Titan x test: NVIDIA Titan X Review — Monster Performance: 4K 60FPS On A Single GPU

Climatic Wind Tunnel Sweden | Vehicle Wind Tunnel Testing Sweden

What would you like to know about?

Search for:

JAMESTOWN TEST CENTER

INNOVATION SURVIVES

only when people believe in their

OWN IDEAS

Heat Transfer

Dedicated wind tunnels with computer connectivity allows the measurement of heat rejection and internal and external fluid pressure drop using the component and engine cooling module heat transfer test. This credible data serves as the base for cooling module simulations further supporting Climatic Vehicle Wind Tunnel Testing. Using this information, the maximization of the individual heat exchanger & cooling module system performance becomes possible. This provides a simple, easily repeatable test which can be performed in continuous conformance testing to ensure the provision of the expected heat transfer over the entire time span of the product by the TitanX Heat Exchanger.

Durability Testing

High application temperature and pressure destructive forces on cooling systems can be reproduced successfully at TitanX owing to our best in class facility. We strive continuously to ensure the upgradation of our testing capabilities in order to accurately simulate the increasing temperatures of modern engines as well as their applications.

Vibration Test Facilities

Simulation of years of usage can be facilitated in a matter of hours on a test bench by the TitanX engineering team after the collection of test track data. We provide 3 types of vibration tables wherein pressurization & thermal cycle load can be added to vibration testing for better simulation of field conditions:

  • Degree of Freedom (DOF)
  • 6 DOF & NARLY
  • TitanX’s 7 DOF

TitanX’s 3 translational & 3rotational DOFs with a chassis twist is known as NARLY. It incorporates load cells for all actuators thereby providing us with excellent capability for modal analysis of complete cooling modules as mounted in a vehicle.

Field and Laboratory Data Recording

Jamestown Division’s field and laboratory data recording has a diverse range of capabilities like:

  • eDAQ system with 48 HLS
  • 8 thermocouple and 24 low level channels conditioning
  • Sensors for 50 acceleration
  • 40 strain channels pressure
  • Displacement transducers

Data processing is done using Glyphworks and inField, while modal analysis is performed using ME-Scope.

Chemistry/Metallurgical Laboratory

At TitanX, we are equipped with state of the art, two staffed chemical and metallurgical laboratories capable of providing intelligent internal & external customer support for microscopic, thermal, chemical and physical analysis of metals, polymers and elastomers. Our simulation models are incorporated with this data as well. The metrology laboratories can provide layout inspection on cooling modules and components to help ensure proper installation.

Vehicle Climatic Wind Tunnel

Our climatic wind tunnel facility is located in Jamestown, NY USA which is used for the development of engine cooling & climate control systems for heavy truck manufacturers. TitanX engineers use this high-quality facility under controlled conditions to optimize product performance under a wide range of operating conditions. These climate-controlled wind tunnels can also be used for the climatic testing of other vehicular systems such as fuel, brakes and other non-vehicle applications. We work with a highly skilled team of engineers experienced in working with customers for the development of standard as well as customized testing programs.

Manage consent

NVIDIA TITAN X Review: The Pascal Beast Unleashed

NVIDIA dropped a bomb a couple of weeks back and shocked almost everyone in the enthusiast community with the surprise announcement of the new, Pascal-based TITAN X. If you missed our original coverage, the story went something like this: NVIDIA CEO Jen-Hsun Huang lost a bet, and out of the blue went and gave away a few TITAN X cards at a special AI event that was taking place at Stanford University. That may sound a little far-fetched, but the way this launch shook out, it seems somewhat plausible.

Regardless, as soon as the announcement was posted, we’re sure many of you had two questions in mind – how fast is it? And what’s the deal with the GP102? Thankfully, we can answer both of those questions and more for you today. We’ve just finished putting the brand new NVIDIA TITAN X through its paces and have the full scoop for you laid out on the pages ahead. First up though, some specifications and a quick tour of the card…

The NVIDIA TITAN X

NVIDIA TITAN X (Pascal)
Specifications & Features
NVIDIA TITAN X
Graphics Processing Clusters 6
Streaming Multiprocessors 28
CUDA Cores (single precision) 3584
Texture Units 224
ROP Units 96
Base Clock 1417MHz
Boost Clock 1531MHz
Memory Clock (Data rate) 5005MHz (Effective Speed — ~10Gbps)
L2 Cache Size 3072KB 
Total Video Memory 12228 MB GDDR5X
Memory Interface 384-Bit
Total Memory Bandwidth 480 GB/s
Texture Filtering Rate (Bilinear) 317. 4 GigaTexels/sec
Fabrication Process 16 nm
Transistor Count 12 Billion
Connectors 3 x Display Port, 1 x Dual-Link DVI, 1 x HDMI
Form Factor Dual Slot
Power Connectors One 8-Pin, One 6-Pin
Recommended Power Supply 600 Watts
Thermal Design Power (TDP) 250 Watts
Thermal Threshold 94°C
Price $1299 MSRP — Find Them At NVIDIA.Com 

In light of its specifications, the new TITAN X seems like an absolute beast. In comparison to a GeForce GTX 1080, for example, the TITAN X has 1,204 additional CUDA cores (3,584 vs. 2,560), more texture units (224 vs. 160), and more ROPs (96 vs. 64). It also has more L2 cache (3,072KB vs. 2,048KB) and a wider memory bus (384-bit vs. 256-bit).

