Who would have thought that over a year after we saw Nvidia release their 10-series GPUs, we would be graced with another one? Likely the direct result of AMD’s Vega 56 and it’s price point, we are presented with the GTX 1070 Ti. The GTX 1070 Ti is a striped down GTX 1080 targeting the 1440p market and priced to offer direct competition to the AMD’s Vega. Nvidia was kind enough to send a Founder’s Edition for us to test, and if you’re in the market for a new card, or simply an enthusiast, you should take a look at this review.
- Base Clock: 1607 Mhz
- Boost Clock: 1683 Mhz
- Memory Clock / Data Rate: 4006 Mhz / 8 Gbps
- Total Video Memory: 8192 MB GDDR5
- Connectors: 3 x Display Port, 1 x HDMI, 1 x Dual-Link DVI
- Form Factor: Dual Slot
- Power Connectors: One 8-pin
- Recommended Power Supply: 500 Watts
- Thermal Design Power: 180 Watts
- Thermal Threshold: 94c
The GPU appearance is the standard style for all the Nvidia Founder’s Edition 10-series GPUs. It features a blower style cooling system that intakes air and exhausts it out the rear of the card where the various output connectors are. It’s not the best style of cooling there is and I’m sure we will see much better cooling systems come out from third party providers. It does, however, keep with the tradition of blower styles mostly being on reference cards that come straight from the manufacturer. That’s not to say that it’s bad - it does it’s job and allows for enough cooling to get the desired performance, but it reaches 83 degrees celsius quickly and thus thermal throttles quickly as well. In addition, we also have ‘GEFORCE GTX’ illuminated in green on the side when slotted in and given power. I was surprised by the single 8-pin connector I’m anxious to see if that’s something we will only be seeing on the Founder’s Edition or if third parties will follow suit.
Test system: AMD Ryzen 1700 OC 4.0Ghz w/ Corsair H100i V2, 16 GB GSkill 3000Mhz RGB RAM, Samsung Evo 850 256GB, Western Digital 7200 RPM 1TB x 2 in Raid-0 configuration, EVGA 650W Power Supply
NVidia 1070 ti Founder’s Edition
FRAPS was the tool used to record the frames per second maximums, minimums and averages. If a game provided an in-game benchmark tool it was utilized. If not, I chose a section of the game that was repeatable to allow for valid comparisons between resolutions. In general, I attempted to pick a portion of the game that represented the graphical load of the majority of its gameplay.
A Quick Note on Overclocking
First things first: you can overclock the 1070 Ti. There were a lot of rumors flying around that it was going to be a locked clock, but that isn’t the case. NVidia has asked its partners (EVGA, MSI, etc) to not offer overclocked cards out of the box. I believe this is to prevent partners from offering an overclocked 1070 Ti that could potentially out perform a non-overclocked GTX 1080. In response to this, many of the third party partners are going to include overclocking profiles in their software that allows for an already tested, one click overclock to be activated by using the included software.
In this review, I chose not to overclock the 1070 Ti. Blower style reference cards like the Founder’s Edition don’t have a lot of cooling overhead to facility an overclock high enough to have a meaningful impact on performance. Also, over the years I have discovered that most gamers aren’t actually overclocking enthusiasts. The testing was done on an overclocked processor to ensure we were bottlenecking the graphics card, not the CPU. Finally, most people familiar with overclocking know about the silicon lottery. Basically, two identical cards will not overclock the same - even in the exact same environment. So, for transparency and consistency you are getting benchmarks from a card straight from the retail box.
After the benchmarks were complete, however, I did test the overclocking capability of the GPU using MSI Afterburner. When Nvidia says that this card is an overclocking powerhouse, they aren’t lying. I was able to achieve a stable 2000 Mhz overclock on the 1070 Ti without thermal throttling with a custom fan curve and no aftermarket cooling solutions. It will be interesting to see what an overclocking enthusiast will be able to do with this card.
NVidia’s goal was to outperform the Vega 56 at the same price point while simultaneously targeting the 1440p (2560x1440) and ultrawide monitor users. I did not have a Vega 56 to run direct comparisons, however, NVidia provided a short list of comparison benchmarks for our viewing pleasure. These specific benchmarks were measured on a i9-7900x with both cards running at default settings and reference clock speeds.
We can clearly see that the 1070 Ti outstrips the Vega 56 in the selected games but I want to provide benchmarks for games played by MMORPG.com patrons, both multiplayer and single player varieties. My attempt was to test a wide breadth of games in order for you to determine if this is the right GPU for you. Benchmarks were measured at both 1080p and 1440p using the exact same settings with any type of frame limiter set to off.
Before we get to the charts I want to say that I was very impressed by this GPU. I’ve owned a 1080 in the past and currently game with a EVGA Hybrid 1080 Ti SC2 so I was expecting to see some major differences, but the frame rates were so high in most games that it was hard to perceive the change. I’m eager to eventually SLI two of them and do a true comparison to the 1080 Ti with 1440p, 4k, and 3440x1440 Ultrawide resolutions.
The chart is pretty self-explanatory, but I do have a few notes. First, FRAPS was not used for measuring the average FPS of Destiny 2. The developers actively decided to not allow numerous overlay programs (FRAPS, Afterburner, Discord, etc) to function with Destiny 2 and as a result the framerates were obtained through simply observing the in-game FPS monitor while playing through the first mission. It is purely an estimate to the best of my abilities.
Second, a note about Rise of the Tomb Raider. For some reason, the beginning of the in-game benchmarks start with an erroneous, low frame rate hitch. I ran the benchmark multiple times, on multiple hard drives and could not seem to ditch the hitch. I even went so far as to swap my 1080 Ti back in. Both cards measured minimums around 15 frames per second so I concluded it was software based. From what I could tell it was just a small handful of frames at the very beginning so the effect on the average should be minimal.
Unfortunately, I did not have a GTX 1080 or GTX 1070 available to draw direct comparisons, but they have been reviewed by our team in the past. Chris reviewed the EVGA 1080 SC2 and we have overlaps on a number of tested games. You can find his review here if you would like to draw comparisons. After a few calculations of my own I found that the GTX 1070 Ti appears to land at about 83-90% of the performance of a GTX 1080. Even though neither system should be bottlenecking either card, take these numbers with a grain of salt since the testing environment was not identical.
Overall, I am impressed with this card and I’m excited to see what Nvidia’s partner companies bring to the table. If you were looking at picking up a GTX 1070 I see no reason why you shouldn’t drop the extra $50 or so to get the Ti - you’ll end up closer to the GTX 1080 side of the spectrum than the pure GTX 1070. It is clear that the GTX 1070 Ti is exactly the card to get if you want to push insane frame rates on a 1080p monitor or enjoy buttery smooth visuals on a 1440p with everything cranked up to 11. Some may argue that the 1070 Ti is a pointless iteration in the 10-series line up, and maybe it is, but to that I say “at least we have more options.”
This product was provided by the manufacturer for the purpose of review.