Re: [ft-devel] gray_render_scanline: very hot when fuzzing

[Kostya Serebryany]

BTW, w/o rendering the profile looks like this:
25.92%  tt_cmap4_next

15.37%  tt_cmap4_char_next

7.32%  FT_Get_Next_Char

5.11% memset

Does this look normal? 

--kcc 

On Tue, Oct 6, 2015 at 9:09 PM, Alexei Podtelezhnikov <address@hidden> wrote:

On Tue, Oct 6, 2015 at 11:43 PM, Kostya Serebryany <address@hidden> wrote:
> On Tue, Oct 6, 2015 at 7:59 PM, Alexei Podtelezhnikov <address@hidden>
>> Anyhow, there is not much left to do or fuzz after
>> rendering, so do not render
>
> If you think that fuzzing the rendering is not that interesting, I just
> won't. It further reduces the time from 71 seconds to 4!

Well, this is where the library converges for scalable fonts. You
definitely pound this code excessively in comparison to other parts.
You've got to stop and say enough at some points, or, perhaps
stochasticly, reduce the frequency of rendering requests 10 fold or
so.