NVIDIA GeForce GTX 1660 Ti Linux Gaming Benchmarks

[cybertraveler]

@Micheal: How about a power draw shoot-out with all cards running at 1920x1080 and capped by vsync @60 fps ?

+1

This would be fun.

[cybertraveler]

Michael - how many times do you run each test? IE for each chart you make, how many times did you run the test for each card?

[Michael]

Michael - how many times do you run each test? IE for each chart you make, how many times did you run the test for each card?

It's a minimum of three times for each benchmark but increases if the variation goes beyond 3.5% between runs. The number of run times, all individual raw results, etc, are available from the OpenBenchmarking.org result file.

[cybertraveler]

Thanks

[JanW]

The power consumption of the GTX 1070Ti seems bizarre. Median power consumption over all tests looks to be 30W lower than GTX 1070. Is this expected?

[cj.wijtmans]

The power consumption of the GTX 1070Ti seems bizarre. Median power consumption over all tests looks to be 30W lower than GTX 1070. Is this expected?

yeah. when something is better/faster at executing something its done faster and consumes less power. the ti is an overclock and not an increase of transistors

[TheYoshiGuy]

Thanks for this Michael! Great range of cards and tests included here.

[darkbasic]

Vega 64 is on par with the 1080

[JanW]

yeah. when something is better/faster at executing something its done faster and consumes less power. the ti is an overclock and not an increase of transistors

This is the first time I hear of overclocking as a means to reduce power consumption - and to be honest I don't believe this explains the numbers. These benchmarks are not a race to idle, they try to crank out as many frames as they can in a fixed amount of time. The 1070Ti rendered many more frames in total than the 1070 during this testing.

EDIT: Just looked up the specs. The 1070Ti has more CUDA cores (2432 vs 1920), but lower clocks than the 1070, since Nvidia does not allow factory-enabled overclocking on the Ti. On page 2 of the article, Michael even quotes 721MHz clock for the GTX 1070Ti vs 1506 for the 1070. (This can't be right? How would a 1070Ti at 721MHz outperform a 1070 at 1506MHz?)

The more I look at this, the more I think that the power consumption data between 1070 and 1070Ti may be switched? The difference is suspiciously close to the difference in TDP (150W vs 180W), just the wrong way. At these resolutions, the cards may be power limited (at least they do not seem to be thermal throttling according to the temps). Or did you raise the power limit for the 1070 and not the 1070Ti Michael?

[Guest]

https://store.steampowered.com/hwsur...platform=linux

1920 x 1080 50.99%
1366 x 768 17.27%
2560 x 1440 6.07%
3840 x 2160 3.63%

It's crazy that people are still gaming on 1080p (unless those are PCs hooked up to a TV in a living room). I got my first 1440p monitor almost 10 years ago, and even before that I got my 24" 1200p screen more than 15 years ago.

I wouldn't even know where to find a non-TV 1080p screen nowadays except maybe on a smartphone or some cheap 13" notebook.