TITAN X (New) GTX 1080 GTX 980 Ti TITAN X (Old)
GTX 980
NVIDIA GPU GP102 GP104 GM200 GM200 GM204
Architecture Pascal  Pascal  Maxwell  Maxwell  Maxwell 
Number of Cores 3584 2560 2816 3072 2048
Core Clock 1417 MHz 1607 MHz 1000 MHz 1000 MHz 1126 MHz
Memory 12GB 8GB 6GB 12GB 4GB
Memory Clock 10000 MHz 10000 MHz 7000 MHz 7000 MHz 7000 MHz
Memory Interface 384-bit G5X 256-bit G5X 384-bit 384-bit 256-bit
Memory Bandwidth 480 GB/s 320 GB/s 336 GB/s 336 GB/s 224 GB/s
TDP 250 watts 180 watts 250 watts 250 watts 165 watts
Peak Compute 11. 0 TFLOPS 8.2 TFLOPS 5.63 TFLOPS 6.14 TFLOPS 4.61 TFLOPS
Transistor Count 12B 7.2B 8.0B 8.0B 5.2B
Process Tech 16nm FinFET 16nm FinFET 28nm 28nm 28nm
MSRP $1,200 $599 $649 $999 $499

 The new TITAN X, however, has lower base and boost GPU clocks. Its effective data rate for the memory is the same though, and the TITAN X has more memory on board too — 12GB vs. 8GB on the GeForce GTX 1080. We should note that the GP102 as it exists on the TITAN X is not the full implementation of the chip. Two SMs have been disabled. The full GP102 has a grand total of 3,840 CUDA cores, as evidenced by the Quadro P6000 that was announced a few days ago.

NVIDIA GP102 Block Diagram — 2 SMs Disabled On TITAN X

The GPU at the heart of the TITAN X has a base clock of 1,417MHz and a boost clock of 1,531MHz. The GDDR5X memory on the card is clocked at 5GHz for an effective data rate of 10Gbps. At its reference clocks, the TITAN X offers up to 480GB/s of memory bandwidth and a peak texture fillrate of 317.4 GigaTexels/s, all within a 250 watt power envelope. The card requires a single 8-pin and single 6-pin supplemental power feed.

Pascal TITAN X — Front and Back

Aesthetically, the new TITAN X borrows heavily from the GTX 1080, minus some silver and plus some additional black. What you see pictured above is the only version of the TITAN X that will be for sale. Partners won’t be offering the TITAN X and it will be available only through NVIDIA.com.  The cooler, however, may (or may not) change at some point to remove the GeForce GTX branding.

Like the GTX 1080, the new TITAN X has a faceted fan-shroud with black and silver elements, a thin, sleek modular back-plate, and an aggressive overall design language that looks great to us. Should you use TITAN X cards in SLI, in adjacent slots, the rear portion of the backplate is also easily removable to allow for better airflow between the cards (we have additional pictures in the gallery below if you’d like to see what the back of the card looks like with half the plate removed).

Vapor Chamber Cooling, Dual Power Connectors…

NVIDIA’s TITAN X is outfitted with a die cast aluminum body which is machine finished and heat treated for additional strength and rigidity. The top edge of the card we received features the signature, lighted «GeForce GTX» logo, but — as we’ve mentioned — that may not be permanent.

The thermal solution on the TITAN X is just like the 1080’s. It features a radial fan and a relatively large vapor chamber linked to a dense heatsink fin array, also like previous-gen flagship cards. The fan pulls air in from above, blows it across the heatsink, and then vents the heated air outside the chassis. At the rear of the card, a second, small heatsink helps wick heat away from the VRM.

The TITAN X features a 7+2-­phase dual-­FET power array for its GPU and memory, extra capacitance in its filtering circuitry, and a power delivery network on its PCB that has been optimized for low impedance. All told, these elements result in increased power efficiency and peak-to-peak voltage noise reduction over previous-gen solutions.

The TITAN X Can Power Up To Four Displays Simultaneously

Like NVIDIA’s other Pascal-based graphics cards, the outputs on the TITAN X consist of a trio of DisplayPorts, an HDMI 2.0b output, and a dual-link DVI output. The DisplayPorts are 1.2 certified and DP 1.3/1.4 ready, which enables support for 4K displays at 120Hz, 5K displays at 60Hz, and 8K displays at 60Hz (using two cables and multi-stream transport). Up to four display outputs can be used simultaneously for multi-monitor or VR setups.

