If you have decided on a suitable 144hz monitor, you should also consider whether your PC or graphics card can even reach a frame rate of up to 144fps in computer games or, in the case of a 4K monitor, can also handle this resolution. If this is not the case, we will of course be happy to help you choose the right graphics card for your gaming monitor!
You can see the Ranking of the best Graphics Cards for 1440p / 4K monitors and for 144 fps Gaming above. Below you will find the Minimum Requirements of the Graphics Card for 1440p / 4K monitors / 144fps and the in-depth analysis of each Graphics Card in the ranking.
Minimum Requirements for Graphics card (GPU) for a 144hz monitor:
The graphics card should reach a certain number of frames per second, so that the purchase of a 120hz / 144Hz monitor is worthwhile. Thus, the monitor gives you an advantage as long as the FPS number remains constantly above 60 frames per second. Ideally, the gaming graphics card should reach an average of 144 frames per second.
- Graphics memory: at least 6 GB
- Model (Nvidia): GTX 1060 or higher
Minimum Graphics card (GPU) for 1440p / 4K monitors:
For normal 1440p / 4K monitors, the current limit is 60 Hertz. However, PCs with 4K monitors have an incredible amount of processing power to handle, so a good graphics card is required.
- Graphics memory: at least 6 GB
- Model (Nvidia): GTX 1070 or higher
Ranking First: EVGA GeForce RTX 2080 Ti OC
- Best Performance for 144 Hz Gaming
- Very Silent
- High Overclocking Potential
- Great Performance costs..
One of the all-inclusive carefree packages please! The GeForce RTX 2080 Ti can be a normal graphics card, but there are also models that are bursting with additional functions and want to offer the highest level of factory overclocking. The EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming is such a model – and that’s what we want to take a look at now.
It remains the same: The start of the GeForce RTX cards can be described as bumpy. The poor availability of the GeForce RTX 2080 Ti continues. A supposedly higher failure rate of the card is still unsolved. The GeForce RTX 2080 can’t prevail against a GeForce GTX 1080 Ti without RTX functions in the games and the GeForce RTX 2070 is still too expensive for many newcomers – additionally the missing RTX and DLSS functions in the games come into play.
From an architectural point of view, there are no differences between the GeForce RTX 2080 Ti models. The TU102 GPU based on the Turing architecture offers 4,352 shader units, 68 RT cores and 544 tensor cores. The graphics memory is 11 GB GDDR6, which is connected via a 352 bit wide memory interface. At a clock speed of 1,750 MHz, the memory bandwidth is 616 GB/s.
When comparing the models with each other, the clock differences become clear. The entry level models of the GeForce RTX 2080 Ti offer a boost clock of 1,545 MHz, the Founders Edition comes up to 1,635 MHz. The gap to the high-end models is very small in terms of clock speed. The EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming should go to work with at least 1.755 MHz. This is considerably more than an ASUS ROG Strix GeForce RTX 2080 Ti OC with 1.665 MHz in OC mode, although in the end it depends on which boost clock can be kept above the minimum requirements. This is exactly what we’ll look at on the next page.
EVGA specifies a thermal design power of 300 W. This is the standard power target and can be increased by 24.3% to 373 W. Thus, the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming gives a lot of leeway in this regard.
Design and Hardware
Not surprisingly, the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming as a triple-slot slot card with a length of 305 mm goes beyond all dimensions. The cooler itself has a height of 2.75 slots, even though EVGA has installed a triple-slot bezel. So you should have enough space in the case to accommodate the card.
EVGA specifies a boost clock of 1,755 MHz for the GeForce RTX 2080 Ti FTW3 Ultra Gaming. In practice, we see clock rates of well over 1,900 MHz – to be more precise, we are in the range of 1,950 to 1,965 MHz. However, the clock also drops slightly below 1.950 MHz every now and then. Overall, this model is the fastest GeForce RTX 2080 Ti in terms of clock rate, as we have been able to test so far.
