gryffyn

I was curious to see what the actual latency was, so I used Pavel Fatin's https://pavelfatin.com/typometer/.

Notes: This system is running Arch, with kernel 7.0.3, scx_bpfland, and i3 on X11. Primary GPU is an integrated Intel GPU.

Using picom:

  Terminal                     Min    Max    Avg    SD
  alacritty                    20.1   111.4  43.8   18.0
  xfce4-terminal               19.1   112.5  49.1   17.9
  xterm                        12.3   152.6  53.1   26.4
  wezterm                      28.3   106.0  56.2   15.4
  kitty                        25.3   115.9  57.2   18.5
  ghostty                      19.2   611.8  63.4   45.5
  konsole                      37.9   138.6  68.8   18.5

Using picom --realtime:

  Terminal                     Min    Max    Avg    SD
  xterm                        5.7    63.6   15.2   8.8
  alacritty                    7.4    48.8   20.2   9.7
  xfce4-terminal               8.7    65.3   28.3   12.5
  kitty                        13.2   64.8   30.8   10.3
  konsole                      17.3   71.2   31.1   9.5
  ghostty                      12.9   63.7   33.2   11.6
  wezterm                      18.0   58.6   33.6   11.1

No picom:

  Terminal                     Min    Max    Avg    SD
  xterm                        4.5    20.4   8.6    1.0
  alacritty                    9.2    13.3   11.1   0.7
  xfce4-terminal               7.4    17.5   11.3   0.9
  ghostty                      11.2   35.3   18.8   3.8
  wezterm                      14.1   34.2   22.7   3.7
  kitty                        12.2   44.3   24.4   5.8
  konsole                      17.2   30.7   25.0   3.6

Running on the NVIDIA GPU just for fun:

  wezterm (no picom, NVIDIA)   21.7   51.2   34.8   8.1
  wezterm (picom rt, NVIDIA)   22.0   70.9   41.4   11.2
  wezterm (NVIDIA)             34.4   146.7  69.0   17.8

Slightly surprised that xterm doesn't top the chart on the picom category.

I've tried kitty, alacritty, ghostty, wezterm, and konsole (all on X11), but I keep coming back to xfce4-terminal because it opens near instantly and has the least input latency of any of the other terminal emulators I've tried. It also uses very little memory in comparison (with 6 windows open and a fair bit in each scrollback buffer):

  Private  +   Shared  =  RAM used Program
  31.2 MiB +  10.7 MiB =  42.0 MiB xfce4-terminal

Germany continues to do that, North Rhine-Westphalia just bought a >1TB dataset in 2025.

https://www1.wdr.de/nachrichten/landespolitik/offshore-steue...

I would bet it's from AI upscaling. The dark edges around high contrast borders, plus the pronounced and slightly off-colour antialised edges (especially visible on the right side of the J) remind me of upscaling models.

Qalculate <https://qalculate.github.io/> is my favourite REPL-like calculator, although it unfortunately lacks an iOS app. It feels similar to using an HP 48-series calculator.

Numbat <https://numbat.dev/> is similar, but more CLI/REPL-focused, and with more of an emphasis on being a programming language.

I use KOReader's RSS reader, I haven't run into any issues yet. It supports RSS and Atom, supports pulling the full article text if it's in the feed, supports images, and puts each feed entry into its own epub.