How to reduce the time of clang_complete search through boost - vim

Question

I like using clang with vim.
The one problem that I always have is that whenever I include boost, clang goes through boost library every time I put "." after a an object name. It takes 5-10 seconds.
Since I don't make changes to boost headers, is there a way to cache the search through boost?
If not, is there a way to remove boost from the auto-completion search?
update (1) in response to answer by adaszko
after
:let g:clang_use_library = 1
I type a name of a variable.
I press ^N. Vim starts to search through boost tree. it auto-completes the variable.
i press "." and get the following errors:
Error detected while processing function ClangComplete:
line 35:
Traceback (most recent call last):
Press ENTER or type command to continue
Error detected while processing function ClangComplete:
line 35:
File "<string>", line 1, in <module>
Press ENTER or type command to continue
Error detected while processing function ClangComplete:
line 35:
NameError: name 'vim' is not defined
Press ENTER or type command to continue
Error detected while processing function ClangComplete:
line 40:
E121: Undefined variable: l:res
Press ENTER or type command to continue
Error detected while processing function ClangComplete:
line 40:
E15: Invalid expression: l:res
Press ENTER or type command to continue
Error detected while processing function ClangComplete:
line 58:
E121: Undefined variable: l:res
Press ENTER or type command to continue
Error detected while processing function ClangComplete:
line 58:
E15: Invalid expression: l:res
Press ENTER or type command to continue
... and there is no auto-compeltion
update (2)
not sure if clang_complete should take care of the issue with boost. vim without plugins does search through boost.
superuser has an answer to comment out search through boost dirs with set include=^\\s*#\\s*include\ \\(<boost/\\)\\#!

Answer 1

So, you have at least two options. Option #1 is to set g:clang_use_library
to 1. Here's what :help g:clang_use_library says about it:
Instead of calling the clang/clang++ tool use libclang directly. This
gives access to many more clang features. Furthermore it automatically
caches all includes in memory. Updates after changes in the same file will
therefore be a lot faster.
This requires working setup of Python Vim integration though.
Option #2 is to set g:clang_complete_auto to 0 by which you disable
automatic completion after ->, ., :: and may use <C-x> <C-o> instead
to manually invoke clang_complete whenever you wish.
I use both.

Answer 2

from here, you can add the following to your .vimrc:
:set include=^\\s*#\\s*include\ \\(<boost/\\)\\#!
(the question of caching the search through boost is still open though)

Answer 3

I've done a lot of performance enhancements to clang_complete, you can check about this at issue #187. Big part of the problem was just weak script performance due to non optimized code.