EVGA has extensively revised the iCX2 cooler for the GeForce RTX series. We see three axial fans in this model. The semi-transparent fan bezel is a trademark of the current series. EVGA also offers certain customization options, which we’ll have a look at later. Taking the PCI Express interface’s contacts in proportion, the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming’s dimensions become clear at first glance – both in length, height and thickness.
On the rear side EVGA installs a backplate which covers the complete PCB. Instead of a completely closed backplate, EVGA provides some openings to prevent heat accumulation. EVGA does not need any illumination of the backplate.
The three axial fans on the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming have a diameter of 90 mm each. These rotate under load with 1,850 revolutions per minute. Due to the iCX2 sensors, EVGA can also control the three fans differently.
The fans are switched off as soon as the GPU temperature falls below 44 °C. However, the fans can also be switched off at an earlier time by individual control. From a GPU temperature of 56 °C, the fans start working and provide fresh air.
A glance at the front side shows the enormous height of the cooler. It occupies a total of 2.75 slots. The semi-transparent cover of the fans is covered with thin aluminium and the name EVGA and the model name are engraved on it. Since there are LEDs underneath, both light up prominently during operation.
Power is supplied not only by the PCI Express slot with a theoretical 75 W, but also via two 8-pin connectors, each of which can supply 150 W. With a maximum power target of 373 W, this looks like “sewn on edge”. But you have to know that especially the PCI Express connectors can handle much more than the 150 W specified in the specifications and can also deliver depending on the power supply.
Like every GeForce RTX 2080 Ti, the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming also offers an NVLink interface with two links. This enables a bandwidth of 100 GB/s, although this is currently not used with GeForce cards. However, the NVLink interface allows two of these cards to be used in SLI.
Right next to the additional power supply, a switch can be seen, labeled “Normal” and “OC”. According to EVGA, the switch should allow switching between different fan curves. But in our case the fan curves looked the same in both cases. A query to EVGA about the exact function behind the switch is currently underway.
A look at the rear end of the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming shows that the PCB and the cooler are somewhat shorter than the cooler’s cover. Some of the built-in heatpipes also end here. In addition, several header connectors can be seen here, which can be used to connect and control a case fan, for example. The same applies to the control of the RGB lighting.
A closer look at the cooler’s cover shows once again the design of semi-transparent plastic and the aluminum inserts. Not everyone will like the new design and unfortunately the plexiglass scratches quite quickly.
The slot bezel is certainly exceptional with a height of three slots and has rarely been used in this form before. EVGA doesn’t use the additional space to blow out warm exhaust air, for example, but closes the metal panel for the most part. Three DisplayPort 1.4 and one each for HDMI 2.0b and VirtualLink are available at the display outputs.
Components and Ports
The EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming makes a good impression with the cooler. Now let’s have a look at what’s under the hood.
Of course we didn’t miss the chance to remove the cooler, because EVGA has created its own PCB, which should improve the already good power supply of the GeForce RTX 2080 Ti in the NVIDIA reference design. Just like the cooler, the PCB has huge dimensions.
As GPU the TU102-300A-K1-A1 is used. This offers a full 18.6 billion on 754 mm² and is thus the largest and most complex consumer GPU from NVIDIA. EVGA uses the GDDR6 memory chips on 11 of 12 slots. These should be placed as close as possible to the GPU and at the same distance from it in order to keep the signal runtimes as equal as possible.
EVGA provides a total of 19 voltage phases for the supply of GPU and GDDR6 memory. Six of these voltage phases are located to the left of the GPU package, the remaining 13 to the right. Three phases are responsible for the memory, the remaining 16 take care of the GPU. However, these are not 16 independent phases, but 16 dual FETs in which two phases are always controlled by a doubler. Such a large number of voltage phases does not directly result in a higher overclocking or overclocking potential, but the waste heat from the power supply is distributed over 19 instead of 10 phases, which makes it easier to dissipate the waste heat.
