The Titan RTX launch was decidedly unceremonious. Members of the tech press knew that the card was coming but didn’t receive one to test. Nvidia undoubtedly knew its message would be obscured by comparisons drawn between Titan RTX and the other TU102-based card, GeForce RTX 2080 Ti, in games. Based on a complete TU102 processor, Titan RTX was bound to be faster than the GeForce in all benchmarks, regardless of discipline. However, its eye-watering $2,500 price tag would be difficult to justify for entertainment alone

aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL0QvODIwMzgxL29yaWdpbmFsL3RpdGFuLXJ0eC10aHJlZS1xdWFydGVy.jpg

But even as gamers ponder the effect of an extra four Streaming Multiprocessors on their frame rates, we all know that Titan RTX wasn’t intended for those folks. Of course, we’re still going to run it through our suite of game benchmarks. However, Nvidia says this card was designed for “AI researchers, deep learning developers, data scientists, content creators, and artists.”

Participants in those segments with especially deep pockets often spring for Tesla-based platforms or Quadro cards sporting certified drivers. Those aren’t always necessary, though. As a result, “cost-sensitive” professionals working in smaller shops find themselves somewhere in the middle, needing more than a GeForce but unable to spend $6,300 on an equivalent Quadro RTX 6000.

Out of necessity, our test suite is expanding to include professional visualization and deep learning metrics, in addition to the power consumption analysis we like to perform. Get ready for a three-way face-off between Titan RTX, Titan V, and Titan Xp (with a bit of GeForce RTX 2080 Ti sprinkled in).

PROS

• 24GB of memory is ideal for large professional and deep learning workloads
• Improved Tensor cores benefit inferencing performance specifically
• Excellent 4K gaming frame rates
• NVLink support (which Titan V lacks)
• Attractive design

CONS

• $2,500 price limits appeal to professionals with deep pockets
• Axial fan design exhausts (a lot of) heat into your case
• Poor FP64 capabilities compared to Titan V

VERDICT

Titan RTX is the right card for the right customer. It's a no-brainer if you're working with large geometry models, training neural networks with large batch sizes, or inferencing trained networks using frameworks like TensorRT with support for the hardware's features. Gaming on Titan RTX doesn't make as much sense when you could have two GeForce RTX 2080 Tis for a similar price.

Meet Titan RTX: It Starts With A Complete TU102

The Tom’s Hardware audience should be well-acquainted with Nvidia’s TU102 GPU by now: it’s the engine at the heart of GeForce RTX 2080 Ti, composed of 18.6 billion transistors, and measuring 754 square millimeters.

As it appears in the 2080 Ti, though, TU102 features 68 active Streaming Multiprocessors. Four of the chip’s 72 are turned off. One of its 32-bit memory controllers is also disabled, taking eight ROPs and 512KB of L2 cache with it.

Titan RTX is based on the same processor, but with every block active. That means the card boasts a GPU with 72 SMs, 4,608 CUDA cores, 576 Tensor cores, 72 RT cores, 288 texture units, and 36 PolyMorph engines.

Not only does Titan RTX sport more CUDA cores than GeForce RTX 2080 Ti, it also offers a higher GPU Boost clock rating (1,770 MHz vs. 1,635 MHz). As such, its peak single-precision rate increases to 16.3 TFLOPS.

2aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL1ovODIwNDAzL29yaWdpbmFsL0dQR1BVLUFyaXRobWV0aWMucG5n.jpg

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, or 0.51 TFLOPS. This is one area where Titan RTX loses big to its predecessor. Titan V’s GV100 processor is better in the HPC space thanks to 6.9 TFLOPS peak FP64 performance (half of its single-precision rate). A quick run through SiSoftware’s Sandra GPGPU Arithmetic benchmark confirms Titan V’s strength, along with the mixed-precision support inherent to Turing and Volta, which Pascal lacks.

The GPU’s GPCs 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. At the same 14 Gb/s data rate, one extra memory emplacement buys Titan RTX about 9% more memory bandwidth than GeForce RTX 2080 Ti.

