The GTX 1080 Ti Founders Edition Tested
Today is the day that all of you have been waiting for, as we get to release the review of the Nvidia GeForce GTX 1080 Ti Founders Edition. This card has been teased, unboxed, and is speculated all over the globe due to its Ti moniker – symbolizing its importance at Nvidia’s current gaming graphics card roster.
The Nvidia GeForce GTX 1080 Ti is the latest Pascal architecture based graphics card from Nvidia featuring a price point of 699 USD or roughly 35, 000 PHP (+ 20% markup locally). It effectively slots it self at the high-end enthusiast market, along with the Nvidia GeForce GTX 1080 and the Titan X. That being said, Nvidia touts it as the ultimate gamer GPU with DirectX 12 and Virtual Reality gaming domination in mind.
A TITAN IN DISGUISE:
The GeForce GTX 1080 Ti is built around the same GP102 GPU utilized by the Titan X Pascal. That said it comes with 3584 CUDA Cores, 28 Streaming Multiprocessors, and runs at a base clock frequency of 1480 MHz. With 224 texture units, the GeForce GTX 1080 Ti’s texture filtering rate is 354.4 Gigatexels per second at its Boost Clock of 1582 MHz. The card also features the same GDDR5X SGRAM, albeit with a gigabyte less inside.
|GTX 1080 Ti||NVIDIA Titan X (Pascal)||GTX 1080|
|Core Clock||1480 MHz||1417 MHz||1607 MHz|
|Boost Clock||1582 MHz||1531 MHz||1733 MHz|
|TFLOPs (FMA)||11.5 TFLOPs||11 TFLOPs||9 TFLOPs|
|Memory Clock||11Gbps GDDR5X||10Gbps GDDR5X||10Gbps GDDR5X|
|Memory Bus Width||352-bit||384-bit||256-bit|
Bearing similar amounts of CUDA cores, Texture Units, Transistor Count, and a great deal of other similarities, the 699 USD GTX 1080 Ti appears to be a Titan killer. That in mind, Nvidia had to disable one memory controller out of the GP102’s 12 controllers – allowing the 1080 Ti to run with 11 GB of total memory capacity. Each memory controller has a Bus Width of 32-bit, and carries 8 Render Output Units. With that said, Memory Bus Width totals at 352-bit, and ROPs is now at 88 instead of the full 96.
With those differences in place, one would think that the GTX 1080 Ti is pegged to deliver a significant performance hit over the Titan X. On paper, that is simply not the case – as this card features 11.5 TFLOPs of performance according to Nvidia’s own tests. That’s a 500 GFLOP differential, but take note it’s the maximum theoretically. Memory Speed is also clocked at 11 Gbps over the Titan X’s 10 Gbps. This together with the clock rate, bus width, and data rate allowed the 1080 Ti to feature 484 GB/s of memory bandwidth. That’s 4 GB/s more than the Titan X.
THE NVIDIA GEFORCEGTX 1080 TI FOUNDERS EDITION:
Since we’ve already unboxed the Nvidia GeForce GTX 1080 Ti on our preview article located HERE, we decided to skip the formalities of reintroducing it on this review. This card bears similarity to the Founders Edition cards before it shroud-wise with the exception of the GTX 1060 Founders Edition. Still, we are going to highlight its major differences over the older iterations, starting off with the power design and cooling solution.
Power delivery has been vastly revamped compared to the older Founders Edition cards for better heat output and efficiency. The VRM for example, is a 7+2 phase power design with 7 phases for the GPU power and 2 phases for the memory. The MOSFETs also saw improvements from a 1xdualFET to 2xDualFET design. Do note that there are only 11 GDDR5X SGRAMs soldered, and there’s a solder path for a DVI port behind the dual DisplayPorts. This card features a 6+8 pin PEG power connector configuration able to input a maximum of 300W together with the PCI-E slot power.
The cooling solution is still powered by an impeller fan, blowing air to the vapor chamber connected to a dense array of heat fins. Since there’s no DVI ports in place, there’s less resistance and more breathing room for the ventilation to work.
