funny that a very simple xterm is 10x slower than Konsole
FryBoyter@reddit
Let's assume that your ‘benchmark’ is correct. Does it really matter that much?
I mainly use my tools for other reasons. For example, because of their range of functions or simply because I prefer them to other solutions.
I therefore think it's wrong to claim that tool A is better than tool B because it requires 2 per cent less of a certain resource.
By the way, I don't use either Konsole or Alacritty.
Gent_Kyoki@reddit
Honestly most of the terminals ive used on linux has performed about the same just use the one you like the most lol.
speedyundeadhittite@reddit
Both running a terminal-heavy app right now, on my 6y old Dell I can't even notice them on the top list, they're buried down with all other processes dominating.
eattherichnow@reddit
What people glaze over is latency while using software such as nvim - often even at expense of raw performance while dumping a huge file. That requires some fancy trickery to measure properly (time from button press to the glyph showing up, in various scenarios e.g multibyte or layout changes).
But ultimately konsole is good enough for me.
landsoflore2@reddit
In my experience, Konsole is just fine - extremely lightweight, highly customizable, can be made to look really pretty, FOSS... What is there not to like if you are already using KDE?
QuickSilver010@reddit
Idk. I just use kitty. It's the most innovative terminal
TheRealRubiksMaster@reddit (OP)
I didnt bother testing that because it seemed liek it had the least real work customizability.
QuickSilver010@reddit
It's very customizable. The config file is also very nice. Plus I like the image display ability of kitty. I intend to develop some apps with that feature
TSG-AYAN@reddit
I have used konsole (and yakuake) for the last year and it has not disappointed me at all. It supports all image formats (kitty, iterm, and sixel), OSC52, and just WORKS for all my needs.
_chococat_@reddit
I don't use alacritty because it is performant, I use it because I like its font rendering better. Latency may be slightly important, but how fast I can dump a ton of text to the terminal is not interesting or useful.
dcherryholmes@reddit
IDK. I checked out alacirtty and kitty and a few other terminals, thinking GPU acceleration and diplaying jpegs in the 'term would be nice. I know about mux'ing from back in the screen days, so that's not alien. But IME I just spent time trying to make it look and work like Konsole. Now I don't even bother and just use screen when I need it, which is almost never. Which isn't to diss on anyone's use-cases. Just MMMY.
draezha@reddit
Tbh I think people just like to knock on the KDE app ecosystem. You're pretty much on point, any difference is negligible, unless maybe you have a very very slow computer... Then there might be something that makes a difference? I like Kitty personally, but that's just cause of how themeable it is. None of the ones I've used have made any noticable differences performance-wise.
TheRealRubiksMaster@reddit (OP)
Yah that why im leaning konsole, imo its alot more customizable friendly than the config only based stuff.
Hug_The_NSA@reddit
So using "hyperfine" which is a terminal benchmark available in many repos I tried some commands myself. I did "ls" and "pwd" just for easy commands. Both times konsole actually won vs alacritty.
ipsirc@reddit
I don't think hyperfine is the right tool to measure this.
Hug_The_NSA@reddit
I'm not sure either but it seemed like something interesting to mess around with for a bit.
SteveHamlin1@reddit
Did you control for other variables? Do tests of each after a fresh reboot for each? Do them under the same DE? Make sure the same services/processes were running / not running for each? Run the exact same tests, off of the same filesystem, for each?
Unique-Dragonfruit-6@reddit
Run a program with a lot of console output and time it. On my system it takes measureably longer on other consoles and alacritty was the fastest. (Like, with "time find" or something.) That said, I ended up switching back to xfce-terminal because I liked the copy/paste support better.
I don't know about konsole specifically but most of the Linux console programs use the same underlying library to emulate the try and draw it to the screen, and it's written for maintainability rather than raw performance. The amount of abstraction bloat each character goes through before it gets to your screen is atrocious, but it gets the job done.
Alacritty decided they wanted to make the trade-off the other way and it's rather impressive how optimized they got it.
TheRealRubiksMaster@reddit (OP)
please read the middle section
Unique-Dragonfruit-6@reddit
You seem to be talking about resource usage. Alacritty's goal was to minimize latency. If they're running at twice the frame rate they would intentionally use more resources to get better responsiveness.
Otherwise you might have an issue with your graphics stack. But my experience has been it's faster and watses less CPU in general.
usernamedottxt@reddit
Sanity check, you have a GPU yeah?
What are you testing? How are you measuring it?
TheRealRubiksMaster@reddit (OP)
5070 yes, also did you skip the entire middle section where i said exactly that?