I guess that could work as long as pass 2 is just as slow or slower than pass 1.
No - it doesn't. The entire point of running two passes is that you need to have ALL the info from the first pass in order to make more accurate decisions for the second pass.
Parallelizing video encoding isn't that hard in essence. x264 - a popular H.264 encoder scales very well even with 8 cores, provided you can feed it fast enough.
Typically it's achieved by splitting the video frame into separate pieces, and encoding those on each thread. You can also split the video into a few pieces, and execute multiple instances of the encoder. This is what QuEnc^n and HC^n do automatically.
Other things like motion-compensated filtering do not multithread well - yet.
If you want to encode MPEG-2 to H.264 at very high speed, learn how to setup MeGUI. Use one of its faster profiles, and make sure it's auto threading. Slower generally means lower quality - I typically let an encode run overnight to get a really good backup.
Doom9 is a great place to read about this stuff - lurk their forums. PM me on there or here with questions..
btw
Nero Recode using AVC h.264 is still the fastest and most efficient encoder I've ever used. Pay good money to get good things.
No. x264 is a lot better, and it's free. Sorry...
~MiSfit