FireShot Capture 031 - Nvidia Titan RTX Review_ Gaming, Trai_ - https___www.tomshardware.com_rev.jpg

Whereas GeForce RTX 2080 Ti Founders Edition utilizes Micron’s MT61K256M32JE-14:A modules, the company doesn’t have any 16Gb ICs in its parts catalog. Samsung, on the other hand, does offer a higher-density K4ZAF325BM-HC14 module with a 14 Gb/s data rate. Twelve of them give Titan RTX its 24GB capacity and 672 GB/s peak throughput.

33aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL0MvODIwMzgwL29yaWdpbmFsL3RpdGFuLXJ0eC1wY2ItZnJvbnQuan.jpg

Lots of extra memory, a GPU with more active resources, and faster clock rates necessitate a higher thermal design power rating. Whereas GeForce RTX 2080 Ti Founders Edition is specified at 260W, Titan RTX is a 280W card. That 20W increase is no problem at all for the pair of eight-pin auxiliary power connectors found along the top edge, nor is a challenge for Nvidia’s power supply and thermal solution, both of which appear identical to its GeForce RTX 2080 Ti.

Like the 2080 Ti Founders Edition, we count three phases for Titan RTX’s GDDR6 memory and a corresponding PWM controller up front. A total of 13 phases remain. 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...

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.

Samsung’s K4ZAF325BM-HC14 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.

55aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL0svODIwMzg4L29yaWdpbmFsL3RpdGFuLXJ0eC1mcm9udC1jb29sZX.jpg

Under the hood, Titan RTX’s thermal solution is also the same as what we found on GeForce RTX 2080 Ti. A full-length vapor chamber covers the PCB and is topped with an aluminum fin stack. A shroud over the heat sink houses two 8.5cm axial fans with 13 blades each. These fans blow through the fins and exhaust waste heat out the card’s top and bottom edges. Although we don’t necessarily like that Nvidia recirculates hot air with its Turing-based reference coolers, their performance is admittedly superior to older blower-style configurations.

66aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8xLzUvODIwNDA5L29yaWdpbmFsL3RpdGFuLXJ0eC1mcm9udC5qcGc=.jpg

Titan RTX’s backplate isn't just for looks. Nvidia incorporates the plate into its cooling concept by sandwiching thermal pads between the metal and PCB, behind where you'd find memory modules up front. The company probably could have done without the pad under TU102, though: our measurements showed no difference when we removed it.

It should come as no surprise that the same PCB, same components, same cooler, and same dimensions yield a card weighing the same as GeForce RTX 2080 Ti, too. Our Titan RTX registers 2lb. 14.6 oz. (1.322kg) on a scale. Even the sticker on the back of Titan RTX says GeForce RTX 2080 Ti. Nvidia differentiates the card aesthetically with a matte gold (rather than silver) finish, with polished gold accents around the fans and shroud. A shiny gold Titan logo on the top edge glows white under power.

How We Tested Titan RTX

Benchmarking Titan RTX is a bit more complex than a typical gaming graphics card. Because we ran professional visualization, deep learning, and gaming workloads, we ended up utilizing multiple machines to satisfy the requirements of those disparate applications.

To facilitate deep learning testing, we set up a Core i7-8086K (6C/12T) CPU on a Gigabyte Z370 Aorus Ultra Gaming motherboard. All four of its memory slots were populated with 16GB Corsair Vengeance LPX modules at DDR4-2400, giving us 64GB total. Then, we loaded Ubuntu 18.04 LTS onto a 1.6TB Intel SSD DC P3700. Titan RTX is compared to Titan V, Titan Xp, and GeForce RTX 2080 Ti in these benchmarks.

For professional visualization, we set up a Core i7-8700K (6C/12T) CPU on an MSI Z370 Gaming Pro Carbon AC motherboard. Again, all four of its memory slots were populated with 16GB Corsair Vengeance LPX modules at DDR4-2400, giving us 64GB total. Then, we loaded Windows 10 Professional onto a 1.2TB Intel SSD 750-series drive. Incidentally, this is the same system running our Powenetics software for power consumption measurement. Titan RTX is compared to Titan V, Titan Xp, and GeForce RTX 2080 Ti in these benchmarks.

