You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
With that I see things like 228 CPU usage from cosmic-term in htop.
Why not ;)
Anyway, it's better, IMHO, to test with something like rand-ascii (which I just quickly wrote), than to dump whatever unicode/non-unicode mess from /dev/urandom on the terminal, especially when comparing with other terminals that don't utilize a sophisticated text stack.
I don't have the time to test new builds, but I tested an old build (cb26aed), and the performance seems to have already been fine!
And I'm leaving this comment here because I've done a few performance optimizations on Monospace fallback code in cosmic-text.
I've tested them with the same old build, but with a lot of local BiDi-related changes on top that have the potential to slow things down (including paragraph support, physical-to-logical column position maps, among other things), and the performance seems to be still fine. So, that's good news.
I used rand-ascii to compare Konsole with COSMIC Term just cuz I was curious. KDE System Monitor showed ~15% CPU usage for Konsole and ~25% CPU usage for cosmic-term.
cat /dev/urandom
is a simple test for this.With that I see things like 228 CPU usage from cosmic-term in htop. So somehow this results in a substantial CPU load on multiple cores.
The text was updated successfully, but these errors were encountered: