Switching to iTerm via Cmd-Tab is slooow
Thanks for filing an issue! Please answer the questions below so I can help you.
- iTerm2 version: 3.3.12
- OS version: 10.14.6 (18G6020)
- My com.googlecode.iterm2.plist
- Debug log made by cmd-tabbing away and back to iTerm2
Attach a screen capture video if it would make the reproduction steps clearer.- Are you reporting a performance issue, excessive CPU usage, or a hang? Yes; sample: Sample_of_iTerm2.txt (made by Cmd-Tabbing into iTerm2; wait; then back to Activity Monitor; then back to iTerm2 (during which it ended the sampling)
- Are you reporting a crash? No
- Are you reporting excessive memory usage? No (real mem is 1.91 GB, virtual is 11.90 GB -- but I have a bunch of windows and tabs, so I figure that's all normal; I tried replicating a similar amount of Windows and Tabs in Terminal.app, and its memory usage was comparable)
Detailed steps to reproduce the problem:
- Have iTerm2 running in the background
- Cmd-Tab into iTerm2 and instantly start typing a few letters
- Wait for iTerm2 to come into foreground and start showing what you typed
What happened:
After about 1 full second, iTerm comes to the foreground; after roughly another second, the letters I typed start appear
What should have happened:
All of this should have been basically instantly. I tried to reproduce this in Apple's Terminal.app, by opening a similar number of windows each with a similar number of tabs to what I have in iTerm2 (this is still flawed, because I also have a bunch of tabs in iterm2 split vertically once; so overall this is still biased towards Terminal.app, but hopefully not overly much).
Remarks
I have turned off things like blurring, transparency and antialiasing. GPU rendering is enabled. This is on a 2018 15" MBP, 2,6 GHz Intel Core i7 Hexacore, with Radeon Pro Vega 16 4 GB; one external screen (4k) is attached and the main display, but the internal screen is also on. I'll be happy to provide more information, including a full system report (however, AFAIK that contains tons of serial numbers etc. so I'd rather not publish that on a public issue tracker).
Any hints as to what might be wrong would be appreciated!