Enough with the technical jargon; let’s hit the benchmarks. ..

specs and benchmarks in 12 games and 3 benchmarks

NVIDIA started selling the GeForce GTX TITAN X on March 17, 2015 for a suggested retail price of $999. This is a desktop graphics card based on the Maxwell 2.0 architecture with a 28 nm process technology. It has 12 GB of GDDR5 memory with a frequency of 1753 MHz, a 384-bit bus, which provides a bandwidth of 7 Gb / s.

The card occupies 2 slots, connected via PCIe 3.0 x16 interface. The reference version is 267mm long, 111mm wide and 38mm high. The card requires one 6-pin and one 8-pin auxiliary power cables, the claimed maximum power consumption is 250 watts. nine0003

Specifications

GPU

GPU option
GM200-400-A1
Architecture
Maxwell 2.0
Manufacturer
TSMC
Process
28 nm

nine0016

Number of transistors
8000 million
Crystal area
601 mm²

Graphic card

Release date
March 17, 2015
Generation
GeForce 900
Predecessor
GeForce 700
Descendant
GeForce 10
Launch price
$999
Market price
$1099
Data bus interface

nine0014 PCIe 3. 0 x16

Frequencies

Base frequency
1000 MHz
Boost frequency
1089 MHz
Memory frequency
1753 MHz

Memory

Memory capacity
12 GB
Memory type
GDDR5
Memory bus
384 bit
Memory speed
7 Gb/s
Memory bandwidth
336.6 GB/s

Render configuration

Shader blocks
3072
Texture blocks
192
Screen blocks
96
Streaming Multiprocessors
24
L1 cache

nine0014 48 KB

L2 cache
3 MB

Rated output

Pixel fill rate
104. 5 HP/s
Texture Fill Rate
209.1 GT/s
Performance FP32

nine0014 6.691 Tflops

Performance FP64
209.1 GFlops

Graphics

DirectX
12 (12_1)
OpenGL
4.6
OpenCL
1.2

nine0016

Vulcan
1.1
CUDA
5.2
Shader model
6.4

Video card design

Occupied slots
2
Length

nine0014 267 mm

Width
111 mm
Height
38 mm
Heat dissipation requirements
250 W
Recommended PSU power
600 W
Outputs
1x DVI 1x HDMI 3x DisplayPort
Power supply
1x 6-pin + 1x 8-pin
Board number
PG600
Maximum temperature
83 °C
Maximum noise
52. 4 dB

Game FPS

Assassin’s Creed Valhalla

(2020)

Better than NVIDIA GeForce GTX 1660 Ti, but worse than AMD Radeon RX Vega 56.

1920×1080 / Ultra / 45.0

nine0323 2560×1440 / Ultra / 39.0

3840×2160 / Ultra / 26.4

Valorant

(2020)

Better than NVIDIA GeForce RTX 2080 Mobile, but worse than AMD Radeon RX Vega 56.

nine0323 1920×1080 / Ultra / 197. 4

2560×1440 / Ultra / 161.4

3840×2160 / Ultra / 140.4

Death Stranding

(2020)

Better than NVIDIA GeForce RTX 2080 Mobile, but worse than AMD Radeon RX Vega 56.

nine0003

1920×1080 / Ultra / 89.4

2560×1440 / Ultra / 69.4

3840×2160 / Ultra / 42.1

Cyberpunk 2077

(2020)

Better than NVIDIA GeForce GTX 1660 Ti, but worse than AMD Radeon RX Vega 56.

nine0322

3840×2160 / Ultra / 27.6

1920×1080 / Ultra / 45.0

2560×1440 / Ultra / 42.0

Apex Legends

(2019)

Better than AMD Radeon RX 5600M, but worse than NVIDIA GeForce RTX 3050 Ti Laptop GPU.

1920×1080 / Ultra / 108.0
nine0327

2560×1440 / Ultra / 80.4

3840×2160 / Ultra / 48. 0

Metro Exodus

(2019)

Better than NVIDIA GeForce RTX 2080 Mobile, but worse than NVIDIA GeForce GTX 1070 Ti.

nine0003

1920×1080 / Ultra / 54.0

2560×1440 / Ultra / 42.0

3840×2160 / Ultra / 26.4

Red Dead Redemption 2

(2019)

Better than NVIDIA GeForce RTX 2080 Mobile, but worse than AMD Radeon RX Vega 56.

1920×1080 / Ultra / 46. 2

2560×1440 / Ultra / 37.8
nine0327

3840×2160 / Ultra / 24.0

PlayerUnknown’s Battlegrounds

(2017)

Better than AMD Radeon RX 570, but worse than NVIDIA GeForce GTX 1650 Mobile.