THE TEST SYSTEM AND BENCHMARKS:
Our test system is based on a Intel Core i5 6600K gaming system, which should be in-line with the latest mainstream gaming components. The operating system used is a 64-bit Windows 10 Pro Edition with power options set to high performance. Third party background applications are disabled unless they are absolutely required.
|TEST SYSTEM SPECIFICATIONS|
|PROCESSOR||Intel Core i5 6600K|
|CPU COOLER||Cryorig C1 Top Flow|
|MEMORY KIT||Crucial Ballistix Tactical @ 2666MHz 4x4GB Kit|
|GRAPHICS CARD||Nvidia GeForce GTX 1080 Ti Founders Edition 11GB|
|INTERNAL STORAGE||Crucial MX200 250GB|
|POWER SUPPLY||CORSAIR RM850X 850W|
|DISPLAY||27″ DELL U2715H + LG 43UF680T 4K UHD TV|
|OPERATING SYSTEM||Microsoft Windows 10 Pro|
The games and benchmark tools should include 720p, 1080p, 1440p, and 2160p results if necessary to the hardware being tested. They are outlined below for your reference.
- 3DMARK 2013 – Firestrike Presets.
- Luxmark – OpenCL benchmark.
- Aida64 Extreme Edition – GPGPU benchmark.
- Overwatch – Settings are set to Ultra with a Render Scale set to 100%.
- Battlefield 4 – Settings are set to Ultra.
- Starcraft 2 – Settings are set to Extreme.
- DOTA 2 – Settings are manually set to maximum with DirectX 10 only mode and FPS cap disabled.
- DIRT: Rally – Settings are set to Ultra.
- Project CARS – Settings are manually set to maximum with anisotropic filtering at x16.
- The Elder Scrolls V: Skyrim – Settings are set to Ultra with V-SYNC turned off.
- The Witcher 3: Wild Hunt – Settings are set to Ultra with Nvidia HairWorks turned off.
FRAMES PER SECOND 101:
FPS is the general unit of which we measure a graphics card’s performance. Since higher FPS means that there are more frames rendered each second – the higher value, the smoother the gameplay should be. Check out the difference of 30 and 60 FPS on this LINK.
FPS values are recorded by FRAPS in-game using a linear route or gameplay for consistency. The results includes the Average FPS and 1% Minimum FPS. The 1% Minimum FPS represents the negative spikes or drops in-game sufficient enough for a gamer to notice. If the game doesn’t support FRAPS, we will use the in-game benchmark tool available.
TEMPERATURE POWER AND NOISE:
The graphics card thermal figures are taken during a 15 minute idle, and a 15 minute load via AIDA64 Extreme Edition’s GPU Stress Test. The ambient temperature is checked at 27°C (±1°C), with values extracted via HWiNFO and GPU-Z.
The total system power consumption are taken during a 15 minute idle, and a 15 minute load via AIDA64 Extreme Edition’s GPU Stress Test. The measurements are taken by a watt-meter.
The graphics card noise levels are taken during a 15 minute idle, and a 15 minute load via AIDA64 Extreme Edition’s GPU Stress Test. The ambient sound level is 34dBA (±1dBA), with measurements taken by a sound level meter situated exactly 12 inches away from the card.
We also checked how the graphics card behaves on games with the frame rate locked. This should help you check out the temperature, power, and noise of the card at certain display refresh rate configurations.
|Overwatch||Temperature||Total System Power||Noise|
|300 FPS (Maximum Allowed)||83°C||303.0W||43.7dBA|
SYNTHETIC OPENCL AND GPGPU PERFORMANCE:
3DMark 2013 is a suite of benchmarks catered for the gamers and enthusiasts. The total system scores from each presets will be used.
LuxMark is a OpenCL cross-platform benchmark tool and has become, over past years, one of the most used OpenCL benchmarks.
The Aida64 Extreme Edition GPGPU Benchmark is designed to measure GPGPU computing performance via different workloads. Every benchmark methods are designed to work on up to 16 GPUs, including AMD, Intel and nVIDIA GPUs, in any combination.