In the back of the PCB are the controller and measurement circuits of the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming. Under the 8-pin connectors, however, you can also clearly see how much space is available on the PCB. However, thick and thus massively expanded conductor paths embedded in the PCB can also be seen.
Each of the inputs to the power supply, i.e. the PCI Express slot and the two 8-pin connectors, is monitored. For this purpose EVGA uses three shunt resistors, which monitor the current flow over the three inputs. These shunt resistors are here green SMT components in the middle of the picture.
Without the built-in cooler we can take a look at the two header connectors again. On the right, the pins for RGB control are visible, on the left the additional fan connector.
To come back to the dimensions of the PCB, let’s have a look at the NVLink connector area. EVGA uses the height of the PCB to arrange the 13 voltage phases to the right of the GPU package on top of each other. This should make it possible to supply the GPU more evenly through the individual phases. There is therefore plenty of free space elsewhere on the PCB.
Now we want to take a closer look at the cooling of the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming.
On the backside of the backplate you can see some heat conduction pads. These should better dissipate waste heat, which is also present on the back of the PCB. So the backplate has a cooling effect, but how big this effect is is difficult to judge. At least EVGA tries to prevent heat accumulation.
The actual cooler consists of a solid copper block that is nickel-plated. Only the GPU rests on this block. Seven heatpipes are designed to dissipate the waste heat and transport it to the large aluminium heat sink. With the height of 2.75 slots, EVGA wants the surface area significantly increased, which should improve the overall cooling.
Two heatpipes lead the waste heat into the front area, the other five into the rear area of the heatsink. In the front area, EVGA flattens the heatpipes and contacts them directly with the aluminum heatsink, since there is not enough space to lead them through the cooler itself.
On the front side, EVGA uses a front plate to cool components such as the GDDR6 memory chips and some more of the power and voltage supply. Both the front and back of the front plate have thermal pads to provide the best possible contact.
In addition to thermal pads, EVGA also uses a thermal paste for the VRMs and a heat pipe for a better distribution of the waste heat.
EVGA offers the possibility to customize some GeForce RTX cards with accessories. The so-called Trim Kits we now also have a look at. Beside the Trim Kits EVGA also offers a Shield, which covers the front of the card.
First we want to mount the shield. To do so, we have to loosen four screws on the front and four more on both sides of the board with the included tool.
After that the shield can be mounted, which covers the front of the card. The shield costs $30 for EVGA and is made of aluminum. We couldn’t detect any negative effect on the cards’ cooling performance. Due to the narrow webs of the shield, the effect should be small anyway.
The trim kits are available in the colors red/black and red/white. They cost $23 each in EVGA’s online shop. The dark elements are built into the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming. We once mounted the red and white color variants on the card. To do this, we clamped it into the cooler’s cover from behind and then fastened it with the lateral screws that were previously loosened.
The white trim kit doesn’t have much of an effect at first sight, which is certainly due to the mounted shield.
When installed and with the light activated, the white Trim Kits provide additional light emission. So the effects of the lighting become a bit more intense.
As a consequence, we have mounted the red trim kits and with these, the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming shows itself in a significantly different look.
The red Trim Kit “blocks” the colored illumination. But with red LEDs they fit very nicely into the overall look of the card.
Finally, let’s have a look at the Precision-X1-Tool from EVGA. With this tool the most important functions of the card can be controlled. Clock of GPU and memory, power limit, fan speed, temperature limits and much more can be adjusted. In addition there is an automatic overclocking via NV scanner and the possibility to adjust the fan curves and much more.
Noise Levels, Temperature and Performance
Since the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming completely shuts off its fans in idle mode, it can be considered silent in this state. Under load it remains astonishingly quiet at 41 dB(A) and is the quietest model of the GeForce RTX 2080 Ti that has passed our tests so far.
