After the expansion of the GeForce RTX 40 lineup with the corresponding announcement of improved SUPER models, the GeForce RTX 4070 Ti SUPER graphics cards aroused the most interest. The increase in the number of processors, memory capacity, and bus bit depth promises a performance boost and good prospects. And all this at the price of a regular GeForce RTX 4070 Ti. NVIDIA partners have offered many modifications of the new line of graphics adapters. Today we are reviewing one of the top options – ASUS ROG Strix GeForce RTX 4070 Ti SUPER OC 16GB.
NVIDIA GeForce RTX 4070 Ti SUPER
The GeForce RTX 4070 Ti SUPER graphics card uses the AD103 GPU in the AD103-275 modification with 8448 active cores, 264 texture modules, and 96 rasterizers. The chip in this configuration also offers 264 tensor cores and 66 RT units for ray tracing. Compared to the performance of the functional units in the processor used for GeForce RTX 4070 Ti, we see that they have increased by about 10%. At the same time, the frequency formula has remained virtually unchanged, with only the declared base value increasing minimally – 2340/2610 MHz vs. 2310/2610 MHz.
But the memory subsystem has been significantly transformed. GeForce RTX 4070 Ti SUPER is equipped with 16 GB instead of 12 GB in the base version. To increase the memory capacity by 25%, the bus bit rate has increased accordingly – from 192 bits to 256 bits. Since the operating frequencies of GDDR6X chips remained unchanged at an effective 21,000 MHz, the bandwidth also increased by a third – from 504 GB/s to 672 GB/s. This is a good option for high-resolution graphics modes with maximum picture quality.
Despite the significant functional changes, the developers claim that the power consumption of the GeForce RTX 4070 Ti SUPER will remain at the same level as the base version of the GeForce RTX 4070 Ti – 285 watts. The cards are equipped with one 12VHPWR connector to connect additional power
|GeForce RTX 4080
|GeForce RTX 4070 Ti SUPER
|GeForce RTX 4070 Ti
|5 nm, TSMC
|5 nm, TSMC
|5 nm, TSMC
|TMU / rasterizers
|304 (4th generation)
|264 (4th generation)
|240 (4th generation)
|76 (3rd generation)
|66 (3rd generation)
|60 (3rd generation)
|GPU clock speeds
|16 GB GDDR6X
|16 GB GDDR6X
|12 GB GDDR6X
|Effective memory frequency
|22 400 MHz
|21 000 MHz
|21 000 MHz
|Video card power consumption (TGP)
Comparing the technical specifications of GeForce RTX 4070 Ti and GeForce RTX 4070 Ti SUPER, it is obvious that the SUPER modification actually uses the platform from the older GeForce RTX 4080 16 GB. However, the latter still retains a significant advantage, having 15% more compute units and other functional units, as well as increased memory bandwidth. Therefore, even at the stage of first familiarization with the parameters of video cards, we can assume that GeForce RTX 4080 will retain its performance advantage.
As for the price, the recommended price of GeForce RTX 4070 Ti SUPER 16 GB for the US market is $799. At the time of the announcement in January 2023, the GeForce RTX 4070 Ti 12 GB had an identical price tag.
In the case of the GeForce RTX 4070 Ti SUPER, NVIDIA will not offer a Founders Edition reference model, relying on the designs of its partners. However, the latter will not have to make significant efforts here, it will be enough to minimally optimize the existing platforms used for GeForce RTX 4080. Manufacturers have offered a wide range of models with various configurations, frequency formulas, and equipment. Today we are reviewing one of the top versions from ASUS.
ASUS ROG Strix GeForce RTX 4070 Ti SUPER OC 16GB
The video card comes in a medium-sized cardboard box with a large photo of the device on the front panel.
The graphics adapter comes with an adapter from three 8-pin connectors to one 12VHPWR for connecting additional power, a short paper user manual, a reusable velcro tie for laying power cables, and an adjustable height stand to support the graphics card.
