Open MatthewZMD opened 3 years ago
Same phenomenon and I end up restarting Emacs much often.
ping @TommyX12
Hey Matthew, sorry to hear about this issue. Unfortunately I wasn't able to reproduce this; it works as fast as a newly opened emacs even when my emacs-uptime
is 16+ days :( (I am using Emacs 26.2 on MacOS)
The next time it gets laggy, could you try to use the CPU profiler to see which part of the tabnine client is being the bottleneck? I will try to pinpoint the issue. Thanks!
After an update to 3.3.57, I find myself unable to reproduce the issue. I will keep an eye on it and next time when there is a problem I will use a CPU profiler to make a detailed report.
Sorry for bothering again.
I just experienced such a laggy again. The issue is my Emacs is almost frozen thus no way that I can do to investigate this issue.
Due to the nature of this issue, i.e., intermittency. It's really hard to create an environment to reproduce the issue.
What I observe that this update version 3.3.57, can somehow solve the laggy issue by itself.
The next time it gets laggy, could you try to use the CPU profiler to see which part of the tabnine client is being the bottleneck? I will try to pinpoint the issue. Thanks!
I'm currently experiencing lag again, apparently Automatic GC
is taking up quite a bit of CPU. Could you share your gc-cons-threshold
with me?
Automatic GC is taking up quite a bit of CPU
Does this coincide? Today, I experience it again and I studied the CPU profile. I found GC is the most CPU consumer.
;;;; GC issue
(add-hook 'after-init-hook (lambda () (setq gc-cons-threshold 100000000)))
(add-hook 'focus-out-hook 'garbage-collect)
I currently have no laggy issue with this update but will keep an eye on it.
I'm currently experiencing lag again, apparently
Automatic GC
is taking up quite a bit of CPU. Could you share yourgc-cons-threshold
with me?
My gc-cons-threshold
is 200000000. Good to hear that it at least mitigated the issue.
After the update the GC threshold, I find most of the time there is no issue.
However, I find sometimes Emacs will be frozen and reject any command. The message shows the following information:
TabNine process #<process company-tabnine--process> received event "exited abnormally with code 1".
Restarting TabNine process.
It is unclear what causes the restarting process of TabNine.
I was wondering if it is possible to make starting process as asynchronous process.
I was wondering if it is possible to make starting process as asynchronous process.
I looked at the codebase, there is a commented company-tabnine-async
variable and a TODO: should we use async or not.
@TommyX12 What's stopping you guys from using async?
Maybe a good way is implement tabnine com https://github.com/emacs-lsp/lsp-mode as a language server, with this, the process run asynchronous, not depending on emacs
This is a great idea
Maybe a good way is implement tabnine com https://github.com/emacs-lsp/lsp-mode as a language server, with this, the process run asynchronous, not depending on emacs
Would this mean I have to choose between either tabnine or say, jsts-language-server? Because, I want both.
Same problem here. GC solution does not seem to work.
I'm working with emacsclient on an org file with company-tabnine (+other company backends) and I have to kill emacs client & server very often (~15min) :{ company-TabNine is clearly the problem as deactivating it in my .emacs make everything running smooth again.
I've tried killing TabNine/WD-TabNine processes, but they automatically spawn again, and Emacs/client get even more laggy...
The next time it gets laggy, could you try to use the CPU profiler to see which part of the tabnine client is being the bottleneck? I will try to pinpoint the issue. Thanks!
Can someone tell me how to "use the CPU profiler", so that I can find where is the problem on my side?
@liar666 and others
Can someone tell me how to "use the CPU profiler", so that I can find where is the problem on my side?
M-x profiler-start
M-x profiler-stop
M-x profiler-report
I don't know if this is duplicate of #39 or company-mode/company-mode#1056 , but similar issue.
In a newly opened Emacs, company-tabine works like magic and super fast, there is no performance issues at all.
However after using it intensively for hours, regardless of in single or multiple files, the completion gets extremely slow and lag literally on every input.
htop
shows tabnine process takes up quite a bit of memory, and executingM-x company-tabnine-restart-server
orTabNine::restart
doesn't fix the issue. Restarting Emacs makes tabnine works great again.Emacs 28 gccemacs, manjaro, config