Finally, we collect gaming results on the same Core i7-7700K used in previous reviews. It populates an MSI Z170 Gaming M7 motherboard, which also hosts G.Skill’s F4-3000C15Q-16GRR memory kit. Crucial’s MX200 SSD remains, joined by a 1.6TB Intel DC P3700 loaded down with games. Titan RTX is compared to Gigabyte’s Aorus GeForce RTX 2080 Ti Xtreme 11G, the GeForce RTX 2080 Ti Founders Edition, Titan V, GeForce RTX 2080, GeForce RTX 2080, GeForce GTX 1080 Ti, Titan X, GeForce GTX 1070 Ti, GeForce GTX 1070, Radeon RX Vega 64, and Radeon RX Vega 56 in these benchmarks.

All of those cards are tested at 2560x1440 and 3840x2160 in Ashes of the Singularity: Escalation, Battlefield V, Destiny 2, 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 Wolfenstein II: The New Colossus.

We invited AMD to submit a more compute-oriented product for our deep learning and workstation testing. However, company representatives didn’t respond after initially looking into the request. Fortunately, we should have Radeon VII to test very soon.

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

As we generate new data, we’re using the latest drivers. For Linux, that meant using 415.18 for Titan RTX and Titan V, and then 410.93 for Titan Xp. Under Windows 10, we went with Nvidia’s 417.26 press driver for Titan RTX, 417.22 for Gigabyte’s card, 416.33 For GeForce RTX 2070, and 411.51 for 2080/2080 Ti FE. Older Pascal-based boards 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 (except for the Battlefield V and Wolfenstein tests, which are tested with Adrenalin Edition 18.11.2).

Special thanks to Noctua for sending over a batch of NH-D15S heat sinks with fans. These top all three systems, giving us consistency in effective, quiet cooling.

Performance Results: Pro Visualization

ArionBench

Madrid-based RandomControl offers the Arion Render physically-based path tracing render engine and ArionFX, composed of HDR image processing algorithms. ArionBench is meant as a proxy for the former, measuring GPU and CPU performance through a light simulation in a 3D scene.

The benchmark package includes executables for testing available GPUs, CPUs, and a hybrid combination of the two.

77aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL1gvODIwNDAxL29yaWdpbmFsL0FyaW9uQmVuY2gucG5n.jpg

This CUDA-accelerated workload runs best on Titan RTX, followed by GeForce RTX 2080 Ti. Titan V trails the gaming card by about 100 points in the hardware-only test, while Titan Xp finishes far behind the more modern boards.

LuxMark v.3.1

The latest version of LuxMark is based on an updated LuxRender 1.5 render engine, which specifically incorporates OpenCL optimizations that invalidate comparisons to previous versions of the benchmark.

We tested all three scenes available in the 64-bit benchmark: LuxBall HDR (with 217,000 triangles), Neumann TLM-102 SE (with 1,769,000 triangles), and Hotel Lobby, with 4,973,000 triangles).

88aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL00vODIwMzkwL29yaWdpbmFsL0x1eE1hcmsucG5n.jpg

Turing and Volta GPUs trade blows depending on the scene you look at. Titan V scores a win in LuxBall and Hotel Lobby, while the two TU102-based boards score higher in Neumann TLM-102 SE. The main takeaway, however, seems to be that Titan Xp is limited to a fraction of the performance achieved by the newer cards.

Cinema4D

ProRender is another physically-based GPU render engine. Unlike Arion Render, however, it utilizes OpenCL. It’s also biased, meaning the renderer’s output is based on estimations rather than pixel-by-pixel calculations. Arion Render is unbiased, performing calculations on every pixel and in turn taking longer.

99aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL1QvODIwMzk3L29yaWdpbmFsL1Byb1JlbmRlci5wbmc=.jpg

In both of our test scenes, Titan RTX is faster than its Nvidia-sourced competition.

