Nvidia GeForce RTX 2080 Ti Founders Edition Review: A Titan V Killer

GeForce RTX 2080 Ti is fast. Nobody will argue against that. What bothers gamers most is its price tag. The Founders Edition board we’re reviewing today costs $1200—71% higher than the flagship GeForce GTX 1080 Ti Nvidia launched back in 2017. Perhaps disappointingly, then, GeForce RTX 2080 Ti averages about 26% faster than its predecessor across our suite of 13 games at 4K.

But what if we told you it also beats Nvidia’s Titan V? In Battlefield 1, GeForce RTX 2080 Ti is 20% faster. In Forza Motorsport 7, the 2080 Ti enjoys a 22% advantage. Based on the two cards’ specifications, we weren’t expecting such a dramatic finish. Although the $3000 GV100-based Titan V is made for deep learning and not gaming, those results sure put GeForce RTX 2080 Ti’s $1200 price into context.

Even more significant for the growing group of enthusiasts with 4K monitors, GeForce RTX 2080 Ti gives you playable performance at 3840x2160 without sacrificing detail settings. Yes, we turned off taxing anti-aliasing modes to test at 4K. However, our benchmarks reflect high-res, high-quality PC gaming above 60 FPS in every single title.

This is a moment we’ve anticipated for years. GeForce GTX 980 Ti was playable at 4K if you dialed down the details. Although GeForce GTX 1080 got us a little closer to gaming nirvana, it clearly wasn’t the stunner needed for smooth performance at max-quality 4K. Then, GeForce GTX 1080 Ti tempted us to finally declare a 4K champion. But we’re glad we held off. GeForce RTX 2080 Ti finally makes those Ultra settings viable at 3840x2160.

TU102: The Makings of a Gaming Beast

How does GeForce RTX 2080 Ti achieve this? Well, if you missed our comprehensive analysis of the card’s inner workings, check out Nvidia’s Turing Architecture Explored: Inside the GeForce RTX 2080. To summarize, though, today’s subject is based on TU102, a 754-square-millimeter GPU composed of 18.6 billion transistors fabricated on TSMC’s 12nm FinFET manufacturing process. It's loaded down with higher quantities of rendering resources that operate more efficiently than anything we've ever tested.

A complete TU102 processor comprises six Graphics Processing Clusters (GPCs) made up of a Raster Engine and six Texture Processing Clusters (TPCs). Each TPC is composed of one PolyMorph Engine (fixed-function geometry pipeline) and two Streaming Multiprocessors (SMs). At the SM level, we find 64 CUDA cores, eight Tensor cores, one RT core, four texture units, 16 load/store units, 256KB of register file space, four L0 instruction caches, and a 96KB configurable L1 cache/shared memory structure.

Multiply all of that out and you get a GPU with 72 SMs, 4608 CUDA cores, 576 Tensor cores, 72 RT cores, 288 texture units, and 36 PolyMorph engines. Those resources are fed by 12 32-bit GDDR6 memory controllers, each attached to an eight-ROP cluster and 512KB of L2 cache yielding an aggregate 384-bit memory bus, 96 ROPs, and a 6MB L2 cache. Each SM does contain a pair of FP64-capable CUDA cores as well, yielding a double-precision rate that’s 1/32 of TU102’s FP32 performance.


GeForce RTX 2080 Ti FE
GeForce RTX 2080 FE
GeForce GTX 1080 Ti FE
GeForce GTX 1080 FE
Architecture (GPU)
Turing (TU102)
Turing (TU104)Pascal (GP102)
Pascal (GP104)
CUDA Cores
4352
29443584
2560
Peak FP32 Compute
14.2 TFLOPS
10.6 TFLOPS11.3 TFLOPS
8.9 TFLOPS
Tensor Cores
544
368N/A
N/A
RT Cores
68
46N/A
N/A
Texture Units
272
184224
160
Base Clock Rate
1350 MHz
1515 MHz1480 MHz
1607 MHz
GPU Boost Rate
1635 MHz
1800 MHz1582 MHz
1733 MHz
Memory Capacity
11GB GDDR6
8GB GDDR611GB GDDR5X
8GB GDDR5X
Memory Bus
352-bit
256-bit352-bit
256-bit
Memory Bandwidth
616 GB/s
448 GB/s484 GB/s
320 GB/s
ROPs
88
6488
64
L2 Cache
5.5MB
4MB2.75MB
2MB
TDP
260W
225W250W
180W
Transistor Count
18.6 billion
13.6 billion12 billion
7.2 billion
Die Size
754 mm²545 mm²471 mm²314 mm²
SLI Support
Yes (x8 NVLink, x2)
Yes (x8 NVLink)Yes (MIO)
Yes (MIO)

Putting It All Together: GeForce RTX 2080 Ti

The TU102 found on GeForce RTX 2080 Ti isn’t a complete processor, though. Whether Nvidia wanted to leave room for a Titan-class model or found yields of fully-functional GPUs unsatisfactory above a certain bin, the RTX 2080 Ti has two of its TPCs disabled, leaving the card with 4352 CUDA cores, 544 Tensor cores, 68 RT cores, 544 texture units, and 34 PolyMorph engines.

One of TU102’s 32-bit memory controllers is also turned off, creating an aggregate 352-bit bus that moves data to 88 ROPs and 5.5MB of L2 cache. Nvidia matches its strategically-hobbled GPU to Micron’s MT61K256M32JE-14:A modules. Eleven of these populate the RTX 2080 Ti’s PCB, leaving one emplacement vacant. Nevertheless, theoretical peak bandwidth rises sharply compared to the previous generation cards due to GDDR6’s higher data rate. At 14 Gb/s on a 352-bit interface, you’re looking at 616 GB/s. In comparison, GDDR5X at 11 Gb/s held GeForce GTX 1080 Ti to 484 GB/s.