The overclocked models of the GeForce RTX 2080 Ti break all barriers for the power consumption of the entire system. This also applies to the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming. In hardware, 98% of the performance increase of the cards is bought by a higher power consumption. The rules of physics cannot be circumvented. However, we have already seen models of the GeForce RTX 2080 Ti that consumed far more.
A similar picture emerges when measuring the power consumption of the card itself. The EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming draws 308.3 W from the power supply.
Both in idle and load mode, the GPU temperatures of the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming remain in the low range. Under load, 61 °C is a very good result.
Thanks to gigantic power and voltage supply and good cooling, the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming is supposed to offer ideal conditions for overclocking. In the meantime, however, it has been shown that the GeForce RTX cards have a certain sound barrier along which everyone moves.
For the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming we set the power limit to +120%, turned up the coolers and then let the NV scanner run. This reached a maximum clock rate of 2.065 MHz, but we weren’t satisfied with that. So we manually continued to turn the clock screw and ended up with a maximum boost clock of 2.145 MHz. We operated the GDDR6 memory at 2,000 MHz.
Among the air-cooled cards, models like the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming are the preliminary final stage in development. From the dimensions of the PCB and the cooler is hardly possible. Also regarding the clocking, not much more is probably feasible. Currently, only ZOTAC is trying out a clock setting of more than 1,800 MHz. But these specifications are largely meaningless, in the end it’s all about the cooling and the specifications regarding the power limit.
This is exactly where EVGA scores with the GeForce RTX 2080 Ti FTW3 Ultra Gaming. The cooler doesn’t let the GPU get warmer than 61 °C. Only when the user turns the power limit, voltage and clock screw do the temperatures rise again. While the VRMs also stay quite cool, the GDDR6 memory chips get considerably warmer. The temperatures go up to just over 90 °C with strong overclocking. But such temperatures don’t pose a problem.
This ensures that the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming keeps a boost clock of 1,950 MHz and a bit more. This makes it the fastest GeForce RTX 2080 Ti that we have been able to test so far. However, the distances between the cards are sometimes very small. We can measure a few more FPS, but the player won’t feel this.
We have already mentioned the very good cooling, which not only provides for low temperatures but also works very quietly. Under load, 41 dB(A) is simply a good argument for the card with the given performance. The fans’ standstill in idle mode is also a plus point. The card buys the performance with a higher consumption, but if you spend already $1400 for a graphic card, you shouldn’t worry too much about an additional consumption of 40 to 50 W.
The EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming offers a few additional functions, apart from the performance and the given readings. These include the fan connector, which allows controlled operation of a case fan. The optics, which can also be adjusted via the trim kits and the shield, are also very appealing. A trim kit is included when buying the card in EVGA’s online shop and offers at least a rudimentary possibility of adapting the card.
Also worth mentioning is the three year warranty that EVGA offers and the manufacturer is also known for its good customer service in the forum.
All in all the EVGA GeForce RTX 2080 Ti FTW3 Ultra Gaming ranks First in the best Graphics Cards for 144Hz Gaming and 1440p / 4K Monitors for its great Performance.
Ranking Second: Sapphire Nitro+ RX 5700 XT
- Great Performance for 4K Monitors and VR
- Great Price
- Not great for Overclocking
The first batch of custom models for AMD navigation maps is available. But we have not yet looked at all of the most important models. According to various recommendations, the Sapphire Nitro+ Radeon RX 5700 XT is one of those models that you should take a close look at. Whilst the performance is almost identical on almost all cards, the cooling is moving further and further into focus. A detailed test of the Sapphire Nitro+ Radeon RX 5700 XT can be found on the following pages.
After NVIDIA has struggled in the past weeks with some super models in between, today we will again focus on a model with AMD GPU. The Sapphire Nitro+ Radeon RX 5700 XT has already received top marks in several tests and is now trying its hand at us. With the reference version of the Radeon RX 5700 and Radeon RX 5700 XT, AMD has left some room for its board partners. Sapphire, among others, wants to use this potential for this model.