On-board memory doesn’t seem to be a factor, since GeForce RTX 2080 Ti easily beats Titan Xp with 1GB less capacity. It’s more likely that Turing/Volta’s compute performance, increased number of schedulers, on-die SRAM advantage, and higher memory bandwidth convey big gains over the older Pascal architecture.

OctaneRender

The latest version of OTOY’s OctaneRender incorporates support for out-of-core geometry, meaning meshes and textures can be stored in system memory while the unbiased GPU renderer works at interactive speeds.

100aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL08vODIwMzkyL29yaWdpbmFsL09jdGFuZS1SZW5kZXIucG5n.jpg

One of Titan RTX’s big selling points is its 24GB of GDDR6. Thus far, our benchmarks haven’t shown a need for that much on-board memory. However, the first test we ran in OctaneRender repeatedly crashed on Titan V and Titan Xp due to running out of memory. A simpler scene allowed us to create a valid comparison, but not before we got our first taste of capacity envy.

Just because we completed runs on the competing cards didn’t mean their outcomes made much sense, though. It’s plausible that GeForce RTX 2080 Ti’s 11GB put it at a disadvantage to Titan Xp’s 12GB, tipping the scale in favor of Pascal. However, Titan V shouldn’t have suffered such a high render time (and low memory utilization number) with just as much RAM. In bouncing ideas back and forth with Nvidia, we could only hypothesize an issue with Titan V’s HBM2 memory subsystem not playing nice with OctaneRender.

SPECviewperf 13

The most recent version of SPECviewperf employs traces from Autodesk 3ds Max, Dassault Systemes Catia, PTC Creo, Autodesk Maya, Autodesk Showcase, Siemens NX, and Dassault Systemes SolidWorks. Two additional tests, Energy and Medical, aren’t based on a specific application, but rather on datasets typical of those industries.

101aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL1cvODIwNDAwL29yaWdpbmFsL1NQRUN2aWV3cGVyZi0xOTAweDEwN.jpg

In some workloads, Nvidia’s DirectX driver allows GeForce RTX 2080 Ti to match or even exceed the performance of Titan V. But Catia and NX, specifically, respond well to the professional driver optimizations that benefit Titan cards. The GeForce even loses to the older Titan Xp in those workloads.

Across the board, Titan RTX beats the still-formidable Titan V.

102aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL1kvODIwNDAyL29yaWdpbmFsL1NQRUN2aWV3cGVyZi0yODAweDIxM.jpg

Titan V scores a slight win in the Energy tests, but again succumbs to Titan RTX everywhere else once we step the resolution up to 3800x2120.

Performance Results: Deep Learning

The introduction of Turing saw Nvidia’s Tensor cores make their way from the data center-focused Volta architecture to a more general-purpose design with its beginnings in gaming PCs. However, the company’s quick follow up with Turing-based Quadro cards and the inferencing-oriented Tesla T4 GPU made it pretty clear that DLSS wasn’t the only purpose of those cores.

Now that we have Titan RTX—a card with lots of memory—it’s possible to leverage TU102’s training and inferencing capabilities in a non-gaming context.

Before we get to the benchmarks, it’s important to understand how Turing, Volta, and Pascal size up in theory.

Titan Xp’s GP102 processor is more like GP104 than the larger GP100 in that it supports general-purpose IEEE 754 FP16 arithmetic at a fraction of its FP32 rate (1/64), rather than 2x. GP102 does not support mixed precision (FP16 inputs with FP32 accumulates) for training, though. This becomes an important distinction in our Titan Xp benchmarks.

Volta greatly improved on Pascal’s ability to accelerate deep learning workloads with Nvidia’s first-generation Tensor cores. These specialized cores performed fused multiply adds exclusively, multiplying a pair of FP16 matrices and adding their result to an FP16 or FP32 matrix. Titan V’s Tensor performance can be as high as 119.2 TFLOPS for FP16 inputs with FP32 accumulates, making it an adept option for training neural networks.

Although TU102 hosts fewer Tensor cores than Titan V’s GV100, a higher GPU Boost clock rate facilitates a theoretically better 130.5 TFLOPS of Tensor performance on Titan RTX. GeForce RTX 2080 Ti Founders Edition should be able to muster 113.8 TFLOPS. However, Nvidia artificially limits the desktop card’s FP16 with FP32 accumulates to half-rate.

