GP106 vs GP107, What is the difference?

[Kiyometa]

Hello fellow nerds.

 

I have a question that has been bothering me and haven't been able to get a very clear answer from anyone.

What is the difference between for example a GTX 1050Ti that uses the GP106 vs GP107?

 

I know that there is different amounts of rendering stuff with the GP106 having more. GP106 has a quite a bit more Shading units (1280 vs 768), TMUs (80 vs 64), and ROPs (48 vs 32).
I also know that GP106 is TSMC 16nm with a density of 22M and the GP107 is 14nm Samsung with a density of 25M.

What does this mean for Gaming performance and/or video decoding/encoding?

 

As a general rule, I've thought the higher density is USUALLY better performance, but its the "same" product. so did they trade amount of units for density to reach a relative performance parity?

That being said, both Nvidia and AMD have released cut down versions of their cards under the same name with little differntiation so is this simply one of those cases where GP107 is the "same" product but a cut down version?

 

Thanks to any illumination anyone can offer.

[Jurrunio]

3 minutes ago, Kiyometa said:

GP106 has a quite a bit more Shading units (1280 vs 768), TMUs (80 vs 64), and ROPs (48 vs 32).

except they cut all of the numbers down to match the GP107 version of 1050ti. In benchmarks they perform equally.

 

The GP106 does have a bigger die area, which means lower heat density and lower temperatures.