The new Nvidia GTX 1660 Ti and older GTX 1070 , are both good gaming graphics cards
In this Blogs we’ll test out and compare 16 different games at 1080p, 1440p and 4K resolutions to help you see the performance difference between them, and find out which is better value for the price.
Let’s quickly take a look at how the GTX 1660 Ti and 1070 actually differ in terms of specs, note that things like clock speed and power will vary between specific cards, these are just the reference specs for each model.
The 1070 has more CUDA cores, and only just slightly higher base clock speed, while the boost clock on the 1660 Ti is higher.
The 1070 also has more memory, but it’s GDDR5 with different speed.
For the testing I’m using the MSI GTX 1660 Ti Gaming X and the Galax GTX 1070 EX, so both cards are factory overclocked out of the box, which will slightly affect frame rates, so expect slightly different results with different cards.
GHz in an MSI Z390 Gaming Pro Carbon motherboard, along with 16GB of T-Force Night Hawk CL16 memory from Team Group running at DDR4-3200 in dual channel.
Check the links in the description for details on all of the components as well as for up to date pricing.
The same Windows updates and Nvidia driver 419.
5 was used for the testing, so let’s get into the results.
Let’s start off with Apex Legends, which was tested with all settings maxed out.
In all upcoming graphs I’ve got the GTX 1070 shown by the top bar in red, while the GTX 1660 Ti is shown below in purple, while the three different resolutions, 1080p, 1440p and 4K are shown on the left.
There were also very nice improvements to 1% lows at lower resolutions too, up to a 27% increase with the 1660 Ti at 1080p.
The 1070 was coming out ahead of the 1660 Ti in all resolutions at this test, with 6% higher average FPS at 1080p, 7% at 1440p, and 15% at 4K.
Overwatch was tested in the practice range as I can easily perform the same test run compared to playing with bots or other players which will differ every time.
While both of these graphics cards were capable of delivering a playable 4K experience, even with ultra settings, the 1070 was consistently ahead in this game.
At 1080p the 1070 was 7.
% ahead of the 1660 Ti in average FPS, 7.
% at 1440p, then 9.
% at 4K.
PUBG was tested using the replay feature with the exact same replay at high settings.
This was another game where 1440p with good settings was quite playable, but I wouldn’t want to step up higher to 4K with either.
The 1660 Ti was coming out ahead in average FPS at all resolutions tested, however it was quite close.
At 1080p the 1070 was actually seeing 12% higher 1% lows, which I’d argue is more beneficial over the 2 FPS difference to the average, given both are already over 100 FPS, however this changed at 1440p with the 1660 Ti now 18% ahead in 1% low.
Watch Dogs 2 is a resource intensive game and was tested with high settings.
In my opinion this one doesn’t need a high frame rate to play, I can get by with a solid 30 FPS, so both cards were actually giving me a playable experience at 4K.
At all resolutions though there was basically no noticeable difference between the two.
Ghost Recon was tested with the built in benchmark and high settings in use, and is another resource heavy game.
The 1070 was just slightly ahead of the 1660 Ti here, but only by a small amount, just 0.
% at 1080p and 1440p, then 4.
% at 4K.
The Witcher 3 was tested with Hairworks disabled, and like most other games was playable at both 1080p and 1440p even with high settings in use.
The 1660 Ti was coming out ahead in all three resolutions, however it was a close match with the 1070 not far behind, with no real practical difference between the two.
Shadow of War was tested using the built in benchmark with high settings, and this is another game that I’ve found to favour Nvidia’s new Turing architecture.
In this case it doesn’t really seem to be making much of a difference though, with the 1660 Ti 1.
% ahead at 1080p, 3.
% ahead at 1440p, and 2.
% at 4K.
In terms of overall improvement, over all 16 games tested with a 1080p resolution on average the GTX 1660 Ti was performing just 0.
% better when compared with the GTX 1070 in terms of average FPS.
There wasn’t really that big of a gain on the 1660 Ti, other than from the games that really seem to favour the Turing architecture, like Rainbow Six Siege.
At 1440p on average over the same 16 games the GTX 1660 Ti was now scoring 0.
% higher average frame rates when compared with the GTX 1070, so only a very small change on average from what we saw at 1080p.
Again we only seem to be in the positive for the 1660 Ti due to Rainbow Six Siege.
At 4K we’re now seeing the scales tip towards the 1070, with more games performing better with it compared to the 1660 Ti, perhaps due to the larger memory of the 1070 which becomes more useful at these higher resolution.
Realistically though neither of these are 4K cards.
When it comes down to it, I don’t think there’s very much of a performance difference between these two graphics cards, they’re quite similar once we average things out, but it does of course depend on the game.
It’s also worth noting that the GTX 1070 does have SLI support, while the 1660 Ti does not, so technically you could expand the 1070 with a second card to improve performance, however considering how hit or miss SLI support is in games I wouldn’t consider that as an upgrade path, and would instead recommend the best single slot card you can get at the time.
I haven’t tested overclocking here, as results aren’t guaranteed and will vary between hardware based on what overclocks you can get, realistically it generally only makes a couple of FPS difference, but I will cover overclocking in the full review videos.
Now for the final difference, the price.
I suggest checking updated prices using the links in the description, as prices will change over time.
At the time of recording, in the US my GTX 1660 Ti I’m testing with is $310 USD.
I am using a higher end 1660 Ti however, it just happened to be a good deal for me here in Australia when I bought it, lower end models can be had for around $280 though.
As for the GTX 1070, well prices are a bit sad there, they seem to be around $450 USD new minimum on Newegg.
Many stores are already no longer even stocking the GTX 1070, as supply starts to dry up.
This will likely mean that over time, the price will continue to go up further as they become difficult to buy new.
As a result the second hand market will eventually become the only practical option, if not already, which completely changes the value proposition.
I got my 1070 second hand for about $220 USD in October 2018 for example.
Looking at brand new prices, I think the 1660 Ti is the clear winner, simply because it’s only just launched it’s much more widely available, and as we’ve seen here it is capable of similar levels of performance on average, though it does of course vary by game, although it is possible that in future the extra memory of the 1070 may be more beneficial.
If you don’t mind not getting brand new with a warranty, then look for a second hand 1070, there’s great value there.
If we plug the numbers in, I spent $220 USD on my 1070 and $310 USD on my 1660 Ti, so I’ve spent 40% more money on the 1660 Ti.
If I got a cheaper $280 1660 Ti instead then I’m still paying 27% more money over the used 1070, but this is of course only the difference in graphics cards.
When you factor in the whole price of a system, the percentage value is much less.
Basically if I was buying new today, I’d just go with the 1660 Ti and get the warranty, but if you’re on a tight budget and want to save some money try and find a used 1070.
Otherwise if you’re looking to upgrade and already have a 1070, well it makes no sense to go a 1660 Ti, I’d be looking at something higher.