Improve rendering of symbols in my terminal emulator

[chilly void]

https://github.com/commmrade/kemul
So I'm making a terminal emulator
And now I'm trying to optimize rendering and would like to get some advices about that
Right now only optimization I have is a texture atlas, where I store different kinds of symbols

Also, I'm gonna store symbols in a 2d vector of Cell structures soon, so I can store bg, fg color and etc.
Cell is gonna like like:

• uint32_t codepoint
• SDL_Color fg_color
• SDL_Color bg_color
• uint16_t flags (underlines, bold, etc) (bit flags)

Would love to hear some advices on how to optimize rendering of Cells as well.

std::vector<uint32_t> codepoints;
        utf8::utf8to32(buffer.get_buffer()[i].cbegin(), buffer.get_buffer()[i].cend(), std::back_inserter(codepoints));

Right now I turn text into codepoints like this
But with Cells it's gonna be weird I think, I will have to extract codepoint out of Cell struct and I suppose it will affect performance, so how should I go about that?

[cobalt ruin]

When your question is answered use !solved to mark the question as resolved.

Remember to ask specific questions, provide necessary details, and reduce your question to its simplest form. For tips on how to ask a good question use !howto ask.

[unkempt wing]

I mean, from this snippet here, you call buffer.get_buffer() which returns, something, then you run [i] which returns, something, and then you access the iterator,
this here is duplicate code which will run twice for no reason,
move the repeating part in some local variable and use that

outside of the obvious like this^, run a profiler and observe where the real bottlenecks are

[chilly void]

As for profiling
I can use something like valgrind right?

[unkempt wing]

oh and the <algorithm> header may cause you to run N deep copies of some heavy object you have, and thus crush your performance
like this guy:
https://www.youtube.com/watch?v=mOSirVeP5lo&pp=ygUVY2hlcm5vIEkgb3B0aW1pemVkIGl0

[unkempt wing]

that's more for bug huntung, not profiling bottlenecks in performance

[chilly void]

what would u recommend fkor profiling bottlenecks

and on linux

[unkempt wing]

I don't have practical experience with profilers sorry

[unkempt wing]

the shell comand gprof maybe

kinda like gdb's cousin