We got a version of the ROG-STRIX-RTX4070TIS-O16G-GAMING video card with additional factory overclocking of the GPU for review. In this case, the GPU frequency formula has been changed from the recommended 2340/2610 MHz to 2340/2670 MHz.
Visually, the model looks like a “technological beam” of considerable size. The overall dimensions of the video card are 336×150×63 mm, and it weighs 1800 g. According to the developers’ interpretation, ROG-STRIX-RTX4070TIS-O16G-GAMING has a 3.15-slot design, and this should be taken into account when purchasing.
The “sturdy build” of the model from the STRIX line is not surprising. The video card is equipped with a three-fan cooling system. This configuration itself largely determines the size of the adapter.
The video card cooler uses a massive radiator unit with two set cassettes of aluminum plates connected by heat pipes to accelerate heat transfer and make more efficient use of the total cooler dissipation area. At the heart of the heat sink is a nickel-plated copper plate, which is in contact with both the GPU and memory chips. There is no evaporation chamber, which was the case with the RTX 4080 from the STRIX line.
As we have already mentioned, three fans are used to blow the radiator unit. In this case, the fans are 100 mm in diameter and are mounted on double ball bearings, made using Axial-tech technology with a solid connecting ring on the outer radius.
This technique, with its specific blade shape, allows for better directionality and increased air pressure at lower speeds. To reduce the possible effect of turbulence, adjacent fans rotate in reverse. Usually, the video card provides a hybrid cooler mode with fans stopping in the idle mode when the GPU temperature does not exceed 55C.
The ASUS ROG-STRIX-RTX4070TIS-O16G-GAMING uses a 10-layer PCB with dimensions of 240×145 mm. The platform offers a 16-phase power supply subsystem for the GPU based on the Monolithic Power Systems MP2888 controller and OnSemi NCP303151 power assemblies with an operating current of up to 50A. The memory subsystem power supply has a 3-phase circuit with a uP9512Q control chip and identical DrMOS assemblies.
The long-awaited moment – GeForce RTX 4070 Ti SUPER has eight memory chips with a total capacity of 16 GB. GDDR6X chips are located next to the GPU. In this case, Micron chips (D8BZC) are used, which, according to the specification, are designed to operate at an effective 21,000 MHz.
The overall design of the adapter is reinforced by an additional metal bracket. On the back of the video card, there is an aluminum protective plate that protects the components and PCB from mechanical damage.
As we have already mentioned, the total length of ASUS ROG-STRIX-RTX4070TIS-O16G-GAMING is 336 mm, while the PCB itself is much more compact (240 mm). In the segment away from the interface panel, additional holes are provided for through ventilation of the radiator cassette. This is another way to improve the efficiency of the cooling system.
To connect additional power, there is a single 12VHPWR connector on the top edge of the video card. A red indicator light is provided near the power connector. The indicator will be on, reminding you to connect the power cable. Pulsation is an indicator that the power supply is insufficient. For a system with ASUS ROG-STRIX-RTX4070TIS-O16G-GAMING, the developers recommend using a power supply with a capacity of 750 W or more. In the official specifications of GeForce RTX 4070 Ti SUPER, NVIDIA indicates that a 700+W PSU is required. So it is not surprising that a slightly larger power reserve is suggested for an overclocked video card.
On the back of the video card, you’ll notice an additional balancing bracket in the area where the GPU is located. The stainless steel frame distributes the clamping force when installing the cooling system.
The video card is equipped with two BIOS chips with different firmware versions that optimize the cooling system operation. A miniature toggle switch on the top edge of the adapter allows you to select one of two profiles – standard P Mode and Q Mode, optimized for noise reduction.
In the case of ASUS ROG-STRIX-RTX4070TIS-O16G-GAMING, the developers allow you to optimize the overall cooling system, if necessary, taking into account the heating characteristics of the video card. Two additional 4-pin connectors are provided on the side surface for connecting external fans. The FanConnect II technology is more interesting for rather specific conditions, but if necessary, you can customize the algorithm of the connected fans depending on the GPU temperature.