1920×1080 / Ultra / 70.0

2560×1440 / Ultra / 54.0

3840×2160 / Ultra / 30.0

Overwatch

(2016)

Better than NVIDIA GeForce GTX 980 Ti, but worse than NVIDIA GeForce GTX 1070 Mobile.

1920×1080 / Ultra / 155.4

2560×1440 / Ultra / 107.1

3840×2160 / Ultra / 55.3

Grand Theft Auto V

(2015)

Better than NVIDIA GeForce RTX 2070 Mobile, but worse than NVIDIA GeForce GTX 1660.

1920×1080 / Ultra / 82.4

2560×1440 / Ultra / 58.2

3840×2160 / Ultra / 28. 6

Counter-Strike: Global Offensive

(2012)

Better than NVIDIA GeForce GTX 1080 Max-Q, but worse than NVIDIA GeForce RTX 2070 Mobile.

nine0323 1920×1080 / Ultra / 241.9

2560×1440 / Ultra / 201.6

3840×2160 / Ultra / 181.4

League of Legends

(2019)

Better than NVIDIA GeForce GTX 1080 Max-Q, but worse than NVIDIA GeForce RTX 2070 Mobile.

1920×1080 / Ultra / 453. 6

2560×1440 / Ultra / 252.0

nine0323 3840×2160 / Ultra / 100.8

Benchmark tests

Octane Render OctaneBench: 125 Better than NVIDIA GeForce GTX 780 Ti, but worse than NVIDIA GeForce GTX 980 Ti.

Passmark: 13001 Better than NVIDIA GeForce GTX 1660 SUPER, but worse than NVIDIA Titan X Pascal.

Unigine Heaven 4.0: 2604 Better than NVIDIA GeForce GTX 980 Ti, but worse than NVIDIA GeForce GTX 1070.

Comparison with other graphics cards

Map 1:

Map 2:

Video reviews and tests

Error 403 The request cannot be completed because you have exceeded your quota. : quotaExceeded

If you find an error or have something to add, write in the comments. Have you had any experience with this video card? Is it worth buying such a card today? nine0003

NVIDIA Titan X Pascal — review and test of video card

Review and test of NVIDIA TITAN X Pascal, probably the most long-awaited video card in mid-summer from the green team, given the market positioning of the GeForce GTX accelerator in 1080, when the price and manufacturing partners are known. After the announcement on July 23, 10 days have passed and, as it was reported, the new product should be available on the market on August 2, 2016. Everything came together here. Here is the next thesis that needs to be checked, because according to the developers, 60% is faster than the previous «Titan» on the Maxwell chip. nine0003

[contents]

Table of contents:

An interesting aspect Nvidia at the last moment decided to separate TITAN into a separate brand, untying the GeForce GTX designation, it becomes the third family after the GeForce and Riva itself. The company thus accomplishes two goals. First, it adds exclusivity, setting TITAN apart from other graphics cards from the company. Secondly, it allows you to maintain a separate pricing policy, which is actually seen from the price of the device.

NVIDIA has changed tactics towards the availability of expensive graphics cards of this generation. Previously, the company has developed a performance chip for the sub-$600 segment and one «big chip» for high-end GeForce, Quadro (professional graphics) and Tesla (HPC) products. nine0003

This is something NVIDIA is avoiding this generation by developing the two «big pieces» of the GP100 debuted with the Tesla accelerator supporting FP64 CUDA cores with exotic HBM2 memory and the new GP102 chip intended for consumer graphics and professional graphics applications that do not use double precision in computing (FP64). This chip also does not use HBM2 in favor of the available GDDR5X.

The tested TITAN X Pascal is NVIDIA’s first product based on the GP102 chip. Note the physical capabilities: 3840 CUDA cores (3584 activated) scattered across 30 streaming multiprocessors in 6 graphics processing clusters. There are also 224 TMUs, 96 ROPs and 12 GB GDDR5X 384-bit bus. But in fact, the GP102 is 1.5 high-end «pebbles» GP104.

Due to the lack of real competition from AMD in this generation, NVIDIA may be more aggressive in pricing. The cost of NVIDIA TITAN X Pascal is set at a staggering $ 1200. The company’s marketing move, the rejection of partner projects (ASUS, EVGA, Zotac, etc.). The decision to sell the video card through your website and OEM channels of gaming PC vendors (CyberPowerPC, Alienware, etc.). nine0003

Architecture

The flagship product is based on the new second largest chip bearing the designation GP102, based on the latest Pascal architecture. The campaign’s «largest» GPU is the GP100, which drives the Tesla P100 accelerator and is built for HPC applications. The GP102 chip is essentially a GP100 minus the FP64 compute for CUDA cores, but the presence of 3840 CUDA cores for FP32 and instead of the 4096-bit HBM2 memory interface, 384-bit GDDR5X. As mentioned earlier, the GP102 makes up one and a half high-end GP104 chips that power the GeForce GTX 1080 graphics cards.0003

