While it is the parallel aspect of the T3 that's the revolution, I'm not quite sure how a regular top + Y-split would fare. The T3 was designed to do parallel while adding a Y-split on another top would be a 'lossy' addition (and a noticeable one at that, I reckon).
I'm very familiar with the numbers the T3 is capable of (been working with skinnee on building an estimator for it) and without giving too much away without his blessing, I'll say what I can about the performance. In a single loop config, its PQ curve is similar to that of the Koolance D5 top. But when 'unleashed' and using both flowpaths (i.e., parallel), the PQ curve is more like the EK's or Detroit Thermo's....and that's not a property of being in parallel, that's just the fact that the top performs better when all ports are used (i.e., as it was intended).
My guess is that a loss similar to the loss that exists when going from LILO+TITO to just LILO or TITO will also exist (though for a different reason) on another top that uses a Y-split.
As for the example that skinnee gave, it's a pretty bad one in the grand scheme of things. Loops that are all-inclusive and have restrictive components will see a big boost from the T3, especially if you isolate the CPU block.
Dual system scenarios also work well

Bookmarks