Overwatch is Blizzard Entertainment’s critically acclaimed First Person Shooter, running on a custom game engine utilizing the DirectX 11 API. Settings are set to Ultra with a Render Scale set to 100%.
Battlefield 4 is Electronic Art’s popular First Person Shooter, running on the Frostbite 3 game engine developed by DICE utilizing the DirectX 11 and the Mantle API. Settings are set to Ultra.
Starcraft 2 is Blizzard Entertainment’s critically acclaimed RTS game, running on the HAVOC game engine utilizing the DirectX 9 API. Settings are set to Extreme.
DOTA 2 is VALVE’s popular F2P MOBA game, running on the Source game engine utilizing the multi-API support. Settings are set to maximum with DirectX 10 only mode and FPS cap disabled.
DIRT: Rally is CODEMASTER’s take on the rally racing game genre. It utilizes the EGO game engine with support for the DirectX 11 API. Settings are set to Ultra.
Project CARS is Slightly Mad Studios’ latest simulation racing game, running on the Madness game engine utilizing DirectX 11 support. Settings are set to maximum.
The Elder Scrolls V: Skyrim is Bethesda’s critically acclaimed Action Adventure RPG. Skyrim utilizes the Creation Engine with DirectX 10 support. Settings are set to Ultra with V-SYNC turned off.
The Witcher 3: Wild Hunt is CDProjekt RED’s award winning Action Adventure RPG. The Witcher 3 utilizes the REDengine 3 with DirectX 11 support. Settings are set to Ultra.
NVIDIA DID IT AGAIN:
First off let us check out the gaming performance of this card under 1080P. At this resolution, performance scaling between games vs the GTX 1080 is alright but not ideal. We’re seeing little to no gains up to 28% gains based on our tests. We’d also like to point out that Project CARS and DOTA2 are limited to 130 and 170 FPS respectively. We are playing with limited time, and we did our best to retest these units only reaching a conclusion that both game’s battle recorder are locked at those said values.
Heading over at the 1440P territory, we could see that this card shines the most at higher resolutions which is just proper for this card’s caliber. Gains are seen from 1.8% to 35.3% which is within the expected performance scaling versus the GTX 1080 according to Nvidia. Games with high resolution textures, AA features, and various post processing effects should sate this card’s appetite.
At 2160P this card really slammed the money into the table allowing itself to gain a huge upper hand over the GTX 1080. Maximum gains were seen from majority of games regardless of frame capping. Suffice to say the GTX 1080 TI is a monstrous card – able to reach what the GTX 1080 cannot. The GTX 1080 TI is definitely a substantial upgrade from the GTX 1080, with plenty of horse power left untapped. It’s practically beyond Titan X level when overclocked but we’ll save that one up for later.
As for the card itself, the Founders Edition card is rather limited thermally. For an instance, when it hits 84c, the Founders Edition starts to dial down the core clock to a thermally acceptable level. Pushing the fan speed over 70% is one solution, and you’d be surprised that a simple fan speed adjustment could change this card’s performance in an instant. With that said, we cannot wait to see the models with better cooling solutions from Nvidia’s AIB partners. The GPU Boost 3.0 sure is one feature that we couldn’t agree to hate.
It is not every day that we get to test a monster – yet here we are with the Nvidia GeForce GTX 1080 Ti Founders Edition. I cannot simply ask you to buy this card really, but if you have a $700 ready to be spend for a graphics card alone then why the hell not. Especially if you’re a guy who just spent the same amount of cash on a GTX 980 Ti that has been released a couple of years ago. GTX 1080 owners are also welcome to upgrade, notably if you really require the power of this card at 1440P and beyond. If your current card can’t drive titles over the said resolution at 144Hz and above, you have a greater chance with the GTX 1080 TI.
With all of those things said, the GTX 1080 TI is now officially the fastest single GPU gaming graphics card, and there’s basically nothing on its way to stop it. Your turn, AMD.