With each successive architecture since Fermi, NVIDIA has enriched the Streaming Multiprocessor (SM) by adding dedicated resources and reducing shared resources within the graphics processing cluster, resulting in large performance gains. The story continues with Pascal. Like the GM200 before it, the GP102 has six graphics clusters (GPCs), specialized units in the GPU, a PCI-Express 3.0 x16 connection slot, a 384-bit GDDR5X memory interface with 12 controllers, and a data transfer rate of 10Gbps respectively. nine0003

Six GPCs are populated with data via the GigaThread Engine (GTE) with 3 MB cache. Graphics clusters carry five streaming multiprocessors (TPCs), in the GM200 the ratio was one to four. In the Pascal architecture, the streaming multiprocessor is equipped with the fourth generation PolyMorph engine, a component for basic rendering setup operations. The streaming multiprocessor (SM) also contains eight texture processing units (TMUs) plus 128 CUDA cores poured into each. Therefore, the GP102 chip has a total of 3840 CUDA cores, of which TITAN X Pascal accounted for 3584 (28 SM). Recall that 2 out of 30 SMs were «cut down» by NVIDIA in order to increase the volume of usable chips. nine0003

Other important features include 224 TMUs and 96 ROPs. The default GPU frequency is 1417 MHz, in Boost mode 1531 MHz. It seems that these are lower frequencies than the GTX 1080 video cards, but you need to understand that the GP102 has a lot more chips, and even at current speeds, its TDP is not designed higher than 250W. For comparison, the original GeForce GTX TITAN X (Maxwell) graphics cards were clocked at 1000MHz and 1089MHz boosted.

The new TITAN X Pascal adopts the features of the GDDR5X standard. The memory operates at an effective frequency of 10 GHz. Theoretically, bypassing the 384-bit interface, its bandwidth is 480 Gb / s. Although, given the architecture features and advanced in terms of optimizing delta color compression, the increase should be 20% and equal to 576 Gb / s. nine0003

The Pascal architecture supports Microsoft’s asynchronous computing standards by adding its own concept of «Dynamic Balancing».

Packaging

Packaging:

  • Graphic Accelerator
  • Documentation

Card

The design of the NVIDIA Titan X Pascal shroud is identical to the younger brother GeForce GTX 1080 Founders Edition. The only change is that it is completely black with no «silver». Also on the front side of the case, the illumination of the GeForce GTX logo is preserved. On the back side, the same plate is found as on the GTX 1080, which can be partially removed for better air circulation in the SLI hitch. Card dimensions are 27 cm x 11 cm.

Card requires two user system slots.

Display connections, includes DVI-D, HDMI and three DisplayPort ports. Unlike previous NVIDIA graphics cards, the DVI port does not include an analog signal, so you will need to use an active adapter. NVIDIA has also updated the DisplayPort 1.2 interface to version 1.3/1.4 certified, providing support for 4K at 120Hz and 5k @ 60Hz, or 8k @ 60Hz via two cables.

The

GPU also comes with a port for HDMI audio devices. It is HDMI 2.0b compliant, supporting HD audio and Blu-ray with 3D movies. The video encoding block on the GPU has been updated to support HEVC in 10 and 12-bit. nine0003

With Pascal, NVIDIA has made some changes to how the SLI configuration functions. In a nutshell, for 4K at 60Hz and above, the company recommends a new high-bandwidth SLI bridge dubbed «SLI HB». These bridges occupy both fingers. Also gone are the traditional three- and four-core SLI configurations. Only in some tests it is possible to run more than a double SLI installation, in which all gaming applications are limited.

The photo above shows the front and back parts of the printed circuit board, disassembled video accelerator. nine0003

Closer

Titan X Pascal cooling system identical to GTX 1080 Founders Edition. All components are cooled: the processor, memory chips and voltage regulator. Compare a well-ventilated evaporation chamber that instantly absorbs the heat from the chip.

As mentioned earlier, NVIDIA has redesigned the rear panel to be thinner than ever, allowing for more airflow to the second card in an SLI configuration. If you want to improve air circulation even more, because the casing consists of two separate plates that can be removed without the main cooling. The photographs show the removal of each of these elements. The plate on top of the GPU is a thermal pad that covers the voltage regulator. nine0003

Special screws from NVIDIA.

NVIDIA uses a 6+8 power connector configuration. Power consumption of 250 watts, of course, all 300 watts will be optimal.

uP9511P voltage regulator, same as found on GTX 1070 and GTX 1080.

GDDR5X memory chips are made by Micron and labeled «D9TXS» decoding the MT58K256M32JA-100 marking. Their specified functionality is 1250 MHz (effective 10,000 MHz).

NVIDIA’s GP102 graphics processor is the company’s second TOP chip using the Pascal architecture. It is manufactured in 16nm process technology in Taiwan by TSMC, Taiwan, with 12 billion transistors and a die size of 471mm². nine0003

Test configuration

  • All results of other video cards were obtained on a system with the same configuration.
  • All games are set to high quality unless otherwise noted.
  • AA and AF through the game settings, not through the driver control panel.

