As far as rolling resistance some manufactures, namely continental and Vittoria claim their testing has shown clinchers to be superior. There are also some other tests, such as the one conducted by Tour, and roues artisanales (conducted at continentals test facilities) that show a noticeable advantage with clinchers. Comparing the spread in the later test between the best clincher and best tubular would indicate you are giving up nearly 16 watts with tubulars.
Now Al Morrison has done probably the most comprehensive testing of road bike tires out there, and his tests show a slight advantage in favor of tubulars. That said he is using vastly different equipment and test protocols, and no disrespect to him and all the hard work he has put into it, but for my money I'm going to go with what the multi-million dollar R&D facilities are showing, and not some guy with rollers in his garage.
Except Al Morrison documents everything: how he glues the tubulars, how he controls for temperature, how he does the measurement. Since we've established glue technique matters, without knowing the glue technique used by the magazine, how can we be sure the results are fair?
For example, I might conclude using tape is the best comparison, since it's reproducable, and glue has innate variability. But we know tape yields worse Crr. Or using no glue at all is reproducable, but gives poor Crr, perhaps counter-intuitively to many.
Note the tire manufacturers have a profit interest in this. Al Morrison has none.