So now follows the Sapphire Nitro+ Radeon RX 5700 XT, which like every other Radeon RX 5700 XT uses the Navi-10-GPU in the expansion stage with 2,560 shader units. We are talking about 40 Compute Units (CU), 64 shader units are available per Compute Unit. Compared to the GCN architecture (Graphics Core Next), these are organized somewhat differently – but we have examined this in detail in our consideration of the RDNA architecture.
In short: AMD can combine two CUs into one workgroup processor. This makes sense if certain workloads previously had to be split between two CUs, but can be better handled in one CU.
Due to the memory connection we see 64 render backends (ROPs). Along with the ROPs comes the identical memory interface with a width of 256 bits. This addresses 8 GB of GDDR6 memory, which is built into eight memory chips. The memory has a clock speed of 1,800 MHz (14.4 GBit/s per pin) and thus achieves a memory bandwidth of 460.8 GB/s. The Sapphire Nitro+ Radeon RX 5700 XT has 160 texture units.
The number of Radeon RX 5700 XT models that promise a boost clock of more than 2,000 MHz is relatively long and is in the range of a dozen cards. With a boost clock of 2,010 MHz, the Sapphire Nitro+ Radeon RX 5700 XT is in good neighbourhood.
Let’s move on to the first technical specifications of the Sapphire Nitro+ Radeon RX 5700 XT, which do not concern the GPU and memory.
The card has a length of 300 mm, so depending on the case, you should make sure that there is enough space. The PCB has a length of 260 mm, so the cooler is slightly longer. The height of the cooler is 2.5 slots, which is not a big problem due to the less and less used additional cards in a system. The three fans have a diameter of 95 mm, so it is also clear that Sapphire is dimensioning the active and passive cooling components as large as possible. From a GPU temperature of 50°C the fans switch off.
The potential of the coolers becomes apparent when looking at the temperatures. A GPU temperature of 70 °C is a first positive sign and also the edge temperature remains pleasingly low at 85 °C. We only reach the boost clock of 2.000 MHz and more for a short time in some games. On average, the Sapphire Nitro+ Radeon RX 5700 XT clocks its GPU with 1.950 MHz for the BIOS, which sets a power limit of 225 W. In the second BIOS it is 200 W, which is still 15 W more than in the reference version.
All benchmarks were done with the OC-BIOS.
At first glance, the Sapphire Nitro+ Radeon RX 5700 XT is a card with a restrained design – at least without being plugged into a slot. Sapphire’s current design language for the Nitro series includes the Octagone, which we find on the cooler.
The Sapphire Nitro+ Radeon RX 5700 XT’s backplate continues the look of the front and goes a few steps further. A lot of silver is interrupted by black stripes. The Nitro logo is RGB lit as soon as the card is turned on. On the right, you can also see that the cooler is slightly longer than the PCB.
The three fans have a diameter of 95 mm each. This gives them a comparatively large diameter – this is in the upper area of the fans installed on a graphics card. From a GPU temperature of 52 °C the fans switch on, at 50 °C they switch off again.
On the front of the Sapphire Nitro+ Radeon RX 5700 XT there is a Sapphire logo, which, like the Nitro logo on the back, is illuminated by RGB LEDs. We will have a look at how this looks in operation in a moment.
At the back end of the card you can find the additional power connections. Two times 8-pin should be more than sufficient for a Radeon RX 5700 XT. Theoretically, 300 W can be fed to the card via this alone. With a maximum thermal design power of the card of 265 W, the additional connections already cover the demand.
The excess length of the cooler is clearly visible at the rear end of the card. On the rear side, Sapphire saves a certain area of the backplate here and thus allows a portion of the air that is blown through the cooler by the rear fan to escape.