Each gaming application has been tested on these screen resolutions:

  • 1600×900: Resolutions for small displays of modern laptops (17″ — 19″).
  • 1920×1080: A common widescreen resolution for large displays (22″ — 26″).
  • 2560×1440: The maximum possible resolution is 16:9 for public displays (27″-32″).
  • 3840×2160: 4K Ultra HD resolution available on the latest high-end monitors.

Anno 2205

Three years have passed since Anno 2070 and Ubisoft returns with an even more futuristic version of the hit, the plot of which will take the gamer to the year 2205, when humanity colonized the moon as part of scientific expeditions, creating a thriving economy being there. In addition to this, the player gets the opportunity to build pretentious cities on Earth in various climatic conditions. nine0003

The engine at the heart of Anno 2205 uses DirectX 11 capabilities, and unlike most RTS games, should be quite demanding on modern graphics hardware given the number of recreated models and their detail.

Assassin’s Creed: Syndicate

The latest Assassin’s Creed Syndicate is set in 19th century London, where an assassin must infiltrate the city’s underworld to take it back from Templar control. Based on Ubisoft’s proven AnvilNext engine, which has served faithfully for several predecessors in the series, the game takes advantage of DirectX 11 capabilities and is heavy on modern hardware. nine0003

Battlefield 3

Battlefield 3 is a combination of infantry gun combat with mechanized warfare, including vehicles: armored personnel carriers, tanks, attack helicopters and aircraft. Role-playing elements make the battles even more exciting. The game has a single player campaign that adds several gigabytes to the disk after installation.

EA-DICEF’s Frostbite 2 engine is behind it all. This makes use of all sorts of DirectX 11 components, including hardware tessellation and new lighting effects. Fast PCs are rewarded with better image quality. nine0003

Battlefield 4

EA’s latest online shooter hit, Battlefield 4 is designed with the PC platform in mind. There is a single player campaign, but the priority is the multiplayer component. Significantly enhanced gameplay and includes the return of Commander Mode, «Levolution», Battlescreen and adds naval combat.

Based on DICE’s polished Frostbite 3 engine, Battlefield 4 benefits not only from DirectX 11, but also from DirectX version 11. 1 compatible GPUs. This is the first game for 64 bit systems by default. nine0003

Batman: Arkham Knight

The console version won awards for excellent gameplay, and at the presentation of the PC version of Batman: Arkham Knight was awarded critical errors, which affected its release and distribution. The latest version is good enough for single GPU systems, we didn’t notice any problems that affected the test, the results were reliable in terms of reliability.

Based on the Unreal game engine built for «next-gen hardware», Batman: Arkham Knight uses DirectX 11 capabilities.

Call of Duty: Black Ops III

The franchise’s product design and visuals are getting better every year, and Black Ops III is no exception. The game uses a heavily modified version of the IW engine used in Black Ops II. This allows DirectX 11 high resolution textures to be used.

Crysis 3

You wake up in a cell to find your long-lost friend and teammate «Psycho», who explains how C. E.L.L practically took over the world by monopolizing energy production in Crysis 3. The game allows the Crysis trilogy to come to its logical conclusion by ending C.E.L.L in the final epic battle. Based on CryEngine 3 and fully compatible with DirectX 11, Crysis 3 is one of the most visually stunning game titles in history. nine0003

Fallout 4

The most anticipated game of 2015, Fallout 4 sets the tone for Fallout 3 and Fallout: New Vegas. You play as a nameless hero who has woken up 210 years after an apocalyptic nuclear world war to fix what’s left of post-apocalyptic Boston.

Drives the game with Bethesda’s Creation engine, its strong first and third person presentation features. The game uses the capabilities of DirectX 11 and moderately «leans» on modern hardware. nine0003

Far Cry Primal

The game draws concepts from the distant prehistoric times of Central Europe, where man still struggles with the forces of nature to become the dominant species on Earth. First person view, prehistoric weapons and tools to hunt, gather and tame wild animals.

Based on Ubisoft’s latest Dunia engine, the game takes advantage of DirectX 11 and puts a lot of strain on high-end GPUs. nine0003

Grand Theft Auto V

Arguably the most anticipated PC release of this decade, Grand Theft Auto V is Rockstar Games’ latest action-packed world where you play not one, but three heroes who switch between missions. Also interesting is the online multiplayer, where the gamer can go on a Grand mission called «heists» with his friends in a cooperative mode.

Released 18 months after being released on consoles, this PC release is a rare occasion where a developer takes the time to perfect a game for a new platform. Enthusiasts consider the PC release to be the best, taking advantage of the higher CPU and GPU power to open up stunning environments that even the latest generation of consoles would die of shame. The game plays well with PC-exclusive technologies such as multi-GPU, multi-monitor scaling, etc. We tested the game with ultra-high settings and MSAA disabled. nine0003

Hitman

The game uses a game engine by IO Interactive, one of the first to use DirectX 12. However, the DirectX 12 implementation of the game is too full of bugs to be integrated into our booth. In this test, we are testing the game in DirectX 11 mode, which is still quite unpretentious on high-performance GPUs.

