The next time I'm shooting in a dimly lit warehouse with a camera that can't go over ISO 100, I'll keep those overall scores in mind.
For those who don't know, the DxOMark Lens Score is based mainly on performance in 150 lux illumination at 1/60 s and ISO 100, and is only slightly influenced by optically important things like sharpness, vignetting, distortion, and CA (or so I infer from comparing many of their measurements, because the don't actually say how their scores are determined, it's a 'black box' calculation). The 50mm f/1.8 II gets an overall score of 28 on the 5DIII (which makes sense in context, because at 1/60 s and ISO 100 in a dim warehouse or an hour before sunrise, I'd pick the f/1.8 lens over the f/4 lenses, too!).
Of course, with the Tamron lens you have the option of using it in a Nikon mount, and if you're relying on DxOMark Scores, you really should use the Nikon version. For example, with the Zeiss 21mm f/2.8 (which Bryan suggests may be the best wide angle lens available), the Nikon mount version gets an overall score of 25 on the D800, whereas the Canon version gets only a 22 on the 5DIII…despite the 5DIII measurements showing higher sharpness and transmission, less distortion, and equal vignetting and CA. The D800 has higher dynamic range than the 5DIII at ISO 100, and that means the Nikon mount version of the same Zeiss lens gets a higher Score.
Basically, I'd recommend ignoring DxOMark's Scores completely. You can get some useful information from their Measurements if you look at the details, but be careful there, too. For example, they initially stated that the 70-200/2.8L IS II was not quite as good as the MkI version of that lens…when they were called on it by pretty much everyone who'd used or tested both, they defended their tests and said 'no mistake' but later quietly fixed their measurements to show the MkII as better. Likewise, their measurements show that the 17-40L wide open is nearly as sharp in the corners as in the center, and that the corners of the 17-40 at f/4 are substantially sharper than the corners of the 16-35 II with the latter stopped down to f/5.6. Both of those are clearly wrong, as you can see from the ISO 12233 comparison.