As our friend Neil said, may be a myth, but although I do not have done a comparative test methodical (and being therefore of questionable value) also observed differences in accuracy between the firmware installed on 3490/3790 (I will not angry if someone disagree, lol..). At the latitude and place where I am and in the averages of observations that I have made with the 3790/3490, "accuracy" has been around for 9 to 12 meters to up, although a few times to have reached 6 meters. Given the imprecision of road maps is perfectly acceptable and usable these values. We also have to admit that there may be variations between these models on a production line, so the need for more rigorous testing to clarify definitively the question (multiple devices, same conditions, etc. ..). But, from what I have observed, I believe that there really difference between the signal quality between the two firmwares, even because the chipsets of 3790 and 3490 are different and should also be the drivers that control and optimize its operation.
Of one thing I am sure, the 3790/3490 (v7.60) is less accurate than the nuvi 2595 (v7.60) because I did a comparison test between the two, on a journey of 1 hour with the vehicle in motion and watching intently values (traveling as a passenger). With both GPS side by side, while the accuracy of the 3790/3490 ranged between 9 and 11 meters to up, of the 2595 was around 3 meters (sorry not having captured images).
By way of curiosity too, despite being among devices of characteristics and goals different, while the Etrex 30 and Oregon 450 maintains precision of 3 meters (without WAAS), the 3790/3490 keeps 10 meters, on average, both side the side and at the same time.
In short, in my simplified tests, but careful, the 3790/3490 is less precise and, despite continuing to be perfectly usable, ideally could maintain the same level of accuracy of his new brothers.