Computer Specs (Any Lag?)

I would like to know the hardware specifications for the computers of anyone on this forum who uses Krita for digital art. I’m guessing I might need a computer upgrade since I’m currently using a six year old Toshiba Satellite C55-B5200 to run Krita and I’m experiencing occasional lag when using certain watercolor presets from the pack by @Pesi at the default size. I am not sure how many users have older computers though, but I just want to make sure I’m not the only user experiencing brush lag for more advanced brush presets.

If anybody has the hardware specifications for his computer model, maybe he can post them here and explain whether he notices any brush lag under any circumstances when using Krita.

This is my Machine (about 5 years old):

Operating System: Manjaro Linux
KDE Plasma Version: 5.18.4
KDE Frameworks Version: 5.69.0
Qt Version: 5.14.2
Kernel Version: 5.4.36-1-MANJARO
OS Type: 64-bit
Processors: 8 × Intel® Core™ i7-4790 CPU @ 3.60GHz
Memory: 31,3 GiB of RAM
Graphics: GeForce GTX 970 with 4096 MB RAM

It doesn’t lag except when I use really big brushes on insane canvas sizes of 8000x8000px plus (depending on the dab spacing of the brush too).

From what I know, Krita does not utilize the graphics card very much so the best thing you can do to boost performance probably is getting as much processor power as possible. I Would probably get a thread ripper today or at least 16 cores.

CPU performance is not the entire story but it’s a big help.

Your laptop specs say it has an Intel Core i3-4005U CPU so a benchmark score and comparison can be seen here:

For @Takiro 's CPU, the benchmark score and comparison is here:

CPU benchmark performance figures are a useful guide if you’re comparing computers based on the CPU. Having fast RAM is also a good thing.

My computer is not bad (2 x Xeon E5450) but it’s let down by having DDR2 RAM (hangs my head in shame).

OMG, did you look outside, its 2020 already :laughing:

I agree fast RAM is important too, but from my experience this mostly affects loading and saving time, not so much lag of brushes. But it is of course true that it’s a problem when you can’t feed the CPU fast enough with data to do the operations on.

Try to deactivate the OpenGL graphic acceleration in your configuration… You’ll see that Krita isn’t as fluid as before :wink: So i think that it use graphic card.

It may not use directly GPU for calculation (but maybe it can, as Krita use Vc library that’s seems to be able to use GPU - if a developer read this, it can confirm or not this point), but it any case if OpenGL is activated, OpenGL acceleration is provided by graphic card :wink:

On my side, using Krita on Linux Debian 10 running on the following computer:

AMD FX™-8350 Eight-Core Processor @ 4Ghz
GeForce GT 630 Rev. 2

Krita is running without any problem, it’s fluid :slight_smile:

(also have a GeForce GTX 1060 3GB on the computer, but the I dedicated the GeForce GT 630 Rev. 2 for Krita use)


It’s like having a big old dog that you take for a walk every day and you have to wait for it to catch up with you. It’s loyal and reliable and I love it :slight_smile:


Krita uses the GPU to draw the canvas, that is, the rendered image is put into textures, and that is zoomed, rotated and panned using the GPU. The canvas decorations – brush outline, guides, assistants etc. are also painting using the GPU, but through QPainter.

Vc is a library that uses vectorization (sse, avx, avx2) to speed up tasks that can be done in parallel.

Krita uses multithreading, too, once Krita gets input, the input is created into a task that is executed in a background thread. Saving is done by making a shallow copy of the document and saving that copy. When applying a filter or rendering the projection, the image is divided into tiles, and tiles are fed to worker threads, so much of that happens simulteanously as well.

If you download really old versions of Krita ( – the oldest binary we have is you’ll see that Krita has become much more responsive over the years. We really do work on improving Krita’s performance all the time, but…

As for hardware, when I started working on Krita in 2003, this was my hardware: – the dell, not the powerbook. Yes, really: 40GB of hard disk, 512MB of memory and 32 MB of GPU memory. It was a monster! Badly built, though, at my ifrst Akademy conference presentation, when I took my laptop and walked to the dais where I was going to give a presentation on krita, the dvd drive sailed right out of the laptop and crashed to the ground. Dell had simply forgotten the screw…

1 Like

Yes, it is used for rendering but not for processing the image. I once wondered if we could have Krita use the Graphics card for image processing since manipulating raster and verctor data is pretty much what they are build for, and it seems there was some discussion about it but copying the data to the cards RAM and back all the time aparently defeats all the benefits of the faster calculations.

Yes, it was why I’ve written


But @boud has provided a detailed and interesting explanation about what is used where and why…

On my side I use a GeForce GT 630 for Krita as I’m using Krita on a FullHD display and this graphic card is enough to render Krita without any lag at 60Hz

But when I use Krita on my 4K monitor, I have to use the GeForce GT 1060 because the GT 630 is not able to reach 60Hz at a such resolution and it generate lags (not even for Krita but for all application: looking a vidéo, scrolling a web page…)

So @Rigognos the hardware specification will also depend of your output: if you have a 4K monitor (or if you think you’ll buy one) you must have a video card that is able to output display at least 60Hz refresh at 4K resolution

If you’re using a FullHD display, I think most of current graphics are able to display output properly and provide OpenGL acceleration, and then you don’t need to have an expensive gamer graphic card :wink:


There are some developments in that area, though, where the GPU can access data in main memory without having to upload and download it. But that’s not universally available, and in any case, we’re drowning in work anyway :slight_smile:

Yes just recently I had a small discussion about what a mess GPU compute still is, which eventually lead to basically throwing OpenCL 2.0 (which basically existed more on paper than for real) into the trash bin and make OpenCL 3.0 the new OpenCL 1.2 + optional extensions. Bye bye shared virtual memory. Yay. Still no sane way to share memory with the GPU.

No offense, but faster memory would probably not make this system look signifficantly less outdated.

8 cores at 3GHz still sound decent on paper, but if you factor in over a decade of IPC improvements, SMT, AVX2 and higher frequencies, I wouldn’t be surprised if even the new budget Ryzen 3 3100 scores signifficantly higher in any benchmark…

Oh and for completeness sake, I got a Core i5-4670, it’s still okay for my painting needs, but compiling feels atrociously slow these days…I hope 8 cores with SMT gets more affordable soon (although it’s at least worth considering at current prices).

Oh definitely. A CPU benchmark score is only useful if you’re comparing computers of the same ‘era’ and the same ‘class’. Even then you have to read the entire device specs carefully and ideally find a review site that carried out a set of full machine benchmarks.
Still, my old dog does what I need it do to, every day without problems :slight_smile: