I've had the exact same problem. Using New Relic (a web application performance app) I was able to trace the issue to the cache purge that Nginx Helper implements. What Nginx Helper does is use wp_remote_get() to "fetch" the URL, with the /purge/ in the URL, to purge the cache. Slightly more than 90% of that update/publish time was due to this cache purging.
Here is my output from New Relic, tracing what's taking time: http://cl.ly/image/2O242I0l3L3c
You can check what your Nginx Helper is purging by setting Log to "Info" and looking at /wp-content/uploads/nginx-helper/nginx.log
I minimized the load time doing two things.
First, I unchecked the options for "Purge Archives", since it was fetching every related URL for the post (date archives, category archives, tags, etc).
Second, in the purger.php file, on Line 205, I switched "feed =" to "false". I don't cache feeds. And what was happening is that Nginx Helper tries to purge the cache of every URL, plus adding /feed/, /feed/atom/, and /feed/rdf/ onto the end of every URL. Each of these requests took time.
Here's the line I changed: https://github.com/rtCamp/nginx-helper/blob/master/purger.php#L205
And voila! It only takes a fews seconds to update/publish a post now. This isn't a perfect solution, since any update will break the purger.php edit, but for now it's at least tolerable.
Make sure to follow up and check what's being purged by looking again at /wp-content/uploads/nginx-helper/nginx.log