Just Cause 3

The long-awaited return of our saboteur/murderer/regime-changer hero Rico Rodriguez, who came to liberate the country from the evil clutches of the dictator in the fictional Mediterranean island of the Medici. The game is based on the same technology as the other gems of 2015, Mad Max from Just Cause 3 and is not deprived of beautiful views and an extremely high degree of detail. nine0003

Rainbow Six: Siege

This latest shooting game from Tom Clancy, in which tactical special squad go after a new global terrorist organization known as The White Masks. Based on Ubisoft’s AnvilNext game engine, Rainbow Six: Siege uses DirectX 11 capabilities, but is only partly resource hungry due to the proximity of large objects in close combat, and partly because the project is trying to win over a massive user base from similar online shooters like Counter Strike: Global Offensive. nine0003

Rise of the Tomb Raider

Sequel to the 2013 reboot of the franchise, Rise of the Tomb Raider. Based on improved Crystal Dynamics, the game uses DirectX 11 and DirectX 12 capabilities. We use DirectX 12 for our testing.

The Witcher 3: Wild Hunt

The Witcher 3: Wild Hunt 2015 action RPG. You are Geralt, the Witcher, who is in search of his lost friend Ciri, the hero will have to explore a huge in-game world. The game is filled with side stories unrelated to quests that make you feel like there’s actually something else going on in the world besides the main story. nine0003

The Witcher 3 uses REDengine 3 specifically designed for open world RPG environments. It uses DirectX 11 features with tessellation, and some post-processing effects. In the interest of neutrality, we played the game, but with limited options.

Power Consumption

Titan X Pascal’s non-gaming power consumption is slightly higher than other GTX 1000-series cards, but the differences are sort of small.

A typical gaming draw is about 40 watts higher than the GTX 1080, with peaks reaching 268 watts. The marginal charge for «food» was presented at the level of 260 watts. nine0003

Considering the performance obtained, the Titan X Pascal is the most efficient graphics card we have ever tested, as it is 6% more efficient than the GTX 1080. to completely eliminate fan noise during idle, light gaming and Internet surfing. The reported noise level of 30dBA isn’t annoying, but the numbers are certainly among the worst we’ve seen in a while. nine0003

In games, the fan turns on for real, reaching almost 40 dBA. In my opinion, this is too much.

Total performance

This page shows a combined summary of the execution of all the tests on the previous pages, broken down by permissions. Each diagram of the tested board shows 100% to other cards, as in relation to it.

Performance per Watt

The following charts show the performance of the cards in our test group. We used relative performance indicators and average energy consumption results. nine0003

Performance per dollar

If you’re looking for the best «feed» per dollar, you’ll love this page. We looked at the current USD price of each card on Newegg and used that and all the relative performance metrics to calculate the performance index per dollar.

Overclocking

Before overclocking, I did a quick test to see how the Titan cooler handles heat and what happens to frequencies.

As you can see, the frequencies start dropping immediately (with the introduction of Boost 3.0) and continue until about 1550 MHz. They will stabilize later as it appears to be the point at which Boost 3.0 finds stable conditions. nine0003

The overclocking results given in this section are achieved with the fan and voltage set to the default settings defined in the VGA BIOS. We choose this approach because it is the most realistic scenario for most users.

Each sample overclocks differently, so our results here can only serve as a guide to what you can expect from your card.

On NVIDIA cards with Boost enabled, the most discussed value is to observe the increase in frequencies after overclocking. For this review, I’ve applied boost to all clock levels. nine0003

The maximum overclock of our sample is +232 MHz to the base frequency of the GPU, which increases the maximum boost from 1835 MHz to 2063 MHz (12% overclock) and 1405 MHz memory (12% overclock).

Titan X is definitely limited by the board’s power limit and temperatures above 83°C. Nevertheless, there is a significant performance boost in real life — take a look below.

Using these clock speeds, we ran a quick test of Battlefield 3 to see if it would benefit from overclocking. nine0003

Actual 3D overclocking performance is 7.9%.

As additional data points, I set the power limit, the maximum temperature limit and the fan speed to 100%. This gives an additional 7% performance boost with base clocks. With active overclocking, we get 11% brings to the overall performance with default frequencies, and with the overclocked option we get an increase of 19.3%!

Temperatures

Idle temperatures are excellent, which is not surprising since the fans do not stop at idle. During gaming sessions, the temperature rises above 83°C, which leads to lower frequencies due to Boost 3.0. nine0003

Frequency profiles

Modern video cards have several frequency profiles to balance power consumption with operational requirements.