The BIOS switch is located in direct proximity to the slot bracket. Here you can switch between the two UEFI versions, which are not only a simple update, but also allow the card to work in different operating modes.
The display connections are located on the slot bracket. Sapphire uses two HDMI 2.0b and two DisplayPort 1.4 connectors, but there’s not much more to say at this point, as all current gaming monitors can be operated via the connections provided.
Plugged into a PCI Express slot and switched on, the Sapphire Nitro+ Radeon RX 5700 XT unfolds its full RGB splendour. There are illuminated elements on the front, in the transition to the backplate and on the backplate itself.
Noise Levels, Temperature and Performance
The Sapphire Nitro+ Radeon RX 5700 XT is already convincing in terms of volume. In idle mode, the fans are at a standstill from 50 °C and below either way, but we also speak of an extremely low volume level under load. The performance of a Radeon RX 5700 XT is combined here as well as possible with a low overall volume.
In terms of power consumption, the Radeon RX 5700 XT models can’t quite keep up with the level that they actually wanted to operate. Especially when the manufacturers want to squeeze a bit more out of the hardware, the cards quickly allow themselves a bit more.
With 241 W, the Sapphire Nitro+ Radeon RX 5700 XT is not really economical as such. With the alternative BIOS and a GPU power of 200 W, a few watts can be saved, but the performance then of course also drops.
We don’t see any problems with the temperatures. The Sapphire Nitro+ Radeon RX 5700 XT’s cooler has no problems at all to keep all components at sufficiently low temperatures.
The Sapphire Nitro+ Radeon RX 5700 XT is one, if not the best model of the current navigation generation. However, among the fastest maps the air is very thin and the distances are therefore extremely small. Nevertheless, this model combines the highest possible performance for a Radeon RX 5700 XT with extremely good cooling.
In terms of performance, it must be said that the Radeon RX 5700 XT is a 1440p card and can compete with the GeForce RTX 2070 Super and sometimes the GeForce RTX 2080 Super. In the absolute high-end segment, NVIDIA remains alone and no Sapphire Nitro+ Radeon RX 5700 XT, no matter how good, can change that.
Let’s get back to the positive aspects again and these are certainly to be found in the cooling, apart from the performance. The Nitro+ Radeon RX 5700 XT’s cooler keeps both the memory and the GPU at low temperatures. Although other cards also reach a value of 70°C for the GPU, paired with the given performance and especially the volume, no model of the Radeon RX 5700 XT reaches this.
We’ve refrained from overclocking the Sapphire Nitro+ Radeon RX 5700 XT or rather, we’ve refrained from taking a closer look at it. We were able to increase the GPU clock by 30 to 50 MHz and the GDDR6 memory could also be overclocked by 250 MHz, but since the card already almost completely exhausts the boost clock in the standard settings, there’s little leeway left for overclocking.
With the alternative BIOS, the Sapphire Nitro+ Radeon RX 5700 XT offers an interesting possibility to use the card in a more efficient operating mode. The necessary settings can also be made manually, but with the pre-defined BIOS, anyone can quickly and easily switch to this mode.
The elaborate RGB lighting is certainly also a factor that influences the purchase decision for some. Sapphire offers here, so to speak, the maximum range with an illumination on all relevant sides of the map. Optionally, even illuminated fans would be possible, which Sapphire offers.
Due to the many positive aspects, especially the performance in combination with the volume, we give the Sapphire Nitro+ Radeon RX 5700 XT the Excellent grade and therefore second Rank in the best Graphics cards for 144Hz Gaming and 1440p / 4K monitors. Also due to its comparatively low price it got our ‘Best Price’ winner in this test.
Ranking Third: ASUS GeForce RTX 2070 Super OC
- Good Performance for 144 Hz Gaming and VR
- Many Ports
- Good Price
- Not a lot Equipment
The Geforce RTX 2070 is the cheapest or rather least expensive Turing graphics card from Nvidia. It is faster and more economical than a Geforce GTX 1080 or Vega 64 and costs almost as much depending on the model. We have tested two Geforce RTX 2070 variants from Asus and MSI.
