Re: [PATCH (performance tracing)] test git-status performance

From: Duy Nguyen <pclouds@gmail.com>
Date: Jun 26, 2014, 12:30 UTC
Message-ID: <CACsJy8Bkfz6=WUy0RsXd0e9dXqocUFXhAaO0dpUg+AjZ5RNeNA@mail.gmail.com>
In-Reply-To: <53AA0228.4080600@gmail.com>

On Wed, Jun 25, 2014 at 5:56 AM, Karsten Blees <karsten.blees@gmail.com> wrote:

Show 21 quoted lines

>  void wt_status_collect(struct wt_status *s)
>  {
> +       uint64_t start = getnanotime();
> +
>         wt_status_collect_changes_worktree(s);
>
> +       trace_performance_since(start, "wt_status_collect_changes_worktree");
> +       start = getnanotime();
> +
>         if (s->is_initial)
>                 wt_status_collect_changes_initial(s);
>         else
>                 wt_status_collect_changes_index(s);
> +
> +       trace_performance_since(start, "wt_status_collect_changes_index");
> +       start = getnanotime();
> +
>         wt_status_collect_untracked(s);
> +
> +       trace_performance_since(start, "wt_status_collect_untracked");
>  }

Now that we have good perf measuring support, perhaps the next step is remove gettimeofday() in advice_status_u_option related code in wt_status_collect_untracked().

-- 
Duy