The following table lists the clock settings for important GPU performance and voltage scenarios.

The card uses NVIDIA’s dynamic overclocking engine, GPU Boost 3.0. It will dynamically adjust frequencies and voltages based on visualization of load, temperature and other factors. nine0003

In the graph below, we have recorded all GPU frequencies and voltages in combination for a 1920×1080 resolution for a benchmark set. The plotted points are transparent, allowing them to be stacked to indicate more commonly used values. A translucent color means the clock/voltage combination is rarely used, while a saturated one means it is active more often.

Rating and review

Titan X Pascal graphics cards are available exclusively from the NVIDIA website for $1200.


  • Fastest single-chip graphics card ever tested
  • High energy efficiency
  • 12 GB VRAM
  • Rear collapsible plate included
  • New technologies from NVIDIA: Ansel, FastSync, HEVC Video and VR
  • HDMI 2.0b, DisplayPort 1.4

  • Extremely high price
  • Low cooling
  • Cooler does not turn off in standby mode
  • Noisy in games
  • Limited floating point performance

Rating:
8.8

NVIDIA recently dropped the Titan X Pascal bomb, surprising almost everyone with the fact that the hero appeared on the market exactly on schedule, on the same day, as announced in the announcement. The card is based on the brand new GP102 GPU, and is pretty much 50% wider in scale than the GP104 version used on the GTX 1080. The new GPU also includes features geared towards the more professional space, like INT8 support, an ability useful for machine learning. However, the card does not support the full speed and accuracy of FP64 and FP16 calculations — unlike previous Titan cards, made more interesting in terms of video display and other content creation tasks. But these features are completely useless for the players, let’s focus on what is important to you. nine0003

Out of the box, at 4K resolution, the Titan X Pascal is 30% faster than the GTX 1080. The performance compared to the flagship of the previous generation GTX 980 Ti is another +37%. Only the GTX 1080 is honored to compete with the new leader in SLI mode, and then in games that scale well in SLI. This means that Titan X Pascal, roughly speaking, «sits» in the middle, between the GTX 1080 and GTX 1080 SLI.

NVIDIA reused the cooling system from the GTX 1080 Founders Edition, the cooler was already too weak to handle the heat dissipation of the GTX 1080. As a result, the GTX Titan X suffocates in thermal gas almost all the time; we tested it and reached the maximum Boost value, a frequency of 1835 MHz, active for a few moments before slowly falling off until the card reached its thermal limit. On average, the monster works for 1692 MHz, which is still much higher than the base frequency of 1418 MHz. I did a quick test with power and temperature limit. In this scenario, the performance gain of the card was about 5-6% for 4K, depending on the game. When I did the same test with our max clocks using overclocking, the performance gain was a whopping 19% over par out of the box! During idle, the fan noise is low, however, the board does not have a low load or idle shutdown feature. In gaming, the card is definitely noisier than the GTX 1080 Founders Edition, reaching almost 40 dBA, which is quite audible. As with the GTX 1080, the cooler is the weakest part of the Titan X Pascal. nine0003

Energy efficiency is an amazing feature of the accelerator, though, as with all other Pascal-based cards. In typical games, the card uses only 205 watts, with peaks up to 268 watts, which is the limit of the power bar configured with the possibilities; even if the limit were higher, the cooler couldn’t handle the extra heat anyway. Compared to AMD’s TPD, NVIDIA is incredibly far ahead. GTX Titan X is almost twice as efficient as AMD Polaris cards.

Titan X Pascal available directly from NVIDIA; there will be no custom designs from partners, and no other stores will sell it. Several system integrators will offer Titan X Pascal as part of a complete system. On the forums, many users have reported problems with the device ordering process, broken «Order Now» buttons, «Deal failed», no emails, and extremely long processing times of two weeks for direct debit. On the other side. I’ve seen a lot of posts from people getting their hands on the Titan X Pascal, so the orders are definitely working. nine0003

NVIDIA set a price of $1200 for the Titan X Pascal, no not $1199, the law of the genre is to save that last dollar. Also, on the subject of pricing, it’s up 20% from the first Titan X, which cost «only» $999. This pricing might sound a little crazy, though, it’s not surprising; NVIDIA completely controls the market in the hi-end segment, AMD barely reaches half the performance and efficiency of this video card. If you were in NVIDIA’s place, you would also use this advantage, right? Looking closely at the performance-per-dollar test, we see that the GTX 1080 is 30% higher, not that much, of course, if there is money to spend on such a device. Titan X Pascal is guaranteed to give you the highest frame rates no matter what game you’re playing. Aside from the price you can either afford or vice versa, the big problem for me is the engineering of the card. NVIDIA once again applied a cooling solution that couldn’t handle the heat dissipation of a fabulously designed GPU, resulting in a significant loss in performance, and without any custom designs. Well, the only alternative is water; the first waterblocks are already appearing on the market — this is the way I would use it.