After the Geforce RTX 2080 Ti and the Geforce RTX 2080, Nvidia has put a third graphics card based on the Turing architecture into the market with the Geforce RTX 2070. In opposition to the two top models, the smaller offshoot is cheaper, halfway available and above all more convincing in terms of price: We tested an overclocked one from Asus and one running with standard frequencies from MSI and got a clear recommendation.
The Geforce RTX 2070 is based on a chip called TU106, whereas the 106 offshoots are usually used for models like the Geforce GTX 1060 – thus, easily one or two price categories below the $520, which is at least due for a Geforce RTX 2070.
Background is that the TU106 with 445 mm² surface and 10.8 billion transistors already represents a high-end and not a midrange CPU according to previous standards. The TU106 of the Geforce RTX 2070 is a full version, so the graphics card has 2,304 shader units and a 256-bit interface with 8 GByte GDDR6 memory.
However, Nvidia differentiates between a TU106-400-A1 and a TU106-400A-A1: The former are used for graphics cards that adhere to the reference specifications for clock speed and power dissipation, while the latter are used for models with a higher clock speed and increased board power.
The Asus Geforce RTX 2070 Strix OC we tested therefore uses a TU106-400A-A1 and the MSI Geforce RTX 2070 Armor a TU106-400-A1. The A-chips are supposedly supposed to be better overclockable, but Nvidia especially allows more generous power targets via software, which is important for overclocking.
Because the architecture is the same as in the Geforce RTX 2080 (Ti), the Geforce RTX 2070 also accelerates integrated raytracing effects in upcoming games and masters inferencing, which is used for DLSS (Deep Learning Super Sampling).
Of both, there are only demos from Nvidia and no commercially available title that supports such an implementation. The probably first one will be Battlefield 5, the dice shooter with raytracing mirrors will be released on November 20th, 2018. Further games like Shadow of the Tomb Raider or Pubg will get new RTX options via patch at an unknown date.
Hardware and Design
The Geforce RTX 2070 ROG Strix Gaming OC is positioned by Asus above the Geforce RTX 2070 Turbo with DHE cooling, the Geforce RTX 2070 Dual and the Strix without OC suffix. The graphic card is 305 mm long and has a 2.5-slot design. Background are the three 90 mm fans, which sit on a powerful cooler.
At the top of the board next to the I/O panel is a small dip-switch with which we switch between performance and quiet-firmware. The temperature is low in both modes – we recommend Quiet with 66 degrees Celsius under 4K gaming load and also because the fans switch off in idle.
A small button next to the firmware switch deactivates the red lights around the propellers, also the red ROG logo on the backside goes out then. The light effects are adjusted by the Aura-Tool. The backplate serves more for stability than for cooling, especially since the GDDR6 memory is on the front anyway.
Asus installs an 8-pin and a 6-pin connector, which gives enough leeway to increase the power target beyond 215 watts. The board has voltage measurement points for GPU and memory as well as the PLL, and there are also two connectors for case fans.
The Strix has two DisplayPort 1.4, two HDMI 2.0b and Virtual Link via USB-C on the connection side. The latter is suitable for VR headsets, for 4K displays and, funnily enough, also for external SSDs, or it charges smartphones. There is a Windows tool called GPU Tweak II for the Asus card: Users can select the OC mode here.
Then the RTX 2070 applies a nominal boost of 1.845 MHz instead of 1.815 MHz and the power target increases by 10 percent. In practice we see 1.925 MHz instead of 1.890 MHz – so the difference is negligible, there is not noticeably more power.