Additional backlighting of the video card remains an option, which in dosed forms can only increase the overall attractiveness of the adapter. In the case of the new model of the STRIX line, the developers managed to strike a balance. For illumination, there is a logo of the game series on the top edge, as well as segmented “lights” around the perimeter of the side panel, which quite organically complement the overall aesthetic background during operation. The use of addressable aRGB elements with AURA Sync synchronization support provides more opportunities for customization and approximation to the “beautiful” with personalized views.
The manufacturer has equipped the interface panel with five digital outputs. There are three DisplayPort 1.4a and two HDMI 2.1a. As standard, GeForce RTX 4070 Ti SUPER models offer a 3+1 configuration. However, an additional HDMI will not be superfluous. Although the graphics adapters allow you to connect up to four screens at the same time, the second HDMI adds more flexibility to possible display configurations.
Test platform configuration
To test the capabilities of the new video card, we used a system from Artline built on the latest Core i7-14700K chip (8/16+12; 3.4/5.6 GHz + 2.5/4.3 GHz) complete with an ASUS PRIME Z790-A WiFi motherboard (Intel Z790) and 32 GB of DDR5-6000 memory.
The configuration is close to the one we recommend for a progressive gaming configuration in the traditional “PC of the Month“. Except for the processor – in the basic version, Core i5-14600KF is offered. Although a more powerful chip in this situation will definitely not hurt, in addition, in the new 14th generation Intel Core lineup, it is the Core i7 chips that have received the greatest acceleration after increasing the number of energy-efficient cores.
If you want to get the most out of your 14th generation Intel Core chips, you need to choose your cooling system carefully. Powerful CPUs have a considerable level of power consumption under load, so a liquid cooling option is desirable for efficient heat dissipation. In this case, the PC is equipped with the brand new ASUS ROG Strix LC III 360 ARGB with an advanced Asetek Gen7 V2 pump and three 120mm ROG Strix AF-12S ARGB fans.
- Processor: Intel Core i7-14700KF (8/16+12; 3.4/5.6 GHz + 2.5/4.3 GHz)
- Motherboard: ASUS PRIME Z790-A WIFI (Intel Z790, DDR5)
- Cooling: ASUS ROG Strix LC III 360 ARGB
- Memory: G Skill DDR5-6000 32 GB (F5-6000J3040F16G)
- Storage device: WD BLACK SN770 1TB
- Power supply: ASUS ROG-STRIX-850G-AURA-GAMING (850 W)
- Case: ASUS TUF Gaming GT502
A video card with such a large cooler simply cannot help but support a mode that stops the fans at low load.
So, ASUS ROG-STRIX-RTX4070TIS-O16G-GAMING is expected to be quiet at rest. The GPU temperature in this state remains at 33C.
During a gaming load, the GPU warms up to 65C, with a peak of 79C in the hottest spot. The temperature of the memory chips was kept at 70C. To achieve these values, the fan speed was raised to 1500 rpm. This is less than half of the maximum allowable level (3200 rpm). As you can see in the screenshot, the GPU accelerated to 2850 MHz at its peak. The average operating frequencies during games are slightly lower, but obviously the standard factory overclocking affects the overall formula.
When using a firmware version with Q Mode enabled, system parameters change. The GPU warms up to 71C, and the GDDR6X chips to 76C. In this mode, the fan speed only goes up to 1150 rpm, so the video card was very quiet. This option can be recommended to those who are trying to get a powerful system with minimal noise.
In Q Mode, the maximum operating frequency of the GPU has not actually changed, but the average performance during games has decreased by 30-50 MHz. Therefore, the difference in performance with P Mode will be minimal.
When evaluating the capabilities of the new model, the most interesting question is the actual difference in performance between the GeForce RTX 4070 Ti SUPER 16 GB and the base model of the series. So we chose the GeForce RTX 4070 Ti 12 GB for comparison. Obviously, even without additional testing, it is clear that the SUPER version will be faster in any case, but it is the actual difference in performance in different conditions that is interesting.
