OSDev.org

Ok, I have googled and wiki'd and the only thing I have been able to find to help me understand Dynamic Translation is QEMU docs...

Does anyone here know about it and care to explain, or know of some page of some sort that gives more in-depth information. I am considering implementing this, so I really need a lot of information about it before I try such a daring thing...

Namely, what I am wanting to know is
1. How can dynamic translation be fast unless it is multithreaded?
2. Can it be portable among different archs?
3. How do I get started with it?
4. How does caching work with it?

I have a basic understanding of wha tit is, but I don't know how it works..

Does this clear anything up?

I guess you tried the wrong keyword:
http://en.wikipedia.org/wiki/Dynamic_recompilation

dynamic translation is fast because most of the time the same code is called over and over(the timer isr comes to mind). QEMU simply stores the compiled pieces after it has been translated and every time that that piece of code is called after, the machine code version is used which is faster.