MSI’s Geforce RTX 2070 Armor hardly turns out shorter than Asus’ Strix with 300 mm, but it is surprisingly high with 135 mm. MSI also relies on a 2.5-inch design, but there are only two idle-stopping 95-mm fans instead of three. A backplate and an 8-pin as well as a 6-pin connector are also present, whereby an 8-pin connector would actually be sufficient for this graphic card.
Two DisplayPort 1.4, two HDMI 2.0b and Virtual Link form the connections. The MSI logo shines in many different colors, and the colors can be set under Windows using the Mystic Light tool.
MSI clocks the Armor with an official 1.620 MHz in boost mode; we measured 1.780 MHz at 63 degrees Celsius. Because the memory bandwidth is the same in the Asus Strix, it’s the GPU frequency that distinguishes the two graphic cards. Both models overtake a Geforce GXT 1080 in the Founder’s Edition or a Radeon RX Vega 64.
As a test system for the Geforce RTX 2070, we again use a Z370 board with a Core i7-8700K (test) with 16 GByte DDR4 memory. All games and Windows 10 x64 v1803 are on a Samsung PM981, but the current Geforce 416.34 is used as the driver. It contains optimizations for new games and reduces the idle power consumption of the RTX graphics cards compared to the older Geforce 411.15.
The Geforce RTX 2070 ROG Strix Gaming OC is on average 17 percent faster in 4K and 1440p than a Geforce GTX 1080 in the Founder’s Edition and thus also clearly faster than overclocked custom designs. The Geforce RTX 2070 Armor from MSI with its lower GPU clock rate can still distinguish itself by 13 percent from Nvidia’s reference graphic card of the previous generation and is thus on the level of various overlocking models.
The difference between the Asus and MSI card turns out to be small at 4 percent, the difference isn’t noticeable. Both Geforce RTX 2070s are well suited for 1440p gaming and, with some concessions, also for 4K resolution.
Both graphics cards require about 12 watts in idle, which is significantly more than in a Geforce GTX 1080 Founder’s Edition. Under 4K gaming load, the Geforce RTX 2070 Armor from MSI reaches 174 watts, so it is more economical and faster than Nvidia’s reference design.
The Geforce RTX 2070 ROG Strix Gaming OC from Asus approves itself measurably more with 212 watts, but the performance difference turns out marginal, as already mentioned. Both the Asus card and the MSI card are very quiet under load; they don’t get significantly louder even in the case. The fans are idle anyway.
If you want to overclock a Geforce RTX, you have to consider the power target and can use Nvidia’s OC scanner, which is, for example, now integrated in the Afterburner tool. Both graphic cards have GDDR6 video memory, which can handle 8 GHz instead of 7 GHz. The MSI card can be set to 114 percent power target, manages a passable 1.935 MHz with overclocking, then needs 195 watts and is 11 percent faster. The Asus model allows 120 percent, runs with higher 2.040 MHz at 247 watts and increases by 4 percent.
The Geforce RTX 2070 Armor from MSI, which runs with a standard clock, delivers a speed that is roughly on par with an overclocked Geforce GTX 1080 – the price-performance ratio of the cards is similar.
The Geforce Geforce RTX 2070 ROG Strix Gaming OC from Asus calculates about 4 percent faster than MSI’s Armor, but is significantly more expensive. For $50 more, there is already a new and clearly faster Geforce RTX 2080 with good cooling, but such cards have less features.
The Strix has features such as voltage measurement points, fan connections and a dual BIOS, which certainly has its charms. And even if the Armor card is already very quiet, the Asus model simply has more reserves for silent friends without running into the temperature target.
The corresponding Geforce RTX 2070 is definitely worth a look for entry level prices under $550 especially because the Geforce GTX 1080 should soon disappear from the market.
Besides a similar performance, the RTX cards also have raytracing and DLSS support, although both functions can’t be tried out outside of tech demos. Nevertheless: This Turing card is worth its money, that’s why we ranked it third for the best Graphics cards for 144Hz Gaming and 1440p / 4k Monitors.