Let’s start with the “synthetics” by comparing the results of video cards in popular test packages. The screenshots show the results of ASUS ROG-STRIX-RTX4070TIS-O16G-GAMING. For easy perception, we’ll add the GeForce RTX 4070 Ti 12GB and present the results in a comparison chart.
The 3DMark benchmarks show a 10-16% increase in performance of the SUPER version. We can immediately notice that the difference is smaller in the basic Time Spy mode, and in Time Spy Extreme with higher resolution and ray-tracing stages, the new product gains a greater advantage.
Another area of application where increasing the memory capacity of the video card and its bandwidth can significantly improve the platform’s performance is rendering, visualization, and media content management tasks. For example, multiple additional filters for adding various effects when transcoding video usually increase the requirements for the memory subsystem.
During the Cinebench 2024 benchmark rendering, the ASUS ROG-STRIX-RTX4070TIS-O16G-GAMING significantly improved the results of the base GeForce RTX 4070 Ti. The performance has increased by 28% and this is the largest difference recorded so far. Tests in Blender 4.0 and V-Ray 5.0 also reveal a significant difference in the capabilities of both graphics cards – 15% and 20%, respectively.
Performance in games
Let’s take a look at the differences in games. First, let’s check the capabilities of the graphics cards in Full HD mode with maximum picture quality without ray tracing and intelligent scaling.
According to the results of tests in ten fairly resource-intensive projects, the advantage of the SUPER version over the base version is about 6-8% on average. It should be noted that ASUS ROG-STRIX-RTX4070TIS-O16G-GAMING has a certain factory overclock, which increases the difference by several percent. So, in the basic versions with reference frequencies, the difference will be even more modest. Wow, that’s a great start. So many technical improvements, and the result is only a conditional +5%? Yes, in 1920×1080 mode, even 12 GB of memory is enough to store textures, so increasing the capacity to 16 GB and increasing the bandwidth no longer brings any special advantages to the SUPER version
When ray tracing is activated with maximum quality, the difference between the graphics cards increases slightly (8-14%). However, even in this case, both adapters provide very comfortable performance. The only exception here is Cyberpunk 2077, which in RT Overdrive mode makes even the owners of graphics cards of this level suffer. However, it should be noted that when switching to RTX 4070 Ti SUPER, the average value increased by 14%, but in terms of the important minimum fps indicator, we recorded an increase of 42%. So in this case, the average fps on the RTX 4070 Ti was the same as the minimum value for the RTX 4070 Ti SUPER.
The addition of DLSS for “difficult” cases with ray tracing can make a fundamental difference. A good example of this is Cyberpunk 2077, where performance actually doubles. As for the power ratio of both test participants, the positions remain almost unchanged. With the exception of Dying Light 2, where activating scaling allowed RTX 4070 Ti to reduce the gap from 11% to 2%.
When using the 1440p graphics mode, the computational load on the GPU increases and the memory capacity of the video card increases. However, even in this case, there is no radical change in the positions of the video cards. The advantage of the overclocked version of GeForce RTX 4070 Ti SUPER has increased to 10% on average. Overall, we have excellent performance for comfortable gaming with maximum quality without DLSS. Even in such heavyweights as Starfield and Remnant II, the RTX 4070 Ti SUPER can achieve average 75-90 fps, and performance in less demanding games exceeds 100 fps.
With maximum ray tracing activated, the video cards also perform well. The expected exception is Cyberpunk 2077 RT Overdrive, where decent fps can only be obtained with DLSS. Scaling usually allows you to seriously improve performance, but there are some nuances here as well. Forza Motosport almost ignores enabling this option.
In general, during the RT and DLSS experiments, the difference in performance between the GeForce RTX 4070 Ti 12 GB and GeForce RTX 4070 Ti SUPER 16 GB actually remained at 10%. Rising to 15% only in the case of Cyberpunk 2077 RT Overdrive.