Training Performance

Our first set of benchmarks utilizes Nvidia’s TensorFlow Docker container to train a ResNet-50 convolutional neural network using ImageNet. We separately charted training performance using FP32 and FP16 (mixed) precision.

103aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL1UvODIwMzk4L29yaWdpbmFsL1Jlc05ldDUwLUZQMTYucG5n.jpg

104aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL1YvODIwMzk5L29yaWdpbmFsL1Jlc05ldDUwLUZQMzIucG5n.jpg

The numbers in each chart’s legend represent batch sizes. In brief, batch size determines the number of input images fed to the network concurrently. The larger the batch, the faster you’re able to get through all of ImageNet’s 14+ million images, given ample GPU performance, GPU memory, and system memory.

In both charts, Titan RTX can handle larger batches than the other cards thanks to its 24GB of GDDR6. This conveys a sizeable advantage over cards with less on-board memory. Of course, Titan V remains a formidable graphics card, and it’s able to trade blows with Titan RTX using FP16 and FP32.

GeForce RTX 2080 Ti’s half-rate mixed-precision mode causes it to shed quite a bit of performance compared to Titan RTX. Why isn’t the difference greater? The FP32 accumulate operation is only a small part of the training computation. Most of the matrix multiplication pipeline is the same on Titan RTX and GeForce RTX 2080 Ti, creating a closer contest than the theoretical specs would suggest. Switching to FP32 mode erases some of the discrepancy between Nvidia’s two TU102-based boards.

It’s also worth mentioning the mixed precision results for Titan Xp. Remember, GP102 doesn’t support FP16 inputs with FP32 accumulates, so it operates in native FP16 mode. The benchmarks do run successfully. But their accuracy is inferior to what Volta and Turing enable. To compare generational improvements at a like level of accuracy, you’d need to pit Titan RTX with mixed precision (644 images/sec) against Titan Xp in FP32 mode (233 images/sec).

Next, we train the Google Neural Machine Translation recurrent neural network using the OpenSeq2Seq toolkit, again in TensorFlow.

105aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL1EvODIwMzk0L29yaWdpbmFsL09wZW5TZXEyU2VxLUZQMTYucG5n.jpg

106aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL04vODIwMzkxL29yaWdpbmFsL09wZW5TZXEyU2VxLUZQMzIucG5n.jpg

Specifying FP32 precision again lets Titan RTX show off the advantage of 24GB of memory. Titan V and Xp started displaying out-of-memory errors after successful runs with batch sizes of 32.

Enabling FP16 precision in the dataset’s training parameter script allows the 12GB cards to accommodate larger batches. However, they’re both overwhelmed by Titan RTX as we push 384-image batches through its TU102.

Inferencing Performance

Before we isolate the performance of Turing’s new INT8 mode, we took a pass at inferencing on a trained ResNet-50 model with TensorFlow.

107aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL1IvODIwMzk1L29yaWdpbmFsL0luZmVyZW5jaW5nVGVuc29yRmxvd.jpg

Across the board, Titan RTX outperforms Titan V (as does GeForce RTX 2080 Ti).

Nvidia recommends inferencing in TensorRT, though, which supports Turing GPUs, CUDA 10, and the Ubuntu 18.04 environment we’re testing under. Our first set of results inference a GoogleNet model pre-trained in Caffe.

108aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8xLzEvODIwNDA1L29yaWdpbmFsL0luZmVyZW5jaW5nR29vZ2xlTmV0L.jpg

109aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8xLzAvODIwNDA0L29yaWdpbmFsL0luZmVyZW5jaW5nR29vZ2xlTmV0L.jpg

110aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8xLzIvODIwNDA2L29yaWdpbmFsL0luZmVyZW5jaW5nR29vZ2xlTmV0L.jpg

With the FP32 numbers serving as a baseline, Titan RTX’s speed-ups thanks to FP16 and INT8 modes are significant. GeForce RTX 2080 Ti benefits similarly.

