The 780 was 87% as fast as the 970 (84/97)
https://www.techpowerup.com/reviews/MSI/GTX_970_Gaming/27.html
The MSI 1050 Ti was 72% as fast as the 970 (100/139)
https://www.techpowerup.com/reviews/MSI/GTX_970_Gaming/27.html
That gives the 780 a 21% speed advantage. VRAM is a non-issue at 1080p
However, if you can manage a 1060... that's a 74% speed bump.
A careful look at techpowerups reviews of the 1060 3GB and 1060 6GB is very revealing. the 3GB model has 10% less shaders on the GPU and this accounts for a speed disadvantage at 1080p of 5-8%. Now if we wanted to look at the impact of VRM between 3 GB and 6GB we could look at the 18 games in the test suite and be able to easily demonstrate substantial changes in that performance difference when we jump to 1440p and it should be astounding at 2160p. Problem is, it never occurs. The performance gap between the two cards increases by just 0 - 2%.
The overemphasis placed by many on VRAM stems from misunderstanding of tools which purportedly measure VRM usage. No such tools exist ... Afterburner does nit do this; GPUU_z does not do this. What they do is measure how mch VRAM the game has decided to allocate to potential use ny the game.
https://www.extremetech.com/gaming/213069-is-4gb-of-vram-enough-amds-fury-x-faces-off-with-nvidias-gtx-980-ti-titan-x
GPU-Z: An imperfect tool
GPU-Z claims to report how much VRAM the GPU actually uses, but there’s a significant caveat to this metric. GPU-Z doesn’t actually report how much VRAM the GPU is actually using — instead, it reports the amount of VRAM that a game has requested. We spoke to Nvidia’s Brandon Bell on this topic, who told us the following: “None of the GPU tools on the market report memory usage correctly, whether it’s GPU-Z, Afterburner, Precision, etc. They all report the amount of memory requested by the GPU, not the actual memory usage. Cards will larger memory will request more memory, but that doesn’t mean that they actually use it. They simply request it because the memory is available.”
This is kinda like what credit reporting agencies do ... when you apply for a loan, the reporting agency reports that you have a $5,000 liability on your Visa card tho you only have $500 charged towards your $5000 limit. You have not "used it" but you could if you needed to.
Alienbabeltech was the 1st to show this when the compared a 2GB and 4GB 770.... most of the titles showed no difference in performance at 5760 x 1080. The few that did, were unplayable ... they even tricked the system when they tried Max payne with the 2 GB it would not allo selecting 5760 x 1080... so they stuck in the 4GB card, ran the game, swapped the cards back and it now let them select 5760, it got the same fps and the same image quality. Extremetech, in the link above could only make the games break 4 GB when at 2160p and highest setting but again, does VRAM really matter when 4GB gets you 15 fps and 8GB gets you 19 ... in all cases, the game was unplayable with either 4 or 8GB. Guru3D and Puget sound did similar testing and got the same results.
At 1080p, 3 GB will be more than fine ... at 1440p, look for 6 GB and at 4k, 12 GB is the target, tjho the best we can do today is the target.
Another way to look at it is ... if the argument is made that 11 GB (or less) is fine for 4k ... a 4K screen = four 1080p screens.... so 11 GB / 4 = 2.75 GB