Personally I found the DVDInfoPro scans with the 108 to be wildly out. Scanning the same disk with Philips/Benq or Liteon drives (using CD-Speed/KProbe) produced similar shaped graphs but the figures were wildly different. Remember the PI/PIF scanning (in DVDInfoPro with the 108) is still in beta and imho needs a LOT of work.
Getting back to G04, of which I have a large sample of different types. I found best examples of the media will burn at 12x with far more consistant and reliable readings from CDSpeed. Most arent what I would call a good burn but while PI/PIE rates are somewhat elevated in the 12x burn area they generally dont exceed the "official" limits.
Average examples will burn at 8x at reasonable quality, and even the real junky examples of G04 will burn at 6x without noticeable difference in quality from 4x burns.
I've never found an example of G04 which will burn at 16x, all the types Ive tried either stick to 12x to the end or even drop down to 8x.
You just got to ask yourself, is your data worth the risk for the small amount of time you save?