Titan V’s INT8 rate is ½ of its peak FP16 throughput, so although the card does see improvements versus FP16 mode, they’re not as pronounced.

Titan Xp’s 48.8 TOPS of INT8 performance prove useful in inferencing workloads. An FP16 rate that’s 1/64 of FP32 throughput means we’re not surprised to see FP16 precision only barely faster than the FP32 result.

111aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8xLzMvODIwNDA3L29yaWdpbmFsL0luZmVyZW5jaW5nUmVzTmV0NTAtR.jpg

112aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8xLzQvODIwNDA4L29yaWdpbmFsL0luZmVyZW5jaW5nUmVzTmV0NTAtR.jpg

113aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL1AvODIwMzkzL29yaWdpbmFsL0luZmVyZW5jaW5nUmVzTmV0NTAtS.jpg

Inferencing a ResNet-50 model trained in Caffe demonstrates similar trends. That is, Titan V doesn’t scale as well using INT8 precision, while Titan Xp enjoys a big speed-up from INT8 compared to its FP16 performance.

TU102, however, dominates across the board. Titan RTX turns in the best benchmark numbers, followed by GeForce RTX 2080 Ti. The desktop card hangs tight with Titan RTX, achieving greater-than 90% of its performance through each change in precision.

Performance Results: Gaming at 2560 x 1440

As it stands, Nvidia actively guides gamers toward its GeForce RTX 2080 Ti instead of Titan RTX. Third-party 2080 Tis with substantial overclocks are often just as fast at half of the cost, after all. But neither TU102-based board is really necessary for gaming at 2560 x 1440. In many titles, even with their quality settings cranked up, they run up against maximum frame rate limits or CPU-imposed bottlenecks. With that said, if you own a high-refresh-rate QHD monitor, expect Nvidia’s Titan RTX to serve up blistering performance in most games at their top detail settings.

Six percent more shading resources, nine percent more memory bandwidth, and a GPU Boost clock rate that’s 8% higher than Nvidia’s GeForce RTX 2080 Ti Founders Edition gives Titan RTX a notable advantage in several of our benchmarks, though it doesn’t sweep the entire suite.

Ashes of the Singularity: Escalation (DX12)

114FireShot Capture 032 - Performance Results_ Gaming at 2560 x_ - https___www.tomshardware.com_.jpg

115FireShot Capture 033 - Performance Results_ Gaming at 2560 x_ - https___www.tomshardware.com_.jpg

116FireShot Capture 034 - Performance Results_ Gaming at 2560 x_ - https___www.tomshardware.com_.jpg

Battlefield V (DX12)

117FireShot Capture 036 - Performance Results_ Gaming at 2560 x_ - https___www.tomshardware.com_.jpg

118FireShot Capture 037 - Performance Results_ Gaming at 2560 x_ - https___www.tomshardware.com_.jpg

Destiny 2 (DX11)

119FireShot Capture 038 - Performance Results_ Gaming at 2560 x_ - https___www.tomshardware.com_.jpg

120FireShot Capture 039 - Performance Results_ Gaming at 2560 x_ - https___www.tomshardware.com_.jpg

Far Cry 5 (DX11)

121FireShot Capture 040 - Performance Results_ Gaming at 2560 x_ - https___www.tomshardware.com_.jpg

122FireShot Capture 041 - Performance Results_ Gaming at 2560 x_ - https___www.tomshardware.com_.jpg

Forza Motorsport 7 (DX12)

123FireShot Capture 042 - Performance Results_ Gaming at 2560 x_ - https___www.tomshardware.com_.jpg

124FireShot Capture 043 - Performance Results_ Gaming at 2560 x_ - https___www.tomshardware.com_.jpg

Wolfenstein II: The New Colossus (Vulkan)

125FireShot Capture 044 - Performance Results_ Gaming at 2560 x_ - https___www.tomshardware.com_.jpg

126FireShot Capture 045 - Performance Results_ Gaming at 2560 x_ - https___www.tomshardware.com_.jpg

Performance Results: Gaming at 3840 x 2160

If you aspire to game at 4K and don’t want to choose between smooth frame rates and maxed-out graphics quality, GeForce RTX 2080 Ti and Titan RTX provide a similar experience. Obviously, Titan RTX is faster. But its win is mostly academic—few gamers could justify spending twice as much money for a geometric mean of average frame rates that’s six percent higher than GeForce RTX 2080 Ti and 4% higher than the third-party model we tested.

