I don't know anything about the J35Z2/3, but the J30A4/5 definitely gains some power running premium gas. And if the old engine does, why shouldn't the new one?
Anyway...
Last winter, I used an Auterra dyno scan tool to log ignition timing advance in my 2007Accord V6.
The first runs were using 87 octane. In order to ascertain the ignition timing advance with the engine going all out, I made several WOT runs through 1st and 2nd gear. I noted the timing curves, intake air temp, and coolant temp, and then filled up with 93 octane after completely draining the tank of 87.
I drove about 500 miles on 93 before conducting the 2nd set of tests in order to give the engine computer time to adjust to the higher-octane fuel. I picked a time with nearly identical weather conditions as during the first set of runs, and went at it.
My results weren't astounding, but they definitely proved that the J30A5 ran somewhat more aggressive ignition timing on 93 octane. As I did WOT runs from 0 to 70, I was able to compare the timing advance between the two grades of gas from as low as 2500 rpm all the way to the redline.
I found that, with 87 octane, ignition timing advance dropped to as low as 12* at 2500 rpm and then reached a plateau of about 23* at 5000 rpm on its way to a definite peak of 24.5* at 6000 rpm. All of the runs were within a degree of each other throughout the powerband, and the peak timing was 24.5* every run.
On 93 octane, the peak only rose a fraction to 25*(there was one run that peaked at 26.5, but the others were steady at 25), but the timing advance in the low-end and mid-range increased dramatically. Whereas the timing advanced dropped to as low as 12 degrees on 87, it never fell below 19 degrees on 93, even at a low 2500 rpm. The curve was much flatter, and the peak was less defined, with a steady 25* of peak timing advance spanning from about 4500 rpm to the redline. On the run with the 26.5* peak, that peak came at 6000 rpm while the rest of the curve from 4500+ rpm was still around 25*, so I disregarded that run as it was an outlier as far as peak timing advance was concerned.
To be sure that all things were equal, I also measured the intake and coolant temps during the runs. The average intake air temp for both sets of runs was about 95 degrees F, and the coolant temp during all of the runs was right around 180 degrees F, give or take a few degress. I wanted to make the comparison as fair as I could in order to obtain the most accurate results possible.
All in all, I concluded from my testing that the J30A5 does indeed benefit from premium gasoline, especially in the low-end and mid-range. I don't know how much power each degree of timing advance is worth, but the extra 7* of timing advance at 2500 rpm when using 93 has to be pretty significant. I can definitely feel the extra low-end on the ole' butt dyno when running premium, and overall the engine behaves in a much more refined manner. When run on 87, my V6 feels and sounds like it's in pain. It pings at low rpm during high-load situations and feels kinda rough at idle and during acceleration. With 93 in the tank, it smooths out quite a bit and the pinging disappears. Based on my experience with the J30A5 on regular 87 octane, I don't see how Honda can confidently recommend 87 octane for this engine. Surely, they know better than anyone that this engine HATES low octane gas, and with a 10:1 compression ratio, why shouldn't it?