The increase in memory capacity to 16 GB allows GeForce RTX 4070 Ti SUPER owners to feel more comfortable in 4K mode. However, the actual superiority of the SUPER version over the basic one, even in this mode, does not seem to be anything out of the ordinary. According to the results of the tests, the difference is about 12% on average in favor of the new model.
From the list of games we used, only Starfield, Remnant II, and Cyberpunk 2077 will require additional settings adjustments to get an average of 60 fps. In other projects, it was possible to use the maximum quality without any reservations.
After enabling the ray tracing effects in 4K, you have to look into the settings section. The same Cyberpunk 2077 RT Overdrive is capable of bringing even a 24GB GeForce RTX 4090 to its knees, so it’s no surprise that the GeForce RTX 4070 Ti SUPER will need some simplifications. Of course, DLSS is a very effective “cure” for such situations, although scaling alone may not be enough. We should pay tribute to the GeForce RTX 4070 Ti 12 GB, which in general held its own in 4K, being 10-15% inferior to the SUPER-novelty. That is, it worked at its level without unexpected failures even when RT was activated.
Well, the initially expected advantage of GeForce RTX 4070 SUPER 16 GB over GeForce RTX 4070 Ti 12 GB has been quantified. And how do the features of the new product compare to those of similar models that did not make it into this “version”? First, we should note the direct competition with the Radeon RX 7900 XT 20 GB. After acceleration, the updated NVIDIA model is actually on an equal footing with the AMD video card, outperforming the latter when ray tracing is activated.
At the same time, the GeForce RTX 4070 Ti SUPER 16 GB is 10-15% less than the older GeForce RTX 4080 16 GB, and, accordingly, its successor, the GeForce RTX 4080 SUPER 16 GB. This means that the price/performance ratio will be better with the RTX 4070 Ti SUPER.
If we compare the performance of GeForce RTX 4070 Ti SUPER 16 GB with that of the previous generation models, then the new product manages to outperform the GeForce RTX 3090 24 GB and even come close to the GeForce RTX 3090 Ti 24 GB in 4K mode. It is worth reminding that the latter was introduced two years ago with a price tag of $1999. Now you can get a similar amount of fps for $799. Well, with some adjustments, of course.
NVIDIA claims that the power consumption of the GeForce RTX 4070 Ti SUPER 16 GB will remain at the same level as that of the base GeForce RTX 4070 Ti 12 GB. Despite the use of a more powerful GPU with a larger crystal and number of active units, increased bus bit depth, and the number of memory chips, the specifications indicate a total power consumption (TBP) of 285 W.
Given the standard overclocking of the ASUS ROG-STRIX-RTX4070TIS-O16G-GAMING, we can obviously expect slightly higher performance. In practice, under gaming load, the video card required 310 watts of power, compared to 289 watts recorded for the GeForce RTX 4070 Ti. So the figures are really close. The stock RTX 4070 Ti SUPER models fit within the announced package. The developers have managed to optimize the parameters of the SUPER version so that it does not require additional power consumption. Given the higher performance, we can talk about further improvements in energy efficiency.
In the context of power consumption, we recall that the competitive Radeon RX 7900 XT 20 GB has a declared TBP level of 300 watts. In this case, the difference is quite small, especially considering that AMD’s model has a 320-bit bus and 20 GB of memory, and therefore there are 10 GDDR6X chips on the board instead of eight in the RTX 4070 Ti SUPER 16 GB and six in the RTX 4070 Ti 12 GB.
Given the ” reserve” of the existing cooling system, it was interesting to check the video card’s potential for additional overclocking at least in express mode. The ASUS ROG-STRIX-RTX4070TIS-O16G-GAMING version has already received basic acceleration from the manufacturer, but of course the video card is capable of more. The model allows you to increase the Power Limit by up to 28%.