With that said, anyone who simply must have the best of the best for gaming, regardless of price (some call this having more money than sense), won’t be able to resist.

Conclusion

Reviewing Titan RTX is daunting for the scope of its capabilities, and yet the testing is largely an exercise in exhibition. After all, if you can afford a $2,500 graphics card, you already know what you’re going to do with it. In some cases, that lofty price tag might actually represent savings to customers who were previously looking at a much more expensive Quadro card.

Physically, Titan RTX is almost identical to GeForce RTX 2080 Ti. Its soft golden color is unique, of course. And the model designator between its two axial fans correctly identifies the card. But its dimensions, weight, and even PCB are the same. Nvidia takes advantage of the overbuilt board design and cooler to enable a fully functional TU102 processor and a twelfth memory IC, pushing Titan RTX’s TDP to 280W. That extra bit of headroom allows this card to behave exceptionally well, even under the duress of FurMark. Whereas Titan V shows the limitations of its blower-style cooler under a 250W ceiling, Titan RTX didn’t stumble once with peaks as high as 300W.

130aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL0cvODIwMzg0L29yaWdpbmFsL3RpdGFuLXJ0eC1mcm9udC1wY2Itd.jpg

Titan RTX also exhibited its prowess in projects able to overwhelm the 11GB and 12GB of memory on cards like GeForce RTX 2080 Ti and Titan Xp/V. The first large-geometry OctaneRender 4 scene that Nvidia sent over for testing simply failed on the older Titans. It had to be pared down before our comparison hardware could complete the render task. Further, stepping up to 24GB of GDDR6 allowed Titan RTX to train networks using larger batches than the cards with less memory, speeding up the time it takes to work through inputs.

New support for INT8 and INT4 precision gives Titan RTX a big advantage in inferencing workloads able to leverage its hardware functionality, too. There, the card blows right past Nvidia’s still-potent (and more expensive) Titan V in our benchmarks. Where Titan RTX falls short is FP64 performance. In situations where double precision matters, Titan V remains the best FP64 card out there.

But Titan V can’t hold a candle to Titan RTX when it comes to after-hours entertainment. Because Titan RTX is based on the same TU102 processor as GeForce RTX 2080 Ti, it benefits from the work Nvidia put in to optimizing for games. As a result, it’s the fastest gaming graphics card you can buy. Now, you’ll have to decide for yourself if a few frames per second, on average, are worth what you’d pay for a pair of GeForce RTX 2080 Tis joined by NVLink.

We didn’t even touch on Titan RTX’s RT cores, which accelerate BVH traversal and ray casting functions. Mostly, there isn’t anything to test outside of Battlefield V and 3DMark's Port Royal. That latter synthetic shows Titan RTX about 6% ahead of GeForce RTX 2080 Ti, with both Turing-based cards way ahead of Titan V. Nevertheless, Nvidia says it’s working with rendering software partners to exploit ray tracing acceleration through Microsoft DXR and OptiX, citing pre-release versions of Chaos Group’s Project Lavina, OTOY’s OctaneRender, and Autodesk’s Arnold GPU renderer.

It’ll be interesting to see how Nvidia enables Turing’s highest-profile fixed-function feature. In the meantime, a complete TU102 processor is an absolute monster for professional applications able to utilize its improved Tensor cores or massive 24GB of GDDR6 memory, including deep learning and professional visualization workloads. Affluent gamers are better off with an overclocked GeForce RTX 2080 Ti from one of Nvidia’s board partners. The Gigabyte Aorus GeForce RTX 2080 Ti Xtreme 11G we tested offered more than 95% of Titan V’s average frame rate. But if money is truly no object, we couldn’t fault you for chasing that extra 5%.

this video provided by my good friend @kirill complement very well the review