On the Founders Edition card, a base core frequency of 1350 MHz jumps all the way up to a typical GPU Boost rate of 1635 MHz, so long as GeForce RTX 2080 Ti is running cool enough. And because Nvidia cites peak compute performance using GPU Boost numbers, its top-end model achieves up to 14.2 TFLOPS of single-precision math.

That frequency is overclocked relative to Nvidia’s reference specification. As a result, the Founders Edition model has a slightly higher 260W TDP rating. A PCIe slot, an eight-pin power connector, and a six-pin power connector would theoretically be capable of servicing such a limit. However, Nvidia instead uses two supplementary eight-pin connectors, giving its GeForce RTX 2080 Ti Founders Edition potential headroom for overclocking.

Moving back from the PCIe connectors to GeForce RTX 2080 Ti's power supply, we count three phases for the GDDR6 memory and a corresponding PWM controller up front. A total of 13 phases remain, according to Nvidia. But that can't be right, can it? Even with six phase-doubled circuits, there would be one phase leftover to act as a load balancer. The numbers just don't add up.

In reality, it appears that five phases are fed by the eight-pin connectors and doubled. With two control loops per phase, 5*2=10 voltage regulation circuits. The remaining three phases to the left of the GPU are fed by the motherboard's PCIe slot and not doubled. That gives us Nvidia's lucky number 13 (along with a smart load distribution scheme). Of course, implementing all of this well requires the right components...

GPU Power Supply

Front and center in this design is uPI's uP9512 eight-phase buck controller specifically designed to support next-gen GPUs. Per uPI, "the uP9512 provides programmable output voltage and active voltage positioning functions to adjust the output voltage as a function of the load current, so it is optimally positioned for a load current transient."

The uP9512 supports Nvidia's Open Voltage Regulator Type 4i+ technology with PWMVID. This input is buffered and filtered to produce a very accurate reference voltage. The output voltage is then precisely controlled to the reference input. An integrated SMBus interface offers enough flexibility to optimize performance and efficiency, while also facilitating communication with the appropriate software.

All 13 voltage regulation circuits are equipped with an ON Semiconductor FDMF3170 Smart Power Stage module with integrated PowerTrench MOSFETs and driver ICs.

As usual, the coils rely on encapsulated ferrite cores, but this time they are rectangular to make room for the voltage regulator circuits.

Memory Power Supply

Micron's MT61K256M32JE-14:A memory ICs are powered by three phases coming from a second uP9512. The same FDMF3170 Smart Power Stage modules crop up yet again. The 470mH coils offer greater inductance than the ones found on the GPU power phases, but they're completely identical in terms of physical dimensions.

The input filtering takes place via three 1μH coils, whereby each of the three connection lines has a matching shunt. This is a very low resistance to which voltage drop is measured in parallel and passed on to the telemetry. Through these circuits, Nvidia is able to limit board power in a fairly precise way.

Unfortunately for the folks who like a bit of redundancy, this card only comes equipped with one BIOS.

How We Tested GeForce RTX 2080 Ti

Nvidia’s latest and greatest will no doubt be found in one of the many high-end platforms now available from AMD and Intel. Our graphics station still employs an MSI Z170 Gaming M7 motherboard with an Intel Core i7-7700K CPU at 4.2 GHz, though. The processor is complemented by G.Skill’s F4-3000C15Q-16GRR memory kit. Crucial’s MX200 SSD remains, joined by a 1.4TB Intel DC P3700 loaded down with games.

As far as competition goes, the GeForce RTX 2080 Ti is rivaled only by the $3000 Titan V. We add that card to our test pool this time around, along with GeForce GTX 1080 Ti, Titan X, GeForce GTX 1080, GeForce GTX 1070 Ti, and GeForce GTX 1070 from Nvidia. AMD is represented by the Radeon RX Vega 64 and 56. All cards are either Founders Edition or reference models. We do have some partner boards in-house from both Nvidia and AMD, and plan to use those for third-party reviews.

Our benchmark selection now includes Ashes of the Singularity: Escalation, Battlefield 1, Civilization VI, Destiny 2,Doom, Far Cry 5,Forza Motorsport 7, Grand Theft Auto V, Metro: Last Light Redux, Rise of the Tomb Raider, Tom Clancy’s The Division, Tom Clancy’s Ghost Recon Wildlands, The Witcher 3 and World of Warcraft: Battle for Azeroth. We’re working on adding Monster Hunter: World, Shadow of the Tomb Raider, Wolfenstein II, and a couple of others, but had to scrap those plans due to very limited time with Nvidia’s final driver for its Turing-based cards.

The testing methodology we're using comes from PresentMon: Performance In DirectX, OpenGL, And Vulkan. In short, all of these games are evaluated using a combination of OCAT and our own in-house GUI for PresentMon, with logging via AIDA64.

All of the numbers you see in today’s piece are fresh, using updated drivers. For Nvidia, we’re using build 411.51 for GeForce RTX 2080 Ti and 2080. The other cards were tested with build 398.82. Titan V’s results were spot-checked with 411.51 to ensure performance didn’t change. AMD’s cards utilize Crimson Adrenalin Edition 18.8.1, which was the latest at test time.

MORE: Best Graphics Cards

MORE: Desktop GPU Performance Hierarchy Table

MORE: All Graphics Content