By expanding the energy limit, we experimentally managed to increase the GPU base value by 130 MHz to 2470 MHz, while the average acceleration value increased from 2670 MHz to 2800 MHz. The memory chips were also overclocked from 1313 MHz to 1500 MHz, so the effective values increased from 21,000 MHz to 24,000 MHz (+14%).
After additional overclocking, the GPU came close to the 3 GHz mark and accelerated to 2985 MHz under load. By changing the Power Limit, we got a corresponding increase in power consumption to almost 360 W. Is it worth it?
Additional frequency boost brings about a 5% increase in performance. Therefore, the question of the practical feasibility of such experiments remains in the rhetorical plane. The cooling system is ready for such challenges, but it is up to each individual to decide whether additional overclocking is justified.
GeeForce RTX 4070 Ti SUPER: conclusions
It was the GeForce RTX 4070 Ti SUPER that was the most anticipated model from NVIDIA’s updated SUPER line. Does it live up to expectations? Most likely, yes, although with some clarifications. The new model is equipped with 16 GB of memory. That’s it, and that’s all we can say about it. The presence of 12 GB of memory in the base GeForce RTX 4070 Ti is the main factor that caused a wave of fair criticism of this model. NVIDIA has done some conditional work on the bugs by increasing the local memory capacity and bus bit depth.
Given the unchanged frequency of GDDR6X chips, the memory subsystem bandwidth has also increased by a quarter. In addition, given the significantly increased number of compute units and related GPU functional units, one could potentially expect the performance of GeForce RTX 4070 Ti SUPER 16 GB to increase by 15-20%. Why did the practical results turn out to be more modest? Even when it comes to the OS version of the card with a natively accelerated GPU, we have a 7-14% increase in fps. If we compare adapters with the recommended frequency values, the difference is generally 5-10%.
In the case of Full HD mode, the capabilities of the GeForce RTX 4070 Ti SUPER are obviously already excessive. Therefore, an additional increase in memory capacity and memory bandwidth does not lead to tangible improvements. For 1080p with maximum quality settings, a GeForce RTX 4070 SUPER 12 GB or even a basic GeForce RTX 4070 12 GB is enough, and alternatively, a Radeon RX 7800 XT 16 GB.
When using the 1440p mode, the advantage of the SUPER modification becomes more noticeable, and more often reaches ~10%. This coincides with a relative increase in the number of GPU function units, although one could assume that a 33% increase in memory capacity and bandwidth would bring even more acceleration. Why didn’t this happen? Practical tests show that even when using 2560×1440 resolution, 12 GB of memory is usually enough. Or the need does not significantly exceed this figure. In such cases, a capacious L2 cache buffer of 48 MB allows you to partially compensate and “smooth out” a certain shortage without significant performance degradation.
As for the 4K mode, the advantage of GeForce RTX 4070 Ti SUPER over its predecessor by ~12-15% looks more convincing. Although to be honest, there were certain expectations that in such conditions the presence of 16 GB of memory and a radical increase in ROM would lead to even more significant acceleration. However, this is a clear confirmation that it makes no sense to evaluate a video card only by its technical characteristics, let alone by its individual parameters. The balance of all components is important.
So how can you take the appearance of GeForce RTX 4070 Ti SUPER 16 GB? In general, it’s positive. Finally, a model of this level from NVIDIA has 16 GB of memory and a certain performance improvement. You always want more, but let’s not forget that the new SUPER version has the same price tag as the original GeForce RTX 4070 Ti 12 GB. So, in fact, we have +5-10-15% performance for the same money and, in addition, confidence in the potential for “heavy” projects that will surely require more than 12 GB of video card memory. A good side bonus of the GeForce RTX 4070 Ti SUPER 16 GB is the significant reduction in the cost of the main competitor – Radeon RX 7900 XT 20 GB.
When we asked Google Bard to write a haiku about GeForce RTX 4070 Ti SUPER, we got the following life-affirming result:
Speed and beauty,
RTX 4070 Ti SUPER,
The game world is yours.